WEBVTT

1
00:04:25.450 --> 00:04:27.540
Lynne Jones: Alright, good morning, everybody.

2
00:04:28.410 --> 00:04:33.149
Lynne Jones: We'll wait a few more minutes, it looks like. There's still people coming in.

3
00:04:44.800 --> 00:04:52.619
Lynne Jones: I hope yesterday was… was useful to people. I know that there was some overlap with what we talked about in the science assemblies, but,

4
00:04:53.420 --> 00:04:54.880
Lynne Jones: I didn't…

5
00:04:55.000 --> 00:05:00.570
Lynne Jones: Not everybody made it to the Science Assembly, and I did try to make it a little bit different.

6
00:05:41.840 --> 00:05:46.760
Lynne Jones: Good morning to those of you just joining us. I'm just gonna…

7
00:05:47.030 --> 00:05:50.559
Lynne Jones: We had a few minutes for people to finish trickling in.

8
00:06:08.700 --> 00:06:17.970
Lynne Jones: Well, I don't want to wait too long, because actually, this first part of today is kind of super packed, so, unfortunately.

9
00:06:18.310 --> 00:06:21.440
Lynne Jones: The content and the time span for that.

10
00:06:21.580 --> 00:06:24.250
Lynne Jones: It turned out weren't as great a match as I thought.

11
00:06:24.390 --> 00:06:30.959
Lynne Jones: So, let's get going. Good morning, and welcome again to the SELC Workshop page, Day 2.

12
00:06:31.460 --> 00:06:38.079
Lynne Jones: This is your SEOC and all of the members, and I… we… on the…

13
00:06:38.240 --> 00:06:43.639
Lynne Jones: from LSST, we really appreciate all of the work the SEOC do.

14
00:06:44.250 --> 00:06:52.750
Lynne Jones: Today, we're going to talk about, more details of some of the things we talked about yesterday.

15
00:06:53.460 --> 00:07:12.829
Lynne Jones: Code of conduct reminder, please be kind. We're here to help make Rubin the most awesome thing it can be together, so let's, be kind, be honest, be inclusive. If you do have a problem, please reach out to me, or Federica Bianco, or your science collaboration point of contact.

16
00:07:14.500 --> 00:07:28.820
Lynne Jones: Okay, quick review of what we did yesterday. We went through the current, options, and we talked about metrics, primarily in sort of a high level, over the whole survey kind of point of view.

17
00:07:29.080 --> 00:07:33.590
Lynne Jones: And today we'll talk about some particular aspects of those

18
00:07:33.720 --> 00:07:37.750
Lynne Jones: Simulations and what we're… what we're actually seeing.

19
00:07:38.870 --> 00:07:43.720
Lynne Jones: From data management, from, various aspects.

20
00:07:44.710 --> 00:07:47.150
Lynne Jones: And at the end of today, we have the…

21
00:07:47.270 --> 00:07:53.710
Lynne Jones: the, a time to talk more specifically about DESI. So…

22
00:07:53.860 --> 00:08:04.279
Lynne Jones: that we'll… we will cover that today, and David, I see you're here. I wasn't able to… to make sure we had the right time, but I think you're up for… for doing…

23
00:08:04.470 --> 00:08:08.049
Lynne Jones: If we're talking about Desi in that last section a little bit, right?

24
00:08:08.210 --> 00:08:09.590
David Schlegel: I'm here all day.

25
00:08:09.590 --> 00:08:11.000
Lynne Jones: Perfect, thank you.

26
00:08:12.270 --> 00:08:13.340
Lynne Jones: Alright.

27
00:08:15.810 --> 00:08:21.439
Lynne Jones: Right, and on Monday, we have this, session to get feedback from the science collaborations.

28
00:08:21.820 --> 00:08:29.799
Lynne Jones: So we can adjust these times as needed for what people's feedback wants to be.

29
00:08:32.450 --> 00:08:34.230
Lynne Jones: Alright, so today.

30
00:08:34.440 --> 00:08:40.229
Lynne Jones: The first thing up is details of the, FBS configuration.

31
00:08:40.669 --> 00:08:48.070
Lynne Jones: and what we're starting the survey with. So this is the chunk of today that is quite dense, and

32
00:08:49.700 --> 00:09:02.040
Lynne Jones: it's hard to talk about the FBS and how it's doing what it's doing in a way that makes sense to people without actually telling you a little bit about what's going on inside the FBS. So that's what I'm going to do.

33
00:09:02.350 --> 00:09:03.870
Lynne Jones: And,

34
00:09:04.280 --> 00:09:16.860
Lynne Jones: If you have absolutely no interest in the inner workings of the FBS, this might not be for you, but I actually think it's pretty fascinating, so hopefully it'll be interesting to you, too. Some of this might be a little bit,

35
00:09:17.170 --> 00:09:20.480
Lynne Jones: fast going through, because, like I said, we…

36
00:09:20.890 --> 00:09:31.630
Lynne Jones: We made the time… the agenda before we made the… before we figured out exactly what had to go in this pot, and but I'm gonna… I'm gonna try for that.

37
00:09:32.320 --> 00:09:39.079
Lynne Jones: Anyway, here we go. Light introduction to the inside of the FBS, and what that means for the start of the survey with V5.3.

38
00:09:39.750 --> 00:09:42.399
Lynne Jones: So… Inside the FBS.

39
00:09:42.800 --> 00:09:57.940
Lynne Jones: how… what we have is the whole scheduler configuration, and inside the FES, there is a list of lists of surveys. The list is what we call tiers, so it's like, this is… you can think of it as prioritizing.

40
00:09:58.100 --> 00:10:02.639
Lynne Jones: A higher tier always has priority over a lower tier.

41
00:10:03.590 --> 00:10:11.369
Lynne Jones: So if we put deep drilling over, a tier with lots of pairs, if the deep drilling survey can request an observation, it will.

42
00:10:11.680 --> 00:10:31.370
Lynne Jones: Within a tier, those surveys compete. So if you have multiple surveys inside a tier, they can compete with each other to say, oh, I really want this observation, or, somebody else may say, I kind of want this one, but they'll go side to side-to-head to see which survey is going to choose the observation for the…

43
00:10:31.800 --> 00:10:33.980
Lynne Jones: Upcoming observations be taken.

44
00:10:34.620 --> 00:10:35.680
Lynne Jones: So…

45
00:10:35.970 --> 00:10:50.339
Lynne Jones: So this is important to keep in mind. We have this concept of tiers, which are priority. If a higher tier survey can take an observation, even if it's a lower priority for that survey than one in a lower tier, the higher tier survey will win.

46
00:10:50.790 --> 00:10:55.259
Lynne Jones: So, tiers prioritize, and then within a tier, the surveys compete.

47
00:10:56.660 --> 00:11:00.219
Lynne Jones: That's the basics. The next thing is, what are surveys?

48
00:11:00.530 --> 00:11:20.120
Lynne Jones: So not every survey makes decisions in the same way, and this is not just true of, like, what goes into the decision in a particular survey, but that they also have different decision-making methods. And so, Peter says, you know, it's a modular decision tree, so we have some of those modules are the tiers

49
00:11:20.120 --> 00:11:39.759
Lynne Jones: And then, within the tiers, the decisions from the surveys. But also, it's because there's a modular range of surveys. So we have things like scripted surveys, which take a list of desired observations, along with some constraints, like, the time window, the hour angle, what the sun altitude is.

50
00:11:39.920 --> 00:11:48.779
Lynne Jones: And then it will try to acquire the observations in that list when the constraints for those… that particular observation is matched.

51
00:11:49.140 --> 00:12:00.000
Lynne Jones: So we use this for the deep drilling fields, you could also use it for engineering tests, and there's lots of use for this, but it's a very, sort of, prescriptive survey.

52
00:12:01.470 --> 00:12:04.100
Lynne Jones: If something can be done, it will, basically.

53
00:12:04.760 --> 00:12:07.160
Lynne Jones: And then we have,

54
00:12:07.180 --> 00:12:26.459
Lynne Jones: Markov, which is our… we actually have many subclasses of this kind of survey. It's basically saying this is a survey that uses a lot of basis functions, potentially a lot of basis functions, to make decisions about which observation to take, and we use these for a wide area survey, for the most part.

55
00:12:26.620 --> 00:12:32.120
Lynne Jones: Sometimes we use them for more like a single pointing, and those are field surveys. They are…

56
00:12:32.450 --> 00:12:36.150
Lynne Jones: They're related, they're… it's just they only have one.

57
00:12:36.590 --> 00:12:38.190
Lynne Jones: Position on the sky.

58
00:12:39.510 --> 00:12:54.449
Lynne Jones: For field surveys, you can put in a sequence to be… to be obtained every time you… that survey would request an observation. For the Markov surveys, it's more like you choose one observation, but you can choose it based on anywhere on the sky.

59
00:12:55.070 --> 00:13:10.659
Lynne Jones: Now, really, to fully define a survey, you also… and to know, like, this survey is this survey, and not a slightly different survey. You do have to dig into all of the basis functions to say, like, how is this survey making its decisions?

60
00:13:10.660 --> 00:13:15.729
Lynne Jones: I am not going to go through all of the basis functions for every survey.

61
00:13:16.350 --> 00:13:20.820
Lynne Jones: Here, because that would not be interesting to people, and it would be very long.

62
00:13:20.910 --> 00:13:31.330
Lynne Jones: But for now, let's just say that for most of the Markov-based surveys, the typical balance we look for is between the footprint, the,

63
00:13:31.370 --> 00:13:46.250
Lynne Jones: what we call M5 difference, which is the 5-sigma point source limiting magnitude that we estimate you would have now over the sky versus the dark sky depth. So it's an estimate, not just of where it's, where it's,

64
00:13:46.580 --> 00:13:53.279
Lynne Jones: Where you have the most limiting magnitude right now, but also just compared to what that would be under dark sky conditions.

65
00:13:53.410 --> 00:13:57.330
Lynne Jones: And then we have, a slew time basis function.

66
00:13:57.990 --> 00:14:06.030
Lynne Jones: And these are sort of our big three. You can think of as, like, if you have a wide area survey, these are the big three.

67
00:14:06.340 --> 00:14:18.319
Lynne Jones: basis functions that you'll be trying to balance. And it goes so that you have in… you can balance even coverage, efficient observations with the slew time, and

68
00:14:18.670 --> 00:14:20.750
Lynne Jones: Photon collection with the depth.

69
00:14:22.540 --> 00:14:28.609
Lynne Jones: And there's… there's numbers that you can use to change how these balance against each other, for example.

70
00:14:28.800 --> 00:14:35.769
Lynne Jones: These are all some of the things that go into making a full survey configuration, but this is the basic, basic.

71
00:14:36.800 --> 00:14:42.119
Lynne Jones: And then another thing we do for the use at the summit, is we have

72
00:14:42.270 --> 00:14:53.789
Lynne Jones: some standard masks, which we apply to every survey. And these are basically that we're… we're putting in the administrative constraints, or the just general con…

73
00:14:54.140 --> 00:14:59.639
Lynne Jones: A little bit more than that because of Venus and Jupiter, but these are constraints that every survey will follow.

74
00:14:59.870 --> 00:15:09.239
Lynne Jones: So we have, avoiding the moon, by about 30 degrees. Every survey will respect the altitude limits for the telescope.

75
00:15:09.630 --> 00:15:28.509
Lynne Jones: So we don't… so this will mask off the sky below our lower limit and around zenith. And every survey, in the FES avoids Venus, Mars, and Jupiter, merely because they're so bright and they move fast, so it's just not worth, us looking at them.

76
00:15:29.600 --> 00:15:39.199
Lynne Jones: And we can also add temporary masks, so for example, if we need to avoid a given azimuth range on a particular night, we can… we can add this in.

77
00:15:43.360 --> 00:15:47.360
Lynne Jones: So that's the basics of what goes into

78
00:15:49.140 --> 00:15:57.690
Lynne Jones: our FPS configurations on the sky. And, like, this is the specifics for what we have now in V5.3.

79
00:15:57.970 --> 00:16:02.689
Lynne Jones: So, you can see there's a lot of things on this list. We have 8 tiers.

80
00:16:03.210 --> 00:16:15.289
Lynne Jones: And then within those tiers, we have 40 surveys, but all those are not necessarily in the same tier, so those 40 surveys aren't all competing against each other, they're only competing within their same tier.

81
00:16:17.010 --> 00:16:33.589
Lynne Jones: Okay, so when we say this is what the FES is doing, what we mean is at a particular time, the FES gets updated with the conditions, so the… where the telescope is, what the seeing is like, what the weather's like, what the sky brightness is.

82
00:16:34.380 --> 00:16:48.630
Lynne Jones: And then it goes through this list of tiers until it comes to a tier that says, yes, this is perfect, I can take an observation now, and then that tier will choose an observation.

83
00:16:48.800 --> 00:16:50.370
Lynne Jones: For the survey.

84
00:16:51.250 --> 00:17:02.650
Lynne Jones: For the whole survey. This is where it gets a little tricky. So we have our survey objects, and then we have the survey we're doing, but here, I mean, I'll try to make sure I'm saying survey by the survey objects.

85
00:17:02.840 --> 00:17:03.700
Lynne Jones: So…

86
00:17:06.540 --> 00:17:15.980
Lynne Jones: These are our tiers. I'm going to go through each tier and just say this is the kind of observations this would take and what they look like in our 5.3 simulation.

87
00:17:17.180 --> 00:17:21.450
Lynne Jones: So we have the target of opportunity tier, the TO tier.

88
00:17:22.390 --> 00:17:38.650
Lynne Jones: This, is a set of… basically a set of scripted observations that are waiting for information that says, oh, this event happened. If the event matches the type of object it's supposed to follow up, then it follows the scripted,

89
00:17:38.720 --> 00:17:45.030
Lynne Jones: the predefined observation plan, which comes from the 2024 TO Workshop Report.

90
00:17:45.870 --> 00:17:47.060
Lynne Jones: So…

91
00:17:47.180 --> 00:17:57.620
Lynne Jones: this is a top-tier survey, the very top tier, so if a TO event comes in, and a survey… one of these, TOO surveys.

92
00:17:57.990 --> 00:18:12.299
Lynne Jones: matches the event, and the observations are possible, like, the list of observations from a scripted survey says, yes, this matches the conditions, then this will be the next thing that's observed.

93
00:18:14.830 --> 00:18:31.000
Lynne Jones: I'm gonna introduce here, not just our sky plot on the left, but on the right. This is an LTAS distribution of where those observations were in V5.3. So you see, these are… these are being picked up over

94
00:18:31.670 --> 00:18:39.380
Lynne Jones: On the left, you can see over it pretty much all over the sky, and that's, by chance, of course, because…

95
00:18:39.580 --> 00:18:42.000
Lynne Jones: We don't know where the real TOs are going to be.

96
00:18:42.420 --> 00:18:49.230
Lynne Jones: And on the right is, like, telling you where, from the telescope's point of view, it was observing in order to

97
00:18:49.390 --> 00:18:52.459
Lynne Jones: to see these duos.

98
00:18:54.660 --> 00:19:06.640
Lynne Jones: Our next tier is, a survey that's trying to coordinate with the Roman Galactic playing field. So, this results in… it's set up to obtain about,

99
00:19:06.640 --> 00:19:16.030
Lynne Jones: just over 2,000 visits, spanning 5 years. This is frequent monitoring while the field is available, so… and it goes through GRIZ bands.

100
00:19:16.060 --> 00:19:18.359
Lynne Jones: And the idea is to coordinate with Roman.

101
00:19:18.600 --> 00:19:30.550
Lynne Jones: This does not request any observations within year one, and it is a survey that we need to update and make sure we're actually coordinating, because we don't have this full information yet. So…

102
00:19:30.650 --> 00:19:37.280
Lynne Jones: These observations in the full 5.3 simulation are a little bit of a placeholder, but it's about what we want to do.

103
00:19:40.760 --> 00:19:45.690
Lynne Jones: Our next year is… The deep drilling fields.

104
00:19:46.870 --> 00:19:52.710
Lynne Jones: So, this is a single scripted survey containing all of the predefined DDF

105
00:19:53.340 --> 00:20:00.129
Lynne Jones: desired observations. So this survey is configured a little bit differently than a lot of our other things.

106
00:20:00.300 --> 00:20:10.979
Lynne Jones: For a scripted survey, you basically can give it a list to say ahead of time, to say, here, these are the observations I want to take. Please try to take these observations if you can.

107
00:20:11.960 --> 00:20:14.939
Lynne Jones: And obviously, each of those observations come with

108
00:20:15.050 --> 00:20:23.670
Lynne Jones: Some constraints, the most important of which is usually the time window that that observation could be acquired in.

109
00:20:24.810 --> 00:20:33.960
Lynne Jones: So, to set up our desired DDF observations, we optimize those independently ahead of time. And this is so that

110
00:20:34.320 --> 00:20:46.219
Lynne Jones: because these are a very limited set of pointings, we can do this. We can say, when would be the best time to observe this field when we have this particular number of observations we want to get in the season.

111
00:20:46.580 --> 00:21:01.559
Lynne Jones: With some cadence constraints, perhaps, and then pick the best time in the night based on the sky brightness, to obtain that observation. And also, that way we can skip times when the moon is bright, and so on.

112
00:21:03.050 --> 00:21:20.200
Lynne Jones: So, the DDFs are handled a little bit differently than most of the other things. We're going ahead, before we even get to the survey to say, this is when we should take these observations over the season. Obviously, we don't know when the weather's going to be good or bad, but that's where the scripted survey can handle that.

113
00:21:20.250 --> 00:21:26.830
Lynne Jones: Because you can give it each of those observations, a time window of when to… to actually run.

114
00:21:27.830 --> 00:21:34.509
Lynne Jones: So when we set up our list, we take into account the time window and flexibility, for example.

115
00:21:35.650 --> 00:21:39.809
Lynne Jones: So, the… we have, again, on the bottom here, the…

116
00:21:39.910 --> 00:21:52.520
Lynne Jones: RADEX, distribution on the sky for the deep drilling fields. I think this is fairly familiar to most people. In the middle, the distribution of those visits over altitude and azimuth.

117
00:21:52.520 --> 00:22:00.579
Lynne Jones: And this is kind of interesting because you can see that there's this little hole in the middle. You probably can't see my mouse cursor, but there.

118
00:22:00.660 --> 00:22:06.259
Lynne Jones: I mean, that's the zenith exclusion region, because we can't track the telescope through zenith.

119
00:22:06.630 --> 00:22:11.090
Lynne Jones: So we can observe on either side of Zenith, but not right at Zenith.

120
00:22:11.580 --> 00:22:18.019
Lynne Jones: And then on the right, here is now, another new plot. So what this one shows is…

121
00:22:18.260 --> 00:22:25.359
Lynne Jones: On the y-axis, it's time away from local midnight, and on the x-axis, it's just,

122
00:22:25.650 --> 00:22:27.890
Lynne Jones: Basically, the night of the survey.

123
00:22:27.920 --> 00:22:46.319
Lynne Jones: So, you can see the length of the night changing over… this is just the first two years of the survey. So, you can see the length of the night changing with those red lines getting narrower and then wider. And then the times of the deep drilling fields and their filters are shown by the little lines on the inside.

124
00:22:46.940 --> 00:22:49.729
Lynne Jones: And it is a little bit hard to,

125
00:22:50.500 --> 00:23:03.979
Lynne Jones: There's a problem with, like, scale and how much data you're trying to show, so some of these points in the plot are not super easy to see, but hopefully you were able to pick up the link to my slides.

126
00:23:04.160 --> 00:23:07.189
Lynne Jones: But you can… what you can see here is that

127
00:23:07.350 --> 00:23:10.770
Lynne Jones: As the field becomes available, we observe the field.

128
00:23:11.390 --> 00:23:15.190
Lynne Jones: At the best time of the night for,

129
00:23:15.520 --> 00:23:26.959
Lynne Jones: for the time where the field is visible. And then you can see the similar thing happening for, like, the other fields, because this is all the deep join fields together. They're not quite at the same RA, but they're very clustered.

130
00:23:27.110 --> 00:23:29.550
Lynne Jones: And then began the same.

131
00:23:31.130 --> 00:23:34.939
Lynne Jones: So that, that is all of the deep drilling fields, so they have slightly different

132
00:23:35.080 --> 00:23:40.069
Lynne Jones: RA ranges and times, but you can see how we're observing,

133
00:23:42.140 --> 00:23:51.269
Lynne Jones: At the best… hopefully at the best time of the night for all of those that we've chosen the time when the sky brightness is the best.

134
00:23:52.080 --> 00:23:53.600
Lynne Jones: Alright, so…

135
00:23:55.610 --> 00:24:10.119
Lynne Jones: Right, I need to say a little bit more about the deep drilling fields. For each deep drilling visit, we have, rotational and translational dithers applied. It's a 3-degree rotational dither between visits, so until we change the filter, we just…

136
00:24:10.380 --> 00:24:19.390
Lynne Jones: Click along by 3 degrees. And then the translational offset is a radial dither around the field center of 0.2 degrees.

137
00:24:19.660 --> 00:24:27.599
Lynne Jones: For, the Euclid deep drilling field, it's a little bit different, because we don't do a radial dither, we do dither towards the other field, but

138
00:24:27.920 --> 00:24:33.080
Lynne Jones: It's… it's easier to just describe these for the rest of the deep drilling fields.

139
00:24:33.520 --> 00:24:35.580
Lynne Jones: The sequences,

140
00:24:35.880 --> 00:24:48.199
Lynne Jones: For these… the sequences we set up in the pre-calculated requests, so this is what we're trying to obtain, are in deep seasons, those are the shallow seasons or deep seasons,

141
00:24:48.940 --> 00:24:58.669
Lynne Jones: It alternates. So on even days, we would get two visits in G and 2 visits in I, and on odd days, we'd get two visits in R and 2 visits in C.

142
00:24:59.520 --> 00:25:09.719
Lynne Jones: And then every few days, so that it lines up with these, but it's not quite every day, we have 3 visits in U and 2 visits in Y.

143
00:25:09.750 --> 00:25:26.850
Lynne Jones: So, as I said, we have time windows for each of these, so you can think of them as expiring. So, the even and odd don't build up over multiple days. They expire very quickly. The U and Y, expire a little bit more slowly, so they can occasionally double up, and so we would get 6 in U.

144
00:25:27.060 --> 00:25:29.110
Lynne Jones: Or foreign and Y.

145
00:25:30.670 --> 00:25:38.529
Lynne Jones: For the EDFS field, the sequences are the same, but the cadence, the frequency, is slower to accommodate the double pointing.

146
00:25:40.340 --> 00:25:53.319
Lynne Jones: And then in ultra-deep seasons, we still have the same even-odd day sequences, but then we also add, frequent, but not everyday, sequences of

147
00:25:53.470 --> 00:25:57.290
Lynne Jones: 8 in U, 4 in G, 18 in R,

148
00:25:57.560 --> 00:26:02.420
Lynne Jones: 55 in I, 52 in Z, and 20 in Y band.

149
00:26:02.800 --> 00:26:07.280
Lynne Jones: So this is very, very red-heavy.

150
00:26:07.470 --> 00:26:12.250
Lynne Jones: And the idea is that this is looking for a high redshift supernova.

151
00:26:19.520 --> 00:26:21.810
Lynne Jones: Now we move along to the template tier.

152
00:26:21.920 --> 00:26:25.470
Lynne Jones: Please do go ahead and raise your hand if you have questions while we're going through here.

153
00:26:28.290 --> 00:26:32.289
Lynne Jones: Okay, so let's talk about tier. This is our new tier.

154
00:26:32.590 --> 00:26:34.350
Lynne Jones: In B5.3.

155
00:26:34.810 --> 00:26:42.070
Lynne Jones: We're acquiring visits separated by about 25 minutes, and good image quality for the template here.

156
00:26:42.170 --> 00:26:53.259
Lynne Jones: And you can see the names of the filters that are paired together for these pairs in the list up at the top. So we have U paired with U, G with G, G also with R, R with A, R,

157
00:26:53.660 --> 00:27:08.570
Lynne Jones: R&I. So, in general, for our standard pairs, we don't pair the same filter with itself, except for Y-band, but for the template tier, because they're, we're trying to get the U-band images under

158
00:27:09.050 --> 00:27:14.230
Lynne Jones: tighter constraints for the image quality, we do pair U with itself and G with itself.

159
00:27:15.690 --> 00:27:16.640
Lynne Jones: So…

160
00:27:17.500 --> 00:27:37.219
Lynne Jones: The template tier will only request observations when the atmospheric seeing, as we're getting the measurement from the DIMM currently, implies that the predicted seeing, full width half max, sorry, for the next visit would be better than the equivalent of, 1.2 arcseconds at zenith.

161
00:27:37.340 --> 00:27:56.450
Lynne Jones: And this is a little bit, tiny bit tricky to explain, because when I say 1.2 arcseconds at zenith, we do let that vary with declination, so I have a plot on the right of what a 1.2 full-width half max at zenith translates to over the declination range of our footprint.

162
00:27:57.500 --> 00:28:01.649
Lynne Jones: Because obviously a higher deck, you don't get as good seeing.

163
00:28:02.160 --> 00:28:11.879
Lynne Jones: And so… That the… the template tier is looking for the declination appropriate,

164
00:28:12.170 --> 00:28:22.220
Lynne Jones: predicted full-width half wax at its location. And it will also only track observations which meet those, those, full-width half-max requirements.

165
00:28:22.380 --> 00:28:34.049
Lynne Jones: There are also some hour angle limits to make sure we're pushing towards Zenith, and there are some additional constraints. Otherwise, many aspects are similar to the standard pair survey.

166
00:28:34.350 --> 00:28:42.770
Lynne Jones: The template tier is important to note, shuts off when 6 visits with the desired full-width half max are acquired.

167
00:28:42.970 --> 00:28:45.650
Lynne Jones: Or at the end of year one, whichever is soonest.

168
00:28:47.130 --> 00:29:04.799
Lynne Jones: And this is in our 5.3 configuration. This is the kind of thing, like, the details of what are… is the cutoff, at Zenith, and how many visits we need, and when should it cut off. Those are things that we'll… we'll go back and adjust as we see

169
00:29:05.560 --> 00:29:11.379
Lynne Jones: As we sync up better with, data management pipeline and requirements.

170
00:29:11.530 --> 00:29:14.390
Lynne Jones: And we see what they, what they can use.

171
00:29:17.820 --> 00:29:20.520
Lynne Jones: The template tier, so this is the…

172
00:29:20.860 --> 00:29:30.849
Lynne Jones: footprint, again, on the sky in our NDAC, and the footprint in altitude and azimuth. So what you're seeing here, that wedge shape, is the hour angle constraints.

173
00:29:31.010 --> 00:29:41.319
Lynne Jones: And then, on the right is that same plot of, like, when the observations are happening. And notice this one's much shorter in time because it only runs for the first year.

174
00:29:42.370 --> 00:29:47.519
Lynne Jones: You can also, now that there's more observations on this plot, you can see,

175
00:29:47.760 --> 00:29:55.680
Lynne Jones: A blank period here, just shortly after the start of the simulation, and that's because we have a downtime period there.

176
00:29:57.310 --> 00:29:58.879
Lynne Jones: In the simulation.

177
00:29:59.530 --> 00:30:05.379
Lynne Jones: Your Casol can notice that This is taking up most of the time of the survey.

178
00:30:07.170 --> 00:30:13.769
Lynne Jones: Not all of it, but a lot of it. And actually, now I think about it, see, if you notice this stripe going through here… oops…

179
00:30:14.210 --> 00:30:20.179
Lynne Jones: That would be because that matches up with our deep drilling fields in the previous pot.

180
00:30:20.590 --> 00:30:25.390
Lynne Jones: And when the deep drilling fields are intensely taking observations.

181
00:30:27.690 --> 00:30:28.650
Lynne Jones: Okay.

182
00:30:29.040 --> 00:30:36.520
Lynne Jones: Our next tier down is triplets, because we have a desire to get a few hour timescale follow-up.

183
00:30:36.690 --> 00:30:49.939
Lynne Jones: So, what happens here is we have a pair taken near the start of the night, which then triggers a follow-up visit two to seven hours later in one of the same filters. So the first pair is GR, RI, or IZ.

184
00:30:50.010 --> 00:31:07.579
Lynne Jones: And the, follow-up visit is in one of those same filters. The pairs come from a standard pair survey, but only runs once a night, only once every four days, so there's more time constraints. And the third visit comes from one of these scripted surveys, which says, oh, you took an observation, let me put that on my list to follow up.

185
00:31:08.680 --> 00:31:20.350
Lynne Jones: There are some hour angle constraints on the first pair, so we push it into parts of the sky which are rising, so we don't accidentally, like, make sure that we couldn't opposite reserve later.

186
00:31:21.500 --> 00:31:27.069
Lynne Jones: And then here you can see what happens. So we have the first pair.

187
00:31:27.460 --> 00:31:43.460
Lynne Jones: being pushed into the parts of the sky that are rising, and then you have the second visit of the pair, or sorry, the triplet after the pair, occurring at a wider range of positions on the sky. Because of that, it could be anywhere from 2 to 7 hours later.

188
00:31:43.520 --> 00:31:47.659
Lynne Jones: And then these are the same plots here, but this is the first part, first pair.

189
00:31:47.800 --> 00:31:50.769
Lynne Jones: So you can see it's stacked up near the start of the night.

190
00:31:50.950 --> 00:31:53.410
Lynne Jones: And only happening once per night.

191
00:31:53.620 --> 00:31:57.980
Lynne Jones: And then, the second… the triplet, sorry.

192
00:31:58.400 --> 00:32:03.930
Lynne Jones: Second survey, triplet of the pair, happening at various other times during the night.

193
00:32:05.870 --> 00:32:16.330
Lynne Jones: Okay, and now we come to our parent survey, which is fairly far down in the tiers, but this is actually the workhorse of our survey. More than 60% of the visits actually come from this tier. This is the…

194
00:32:16.810 --> 00:32:29.490
Lynne Jones: the standard pairs all over the sky, survey that we have. So this is where we're, again, we're balancing the M5 difference, the footprint, the slew times, make generally balanced choices, and then

195
00:32:29.490 --> 00:32:40.000
Lynne Jones: We add some extra basis functions to do things like make sure we get, at least 4 visits per pointing in each band per year,

196
00:32:40.480 --> 00:32:47.520
Lynne Jones: So that we don't try to take more than two visits per night at any point in the sky. The pairs are,

197
00:32:47.760 --> 00:32:50.440
Lynne Jones: The first visit of… of…

198
00:32:50.790 --> 00:32:56.969
Lynne Jones: The pair is followed by a second in a different filter, and you can see the filter combinations up at the top.

199
00:32:57.410 --> 00:32:58.290
Lynne Jones: But…

200
00:32:58.620 --> 00:33:15.029
Lynne Jones: And we… the goal revisit time for the second visit in the pair is 33 minutes, and you can see on this plot here, which is, over the whole sky, what is the typical time to coming back to the field, that we do have this good, strong peak at,

201
00:33:15.090 --> 00:33:29.609
Lynne Jones: 33 minutes. You also see this peak here, and this is just because of how this metric works, and this is from field overlaps. So when we take a visit, we're most likely to take the next visit right next to it, and then that gives us a little bit of overlap, so that has a very short

202
00:33:30.980 --> 00:33:33.520
Lynne Jones: Interval between visits in there.

203
00:33:34.580 --> 00:33:41.150
Lynne Jones: So, at 5.3, one of the new things we've done is that the visits in the pair

204
00:33:41.380 --> 00:33:51.260
Lynne Jones: The interval is trying to make 33 minutes, but it will scale that slightly depending on, the best visit time available.

205
00:33:51.360 --> 00:34:02.639
Lynne Jones: Sorry, how to best fill the time of the night. So the goal is 33 minutes, that's the most common time, but it can range, coming from the same survey, anywhere from, like, 40, 20,

206
00:34:02.840 --> 00:34:04.410
Lynne Jones: to,

207
00:34:05.260 --> 00:34:12.500
Lynne Jones: It's 40 or 45 minutes, or something like that. So we… it could be a little bit longer, depending on how much time you have in the night.

208
00:34:13.500 --> 00:34:31.080
Lynne Jones: This replaces, in our… if you were more familiar with our old survey configuration, we had this 15-minute pair and a 33-minute pair. We've consolidated them with this, dynamic time scaling. We only have this one pair survey now. So, this is good, simplifies by…

209
00:34:31.909 --> 00:34:34.949
Lynne Jones: How are these observing over the whole sky?

210
00:34:35.120 --> 00:34:36.540
Lynne Jones: Pretty much everywhere.

211
00:34:36.960 --> 00:34:52.209
Lynne Jones: And then on the right, you see this same plot of visits over time from 5.3, and notice there's very few visits in pairs in the first year, and that's because most of the time we're taking templates. And then after the template tier shuts off, we start taking, pairs.

212
00:34:52.750 --> 00:34:56.150
Lynne Jones: the standard pairs. Remember, the templates are in pairs as well.

213
00:34:57.790 --> 00:35:01.890
Lynne Jones: Alright, now we have one of our microsurvey,

214
00:35:02.540 --> 00:35:19.560
Lynne Jones: tiers, our only microsurvey tier, and this is the Twilight Near Sun microsurvey. It takes observations in RI and Z. It's trying to get low, solar elongation, nearly ecliptic, observations during twilight, so it…

215
00:35:20.120 --> 00:35:25.680
Lynne Jones: runs, it will only start if the sun altitude is,

216
00:35:25.930 --> 00:35:32.739
Lynne Jones: above minus 15 during this time period, and it does short exposures, 15 seconds. So…

217
00:35:33.620 --> 00:35:37.829
Lynne Jones: It runs in the target area. It defines a target area.

218
00:35:38.410 --> 00:35:56.900
Lynne Jones: One of these RIZ will win in that tier, and then it will try to get four images for pointing in the target area. And the area means that the total time we need to complete the whole microsurvey in, like, the morning or night of when it's triggered is about 20 minutes.

219
00:35:57.940 --> 00:36:05.000
Lynne Jones: So… These do successfully get low solar elongation visits, you can see that on the right.

220
00:36:05.360 --> 00:36:09.939
Lynne Jones: And you can also see that these are being pushed to the,

221
00:36:10.670 --> 00:36:14.269
Lynne Jones: The far edges of the, of the,

222
00:36:15.210 --> 00:36:22.310
Lynne Jones: Of the sky, and this triggers in the morning and the evening, but not every day, it's every, 4 days.

223
00:36:22.780 --> 00:36:23.440
Lynne Jones: I think.

224
00:36:23.690 --> 00:36:24.500
Lynne Jones: Yes.

225
00:36:27.210 --> 00:36:43.329
Lynne Jones: All right, and then the very last tier we have is the greedy tier. So, this tier serves as a filler. When no other survey can schedule visits, the greedy survey is there for you. It just schedules one visit at a time, and it doesn't,

226
00:36:43.600 --> 00:37:03.390
Lynne Jones: have a lot of other time constraints. It sort of is discouraged from revisiting the same part of the sky more than once in a night, but other than that, it just is going to balance what's the depth, what's the footprint, what's the sleeve time, so that it can make generally balanced choices. But there are no pairs required, there's no,

227
00:37:03.490 --> 00:37:05.879
Lynne Jones: You must finish,

228
00:37:06.070 --> 00:37:16.959
Lynne Jones: because it's only one visit, it doesn't have to consider too far into the future. And so, a lot of those other basis functions from all of the other surveys have dropped. It's simple,

229
00:37:17.150 --> 00:37:25.550
Lynne Jones: You wouldn't want to take most of your survey visits like this, because it also does things like get trapped in particular areas of the sky, but

230
00:37:26.000 --> 00:37:28.440
Lynne Jones: Yeah, this is our… this is our…

231
00:37:28.730 --> 00:37:31.650
Lynne Jones: our single visits over the sky.

232
00:37:31.770 --> 00:37:35.659
Lynne Jones: Backing up everything else. This runs in RIZNY.

233
00:37:38.730 --> 00:37:55.539
Lynne Jones: Look at that, I made it through all of the tiers. I'm actually going to talk a little bit about, like, what else went into our simulations and what we didn't know next, so if there are questions about, like, the survey configuration itself, the scheduler configuration.

234
00:37:55.780 --> 00:37:58.550
Lynne Jones: Could, take just a quick moment.

235
00:38:02.400 --> 00:38:03.420
Lynne Jones: David?

236
00:38:03.730 --> 00:38:11.679
David Schlegel: Yeah, for the wide, fast, deep survey, then, is that… that's the combination of the template survey and then the para-survey?

237
00:38:12.380 --> 00:38:18.830
Lynne Jones: So, this… this is a great question, actually. I didn't… really,

238
00:38:19.190 --> 00:38:34.840
Lynne Jones: So the… so the wide, fast, deep is more of, like, an area and goal number, of visits over the sky. So, when we define the footprint that goes into the greedy survey, or the pairs survey, or the temp…

239
00:38:35.430 --> 00:38:41.289
Lynne Jones: Not so much the template, but a little bit the template. But when we define the footprint that goes in here.

240
00:38:41.720 --> 00:38:51.290
Lynne Jones: and the greedy survey. We use our footprint over the whole sky, and so the whitefast deep footprint is encoded into…

241
00:38:51.490 --> 00:38:54.890
Lynne Jones: That information that we pass to each of these surveys.

242
00:38:55.130 --> 00:38:56.180
Lynne Jones: So…

243
00:38:57.680 --> 00:39:06.590
Lynne Jones: The Wide Fast Deep… there is not really a survey that is charged with acquiring Wide Fast Deep from the scheduler point of view.

244
00:39:07.270 --> 00:39:17.759
Lynne Jones: The… where this comes in is that we, as the survey strategy team, know what our coverage should be, and so we say, yes, that's what the footprint should look like, so that we do have a wide, fast, deep area.

245
00:39:18.040 --> 00:39:29.299
Lynne Jones: And then we put… place that into the footprint. And because we know we want pairs in the YFSD, most… we arrange things so that the pair survey gets most of those visits.

246
00:39:31.570 --> 00:39:41.689
Lynne Jones: So… From within the scheduler configuration, there really is no Wide Fast Deep versus North Ecliptic spur.

247
00:39:42.230 --> 00:39:47.160
Lynne Jones: But that all goes in as part of the information we feed into the survey.

248
00:39:47.380 --> 00:39:54.759
David Schlegel: So, sorry, just a quick follow-up. So, but it must be that things like the twilight observations wouldn't count towards that coverage?

249
00:39:55.170 --> 00:39:57.240
Lynne Jones: Yeah, so, so…

250
00:39:58.040 --> 00:40:16.579
Lynne Jones: Also a good question. This is true, because every one of these surveys that I talked about, you not only have the option to mask off parts of the sky, but you can also tell it to ignore certain kinds of observations. So there is a piece of metadata that actually gets into all of the

251
00:40:16.580 --> 00:40:29.940
Lynne Jones: information from the visits that is later available to everybody, called the scheduler note. And this is our piece of metadata that sort of tracks which of these surveys it came from, and sometimes extra information as well.

252
00:40:30.540 --> 00:40:41.380
Lynne Jones: And so, when this visit comes from the Twilight Near Sun Survey, it carries a piece of information that says, this came from the Twilight Near Sun Survey.

253
00:40:41.510 --> 00:40:42.880
Lynne Jones: And so…

254
00:40:43.310 --> 00:40:50.769
Lynne Jones: The pair survey is told, ignore those. Those are not relevant for you, you don't use that to build up the footprint.

255
00:40:51.220 --> 00:40:53.730
Lynne Jones: So…

256
00:40:53.890 --> 00:41:05.339
Lynne Jones: that's how… that's how this works. And we also tell the the par survey, for example, to ignore the TOO's observations, because they were putting dents in our footprint.

257
00:41:05.450 --> 00:41:10.269
Lynne Jones: And so there's… There's a lot of…

258
00:41:10.400 --> 00:41:14.240
Lynne Jones: Room to go in and think about, like, how can you…

259
00:41:14.440 --> 00:41:24.080
Lynne Jones: change how the survey is behaving, and this is one of the challenges, but also one of the strengths about the FBS configuration. There's usually more than one way to do a thing.

260
00:41:29.390 --> 00:41:30.070
Lynne Jones: Bye.

261
00:41:31.030 --> 00:41:32.719
Lynne Jones: So, I will…

262
00:41:33.890 --> 00:41:53.049
Lynne Jones: Unless there's other questions. I'm gonna go and talk a little bit about, like, so yesterday we talked about the V5.3 simulation, we talked about, oh, there's 10% lost visits in the 5.3, yeah, sorry, Sarah pointed out the trip list here also contributes to YFAST. There's a lot of things that contribute to WideFastD.

263
00:41:53.130 --> 00:41:54.990
Lynne Jones: And even the greedy survey does.

264
00:41:55.690 --> 00:42:01.869
Lynne Jones: So, yesterday we said there's 10% fewer visits in V5.3.

265
00:42:02.530 --> 00:42:19.750
Lynne Jones: And, I just want to take a moment to acknowledge that this… as Jocko did as well, right, we don't know what the weather is going to be. We also don't know some other things. So, the visit gap, the time between visits we talked about yesterday,

266
00:42:20.280 --> 00:42:26.099
Lynne Jones: I said it has some overhead that's for every slew that is kind of variable.

267
00:42:26.750 --> 00:42:32.299
Lynne Jones: And… at the moment, I think that variability might be,

268
00:42:32.480 --> 00:42:42.659
Lynne Jones: at least might lead to a variation in the total number of visits, that it may be that we get the number wrong by about 10%.

269
00:42:42.870 --> 00:42:45.990
Lynne Jones: That might be… not be too high right now.

270
00:42:46.330 --> 00:42:50.920
Lynne Jones: But after 6 months, I think that we should not be, like.

271
00:42:51.110 --> 00:43:03.710
Lynne Jones: incorrect by this much, and not just because we will know better, like, how the slew time is, but because the slew time itself will improve and match our model. But we don't know exactly how fast that's going to improve.

272
00:43:04.130 --> 00:43:08.739
Lynne Jones: And so, we didn't put it into the simulation.

273
00:43:08.870 --> 00:43:19.939
Lynne Jones: So, in the first… in the first period of time, our rate of acquiring visits in the simulation, V5.3, is probably a little bit too high.

274
00:43:21.180 --> 00:43:28.170
Lynne Jones: The actual engineering time needs. So, we put in 2 half-nights for the first 6 months.

275
00:43:28.420 --> 00:43:40.560
Lynne Jones: But that could be 2 full nights for the first 6 months. And so that kind of gives you that we might be as much as, 15% too high in the total number of visits in the first 6 months.

276
00:43:40.730 --> 00:43:48.259
Lynne Jones: But, after a full year, when… if we're not doing that engineering time anymore, which we… is not the

277
00:43:48.590 --> 00:43:50.010
Lynne Jones: current estimate.

278
00:43:50.230 --> 00:43:52.709
Lynne Jones: Then we might… we wouldn't be that

279
00:43:53.060 --> 00:43:56.129
Lynne Jones: incorrect. That number reduces over time, right?

280
00:43:57.200 --> 00:44:02.560
Lynne Jones: Also, it doesn't take into account that we might get science images from those engineering tests.

281
00:44:04.680 --> 00:44:15.250
Lynne Jones: Because sometimes they do run the FES during their engineering night, sometimes they run the very specific kinds of observations that are still valid observations for engineering.

282
00:44:15.410 --> 00:44:23.739
Lynne Jones: So there's… there's sort of uncertainty here about what is… what will happen with that engineering time, and

283
00:44:23.940 --> 00:44:26.339
Lynne Jones: What use it might be able to be used for.

284
00:44:26.780 --> 00:44:40.009
Lynne Jones: there's the fault downtime and timescale of improvement. So, we do have a model for faults in B5.3, which is an improvement over our earlier sims. I do think it's

285
00:44:40.550 --> 00:44:48.330
Lynne Jones: more accurate and relatively accurate, but there is a lot of variability on this, and so this, I think, is a little hard to say.

286
00:44:48.800 --> 00:44:57.960
Lynne Jones: I'm hoping we have guessed fairly right, but this is… it's hard to put a number on how accurate or not accurate,

287
00:44:58.420 --> 00:45:00.359
Lynne Jones: The values we've come up with are.

288
00:45:01.190 --> 00:45:18.599
Lynne Jones: Then we have the system contribution to seeing, and the timescale for improvement. So we're just not modeling this at present. We can look at doing that, especially during the first year, but also we don't know the timescale for that improvement, or,

289
00:45:20.300 --> 00:45:28.520
Lynne Jones: Or… or… what will happen there? Most likely, as we get more louvers, we'll see changes, and…

290
00:45:28.630 --> 00:45:33.159
Lynne Jones: That will be, will be something we track, for sure.

291
00:45:34.230 --> 00:45:51.240
Lynne Jones: But then we have the effects of cloud transparency, which we don't model at present, and dodging clouds, which we haven't… we're working on, and we actually have a couple of things in active development, but,

292
00:45:51.260 --> 00:45:55.170
Lynne Jones: There certainly is room for future improvement here, so…

293
00:45:55.710 --> 00:46:02.509
Lynne Jones: And then the last thing is the weather downtime. So, we don't have any clear indication of issues, but obviously we'll be monitoring this.

294
00:46:02.770 --> 00:46:11.750
Lynne Jones: So, As you look at smaller and smaller chunks of time out of V5.3.1, I think your uncertainty,

295
00:46:11.750 --> 00:46:25.889
Lynne Jones: is higher, right? As you look at the full 10 years, that's probably not as uncertain as if you're looking only at the first 6 months. So I think this is the main thing I wanted to say here, is just to make sure we… you understand, like.

296
00:46:26.150 --> 00:46:31.110
Lynne Jones: If you're looking at a small timescale, The uncertainty is definitely higher.

297
00:46:34.040 --> 00:46:34.930
Lynne Jones: Right.

298
00:46:35.120 --> 00:46:37.349
Lynne Jones: We took a couple questions along the way.

299
00:46:37.930 --> 00:46:42.190
Lynne Jones: But let's… if there's any questions here… and then I have…

300
00:46:42.540 --> 00:46:48.299
Lynne Jones: Taken up most of the, most of the question time.

301
00:46:48.620 --> 00:46:51.200
Lynne Jones: With the presentation up to this point, but…

302
00:46:51.300 --> 00:46:53.729
Lynne Jones: We'll move on to TOs a little bit late, maybe.

303
00:46:55.390 --> 00:46:59.700
Lynne Jones: Okay, so if there's any questions… Or comments?

304
00:47:00.170 --> 00:47:02.730
Lynne Jones: About this part, we can take them here.

305
00:47:12.150 --> 00:47:13.970
Lynne Jones: R&I, okay.

306
00:47:19.130 --> 00:47:23.550
Lynne Jones: Well, in which case… I'm on time.

307
00:47:24.940 --> 00:47:37.379
Lynne Jones: So, targets of opportunity. We have the… I talked about the target of opportunity tier, and…

308
00:47:37.500 --> 00:47:39.499
Lynne Jones: What it,

309
00:47:39.950 --> 00:47:48.049
Lynne Jones: just briefly, like, how it works. Like, we have these… this list of scripted surveys that are all in the TOO tier,

310
00:47:48.240 --> 00:47:51.780
Lynne Jones: They would compete against each other, but it's just that we don't…

311
00:47:51.920 --> 00:47:55.410
Lynne Jones: expect to get, multiple TOs at the same time.

312
00:47:58.100 --> 00:48:01.280
Lynne Jones: If we did, they would just queue up behind each other.

313
00:48:02.030 --> 00:48:07.479
Lynne Jones: Up to 3% of the survey time may be directed to TOL follow-up.

314
00:48:07.600 --> 00:48:14.719
Lynne Jones: The general plans for TO follow-up at present are well described in this, 2024 TO Workshop report.

315
00:48:15.010 --> 00:48:16.319
Lynne Jones: There…

316
00:48:16.730 --> 00:48:27.360
Lynne Jones: There is another follow-up report, so Sean McBride has been doing a lot of work with the observatory,

317
00:48:27.700 --> 00:48:29.200
Lynne Jones: To, to sort of…

318
00:48:29.440 --> 00:48:41.190
Lynne Jones: pull in the different TOL workshop report, and the very few cases that we didn't simulate, because they were over large amounts of area.

319
00:48:41.870 --> 00:48:45.040
Lynne Jones: And so…

320
00:48:45.220 --> 00:48:53.589
Lynne Jones: there is an update for this somewhere, and I have heard that there was a workshop this year that will probably be,

321
00:48:53.880 --> 00:48:58.059
Lynne Jones: Talking… looking into updating these follow-up plans as well.

322
00:48:58.280 --> 00:49:07.089
Lynne Jones: But what we have now, does match the TOL workshop, and also some… I'll find the link for the other documentation.

323
00:49:07.640 --> 00:49:11.590
Lynne Jones: We have a Rubin TO Advisory Board,

324
00:49:12.120 --> 00:49:19.020
Lynne Jones: This is a link to their documentation, their charge, and they provide guidance regarding specific TO events.

325
00:49:22.180 --> 00:49:29.340
Lynne Jones: Much of the TO process is intended to be automated, so we get an alert from an external program.

326
00:49:29.580 --> 00:49:42.500
Lynne Jones: It's evaluated by, the Rubin TOL Producer, which is a separate, not, not, not FBS scheduler, but it's a, Summit piece of software and framework.

327
00:49:42.830 --> 00:49:52.559
Lynne Jones: It evaluates the… the TOL producer evaluates the alert, and if it meets criteria, then it passes the alert information into

328
00:49:53.240 --> 00:50:00.980
Lynne Jones: the FBS, eventually. It goes through some systems, like the engineering facilities database, the EFD at the summit, and then it gets to the FBS.

329
00:50:01.800 --> 00:50:07.249
Lynne Jones: And so one of the FBS2O surveys, upon receiving that alert.

330
00:50:07.690 --> 00:50:14.410
Lynne Jones: We'll create the list of desired observations that match the predefined plan for that kind of table.

331
00:50:14.660 --> 00:50:17.720
Lynne Jones: And then… off we go. So…

332
00:50:18.210 --> 00:50:28.119
Lynne Jones: This… if the criteria are defined appropriately, this can all happen automatically and quickly without needing a human in the loop.

333
00:50:30.850 --> 00:50:40.540
Lynne Jones: The external events that are currently received by the TWOL producer include gravitational wave events, of the LVK alerts.

334
00:50:40.660 --> 00:50:47.119
Lynne Jones: It can obtain observations of potentially hazardous asteroids from JPL Scout.

335
00:50:48.400 --> 00:50:53.370
Lynne Jones: high-energy neutrinos from IceCube, and

336
00:50:53.900 --> 00:50:59.050
Lynne Jones: the galactic supernova from Super Kami Okande.

337
00:50:59.390 --> 00:51:09.989
Lynne Jones: The… there's another event producer that they're working on adding. It should be within the next year or so, but is not available yet.

338
00:51:11.380 --> 00:51:12.230
Lynne Jones: No.

339
00:51:12.740 --> 00:51:24.229
Lynne Jones: that's the automated route, but if the… if there is an alert or some kind of event that Ruben should follow, but is not one of these automated,

340
00:51:25.110 --> 00:51:31.910
Lynne Jones: sets of events. Then, the TO Advisory Board, whose members are listed here.

341
00:51:32.120 --> 00:51:40.460
Lynne Jones: will activate, and they will sort of activate that to you all, and they can provide guidance, on what to do here.

342
00:51:40.670 --> 00:51:41.760
Lynne Jones: So…

343
00:51:43.400 --> 00:51:56.880
Lynne Jones: The TO Advisory Board is intended to provide general guidance on the TO program on short timescales, like less than 24 hours for these things that don't fit into our current criteria and wouldn't match the

344
00:51:56.880 --> 00:52:04.290
Lynne Jones: automated response. You can do medium time scale response for

345
00:52:05.000 --> 00:52:08.420
Lynne Jones: Perhaps following up a multi-day, TO.

346
00:52:08.530 --> 00:52:21.149
Lynne Jones: And then they also do feedback on longer timescales, like weeks and months. And so one of the things, obviously, that we will have to do is keep track of how the TO events are…

347
00:52:21.970 --> 00:52:30.889
Lynne Jones: performing and how they're matching up with the rest of the survey, so that if we need to go and change the criteria for the automated triggering, then we can do that.

348
00:52:35.670 --> 00:52:43.290
Lynne Jones: The general TOL workflow looks like this, so we have our event… external event producers, LVK, IceCube.

349
00:52:43.430 --> 00:52:45.550
Lynne Jones: Super K JPL.

350
00:52:46.790 --> 00:52:53.080
Lynne Jones: And then that goes into SIMA. So Sima Hopscotch is where we receive the alerts from.

351
00:52:53.090 --> 00:53:07.550
Lynne Jones: Then this is part of the TOL producer, so it parses the alert for quality. Does it meet criteria, or does it meet criteria? Does it fail? If it's failed, it's just rejected. If it passes the quality criteria.

352
00:53:07.600 --> 00:53:15.590
Lynne Jones: Then it goes to the Rubin EFD. Then it goes to the Scheduler CSE, which is the framework that the FBS runs within at the summit.

353
00:53:16.080 --> 00:53:24.360
Lynne Jones: And then this scheduler creates an observing strategy. This is the FBS, creates this observing strategy, and then

354
00:53:25.120 --> 00:53:29.520
Lynne Jones: passes that back to the scheduler CSC, so Ruben executes the observation.

355
00:53:30.540 --> 00:53:39.920
Lynne Jones: And then afterwards, we have all of our processing to understand what the alert is. I pulled this, image from, Sean McBride's SBIE paper for this year, so…

356
00:53:40.620 --> 00:53:42.070
Lynne Jones: This is hot off the press.

357
00:53:43.640 --> 00:53:49.619
Lynne Jones: And of course, the TO Advisory Committee can decide to observe a non-standard TO

358
00:53:52.710 --> 00:54:03.399
Lynne Jones: Yeah. During commissioning, we did get a chance to test this workflow. It was, not as automated as some of the previous

359
00:54:05.320 --> 00:54:15.049
Lynne Jones: as the previous steps may have implied. But that was kind of on purpose, because we were in commissioning, and we didn't want to just be passing through alerts unexpectedly, so…

360
00:54:15.560 --> 00:54:23.330
Lynne Jones: every, alert that was actually passed to the FBS was done by hand, so we didn't get more than we expected.

361
00:54:24.080 --> 00:54:27.340
Lynne Jones: We tested the TWO workflow, of…

362
00:54:28.050 --> 00:54:33.810
Lynne Jones: passing the alerts through the EFD and into the FBS, on two…

363
00:54:34.840 --> 00:54:53.919
Lynne Jones: two, well, actually three, but let's… let's focus on these two, which got, more observations. So there was, 3i Atlas, and this was a very successful TOL program, and, resulted in, I think, probably the first paper coming out of,

364
00:54:54.530 --> 00:54:55.990
Lynne Jones: ribbon commissioning.

365
00:54:56.200 --> 00:54:58.380
Lynne Jones: in the SV survey over the summer.

366
00:54:58.680 --> 00:55:03.869
Lynne Jones: This was… this would definitely have been a non-standard TO, no matter what,

367
00:55:04.450 --> 00:55:13.919
Lynne Jones: But we did… we did observe it several times. You can see the, observation footprint here. This little spot here was,

368
00:55:14.030 --> 00:55:21.509
Lynne Jones: We were testing this by hand, remember? And this was not a mistake, it was just that, we obs…

369
00:55:21.820 --> 00:55:30.770
Lynne Jones: On one night, we tried to observe this field, and our masks in the FBS did their job and said, you cannot observe that field right now because it's too close to the moon.

370
00:55:31.210 --> 00:55:40.399
Lynne Jones: And so they said, okay, we still want to test the TOO procedure, so we're just going to put a different pointing center into the information we pass to the FBS.

371
00:55:40.430 --> 00:55:58.050
Lynne Jones: So we have this thing here, which we should have updated the labeling for, but we didn't, and so the commissioning, right? But I thought it was really great, because it also illustrated the FBS is doing its job with the masks, and it's making sure that we're not,

372
00:55:58.200 --> 00:56:02.989
Lynne Jones: We're not doing things that this… that the telescope is not supposed to do.

373
00:56:04.520 --> 00:56:10.910
Lynne Jones: On the right was our bigger follow-up, TO, so this was…

374
00:56:11.980 --> 00:56:16.769
Lynne Jones: Gravitational wave candidate, S251112.

375
00:56:17.060 --> 00:56:17.950
Lynne Jones: CM.

376
00:56:20.920 --> 00:56:30.500
Lynne Jones: we do have some work to do on our metadata, because it came through looking at, like, one. But we actually… we fixed that already, we just haven't had another chance to test it.

377
00:56:30.630 --> 00:56:45.889
Lynne Jones: So this TO was a gravitational wave that had a very large footprint, and it covered the whole sky, almost through… through here. This is the… what we deemed at the time to be the high priority area of the footprint.

378
00:56:45.950 --> 00:56:51.769
Lynne Jones: with some over here. So this, in this case, this was not a mistake, it was just a very far part of the footprint.

379
00:56:53.130 --> 00:56:56.620
Lynne Jones: And we observed this TO for a couple of nights.

380
00:56:56.810 --> 00:56:58.820
Lynne Jones: In G&I band.

381
00:56:59.290 --> 00:57:10.840
Lynne Jones: And so we followed this area, we did it again, we did it again. And so this was following the planned observations. One of the things about the…

382
00:57:10.840 --> 00:57:19.820
Lynne Jones: 3i Atlas being a non-standard TOO is that we had to actually put in the observations we wanted, because it didn't really,

383
00:57:21.900 --> 00:57:25.299
Lynne Jones: We had to make it look like one of the cases we expected.

384
00:57:25.750 --> 00:57:32.750
Lynne Jones: Okay, and over here, this one did, so we followed it. What we found was that there are…

385
00:57:33.880 --> 00:57:51.950
Lynne Jones: there were some details about how the FBS runs at the summit that are a little bit different than how it runs when we run it in a simulation. And so there were some things… there were definitely things we learned from this experience. It was also really good to see us follow this large area.

386
00:57:51.950 --> 00:57:59.219
Lynne Jones: I think it has proved difficult to process this without, like, production-grade templates.

387
00:57:59.450 --> 00:58:04.350
Lynne Jones: Yeah, and so we're… we'll see how that… Gentles.

388
00:58:05.990 --> 00:58:07.700
Lynne Jones: Okay.

389
00:58:10.060 --> 00:58:12.140
Lynne Jones: So, we have,

390
00:58:15.410 --> 00:58:20.250
Lynne Jones: Excuse me. We have… Children

391
00:58:20.600 --> 00:58:29.169
Lynne Jones: So, we did the gravitational wave case, and we learned things, and we're implementing updates for this in the FPS.

392
00:58:31.620 --> 00:58:37.259
Lynne Jones: And then we have questions for any of this so far.

393
00:58:43.070 --> 00:58:44.360
Lynne Jones: Roberta?

394
00:58:45.630 --> 00:58:55.059
Roberto Assef: Yes, thank you very much. Just a couple of very quick questions. So, is there, thinking… I'm thinking mostly of gravitational waves

395
00:58:55.060 --> 00:59:05.799
Roberto Assef: follow-up, is there a ramp-up, considering the number of requests towards the later years? Because I will assume that, sort of, this will keep increasing and increasing.

396
00:59:06.850 --> 00:59:09.790
Lynne Jones: It kind of goes, with the…

397
00:59:09.950 --> 00:59:13.680
Lynne Jones: With the observing seasons for,

398
00:59:15.050 --> 00:59:30.899
Lynne Jones: For LIGO, Virgo, for… so they're not actually spread evenly throughout the whole survey, and they are concentrated in… I think it's year 2. Peter might have a better answer on when exactly it is.

399
00:59:31.020 --> 00:59:38.100
Lynne Jones: They are concentrated in time. This is also something we'll have to consider, because obviously.

400
00:59:38.330 --> 00:59:48.510
Lynne Jones: we've said, okay, this is when we think that the alerts will be coming in, and we've based that time span and the rates on what came out of the TOO workshop.

401
00:59:48.750 --> 00:59:53.060
Lynne Jones: But… We don't know, for sure.

402
00:59:54.440 --> 01:00:13.109
Roberto Assef: Okay, and… and can I ask one more question? The… so, right, a lot of what we… what happened by the change to 5.3 was this… the last 10% lower on Sky time. So I'm wondering, is it too difficult to, sort of.

403
01:00:13.130 --> 01:00:30.990
Roberto Assef: after the fact, see if the images obtained for TO could fit within the strategies of the WFD and the DDFs. It could add to zero, because we don't know where the TOOs will be and when they will be, but if they add, that will help recover some of that.

404
01:00:32.130 --> 01:00:40.870
Lynne Jones: So… Potentially. The… I mean, and the…

405
01:00:41.730 --> 01:00:50.520
Lynne Jones: It's difficult because, when we're observing, like, with the PARS survey, or, all the other

406
01:00:50.710 --> 01:01:06.609
Lynne Jones: tiers that go into making our normal observing mode. We don't really have, a way to force it to say, okay, on tonight, actually really go do this part of the sky. So, you would still be activating a TOO, essentially.

407
01:01:06.930 --> 01:01:08.140
Lynne Jones: And…

408
01:01:08.480 --> 01:01:18.200
Lynne Jones: we… what it would be is maybe what we would… could do is make the TO gentler, but that's kind of the opposite of what you want for the TOO, right?

409
01:01:19.770 --> 01:01:36.559
Lynne Jones: So… so there… yes, there is this tension between, like, following up the TOO as it… as it ideally would be, and as quick as it would be, versus, like, doing it in a way that might just look like the normal survey.

410
01:01:37.210 --> 01:01:44.559
Lynne Jones: And… the… looking like the normal survey is… I'm not…

411
01:01:46.190 --> 01:02:04.999
Lynne Jones: Yeah. We did have… we… the reason we… so, originally, when we made TOs, we counted them towards the… the footprint for the pairs survey, but, what we found was that disturbs the cadence we're trying to get with the pairs as well. So, it's kind of like they…

412
01:02:05.020 --> 01:02:11.530
Lynne Jones: Don't go together super well, but certainly, you know, if this turns out to be a really big

413
01:02:11.720 --> 01:02:22.280
Lynne Jones: problem that we really have many more TOs than we… that we need to follow up than we can accommodate. This would be something we could go and revisit, but it… it is difficult.

414
01:02:23.450 --> 01:02:25.690
Lynne Jones: And I'm not sure it'd be successful.

415
01:02:27.170 --> 01:02:27.870
Roberto Assef: Thank you.

416
01:02:29.370 --> 01:02:30.380
Lynne Jones: Bindu?

417
01:02:31.480 --> 01:02:44.719
Bindu Rani: Yeah, hi, Lynn. Really quick, is there a fixed amount of time assigned for the TO observations? And, if it is, is there a plan, like, you know, in case if that time has not been used,

418
01:02:45.020 --> 01:02:46.710
Bindu Rani: How that time will be used.

419
01:02:46.950 --> 01:02:47.880
Bindu Rani: Otherwise.

420
01:02:48.450 --> 01:02:53.650
Lynne Jones: Yeah, good question. It's not… there is a fixed cap… well.

421
01:02:54.070 --> 01:03:05.430
Lynne Jones: in theory, a fixed cap is… we don't enforce it in the FBS. To enforce the cap on the TOs, it would have to come into, like, the criteria for triggering on the TOs.

422
01:03:05.750 --> 01:03:13.830
Lynne Jones: So… The TUL program is expected to use up to 3% of the total survey time.

423
01:03:14.520 --> 01:03:31.369
Lynne Jones: there's nothing that, like, if they… there are no TOOs, it's not like we're not observing. We just don't observe a TOO, and we do observe all the normal stuff we would observe, so all of the other tiers lower in the scheduler just would activate.

424
01:03:31.570 --> 01:03:46.380
Lynne Jones: And, like I was saying yesterday, our footprint is just a ratio, so if we have more time, we just get more observations distributed over the footprint, and if we have less time, we just get less observations distributed over the footprint.

425
01:03:46.630 --> 01:03:47.280
Bindu Rani: I still think.

426
01:03:47.280 --> 01:03:53.830
Lynne Jones: So, it just… if we… if we didn't have any TOs, we would have a little bit more time

427
01:03:54.280 --> 01:03:58.640
Lynne Jones: That would go into everything else in the survey, and if we had more TOOs.

428
01:03:59.080 --> 01:04:01.250
Lynne Jones: Then it takes the time.

429
01:04:02.900 --> 01:04:03.460
Bindu Rani: Thanks.

430
01:04:06.680 --> 01:04:07.650
Lynne Jones: Rachel?

431
01:04:09.870 --> 01:04:17.370
Rachel Mandelbaum: Yeah, hi, Lynn. Just to… to add to that and make sure I've understood properly, especially given an answer you gave to another question.

432
01:04:17.510 --> 01:04:27.969
Rachel Mandelbaum: Is it correct that, like, you know, since We do anticipate that The number of gravitational wave…

433
01:04:28.270 --> 01:04:30.680
Rachel Mandelbaum: Events will depend on, like, weather.

434
01:04:31.050 --> 01:04:35.870
Rachel Mandelbaum: whether there's, like, a LIGO ROGO CAGRA observing run on or not.

435
01:04:36.120 --> 01:04:36.580
Lynne Jones: They're.

436
01:04:36.580 --> 01:04:43.649
Rachel Mandelbaum: There might be some years where the TOO time could exceed the 3%, and the 3% is more across the whole survey.

437
01:04:43.870 --> 01:04:44.300
Lynne Jones: Yeah.

438
01:04:46.890 --> 01:05:04.580
Lynne Jones: Yeah, the… I mean, we do expect it to be variable, because most of the time in the TOs is for these very large events, which are not as frequent, so they will be bunched up. Sorry, not bunched up, but the TO time overall will be bunchy.

439
01:05:05.400 --> 01:05:09.810
Lynne Jones: I don't think we have a…

440
01:05:10.340 --> 01:05:16.210
Lynne Jones: cap on how much TOL time can be used within a particular time frame.

441
01:05:16.380 --> 01:05:21.079
Lynne Jones: And that… that is something that I think the SEOC have…

442
01:05:21.340 --> 01:05:30.740
Lynne Jones: decided to think about, but it is hard to make progress on that without understanding what the actual,

443
01:05:31.480 --> 01:05:35.500
Lynne Jones: event rate is going to be, because there's… there's uncertainty on it, so…

444
01:05:36.690 --> 01:05:45.689
Lynne Jones: And it's also hard, because it is hard then also to say, okay, what effect will that have on the rest of the observations? So…

445
01:05:46.290 --> 01:05:54.570
Lynne Jones: Yeah, I do think this is a thing that we need to think about, because it would go back to the,

446
01:05:55.030 --> 01:06:10.820
Lynne Jones: to the TO Advisory Board, and they would say, hey, this… this time is being used too… too quickly, and we should… we should change the criteria. Igor, and you are on the, TO Advisory Board, so hopefully you can say a few words.

447
01:06:10.990 --> 01:06:17.739
Igor Andreoni: Yeah, maybe just complimenting a little bit, all you've said was in response to Rachel.

448
01:06:18.160 --> 01:06:19.190
Igor Andreoni: the…

449
01:06:19.760 --> 01:06:31.860
Igor Andreoni: We have tried to rally up the community, bringing it together to discuss and develop strategies and come up with realistic budgets.

450
01:06:32.050 --> 01:06:36.700
Igor Andreoni: for TU observation requests. So this was done in 2024.

451
01:06:36.870 --> 01:06:44.269
Igor Andreoni: And we are doing it again right now. Hopefully, we will have the new documents published by the end of July.

452
01:06:44.390 --> 01:06:46.430
Igor Andreoni: So, I would suggest…

453
01:06:46.790 --> 01:06:54.749
Igor Andreoni: If you want to have, like, a preview, you can check on the archive. We have put the report from 2024.

454
01:06:54.860 --> 01:07:00.400
Igor Andreoni: I think if you search the Rubin TO program on ADS, should be the first thing that comes up.

455
01:07:00.730 --> 01:07:05.870
Igor Andreoni: Otherwise, waiting a month, you will have more solid numbers.

456
01:07:06.080 --> 01:07:15.759
Igor Andreoni: These numbers that we've come up with, try to be more or less within 3% over the course of the whole survey.

457
01:07:15.970 --> 01:07:21.110
Igor Andreoni: We are trying to be very, mindful of,

458
01:07:21.600 --> 01:07:34.260
Igor Andreoni: minimizing disruption, of course, and when I say we, I really mean everyone who is interested in TOs, from gravitational waves to other types of messengers, to, potentially hazardous asteroids.

459
01:07:34.850 --> 01:07:36.649
Igor Andreoni: In addition.

460
01:07:37.080 --> 01:07:54.649
Igor Andreoni: at the moment, the go-to policy is whenever a counterpart we're searching for is found, we stop observations, so that the other telescope with a smaller field of view can perform the follow-up if they're available. So there are…

461
01:07:54.920 --> 01:07:57.780
Igor Andreoni: I think…

462
01:07:58.620 --> 01:08:11.529
Igor Andreoni: There are ways we're actually perform… gonna perform the observations that are maybe not fully encapsulated in the simulations, because again, they're so… honestly, so difficult to predict.

463
01:08:11.690 --> 01:08:16.240
Igor Andreoni: But my hope is that the…

464
01:08:16.810 --> 01:08:20.969
Igor Andreoni: We will use the time well in a very efficient manner.

465
01:08:21.220 --> 01:08:27.159
Igor Andreoni: Possibly all the… what's available, and if not, observation will go towards

466
01:08:27.330 --> 01:08:34.739
Igor Andreoni: The minimum… all the meaningful science that can be achieved with the regular survey.

467
01:08:35.050 --> 01:08:51.330
Igor Andreoni: And all of this keeps changing. The rates, for example, for gravitational wave sources keep changing as more and more light runs are performed. And at the same time, new ideas for interesting science cases are brought up by the community. So we…

468
01:08:51.330 --> 01:09:02.239
Igor Andreoni: Every couple of years, we'll try to rethink a little bit how to structure the program and try to make it as effective and efficient as possible.

469
01:09:09.229 --> 01:09:10.109
Lynne Jones: Thanks, Igor.

470
01:09:10.539 --> 01:09:13.309
Lynne Jones: I'm glad you… I'm glad you, got a…

471
01:09:13.429 --> 01:09:16.949
Lynne Jones: Though we got to hear from somebody from the TO Advisory Board.

472
01:09:17.439 --> 01:09:18.329
Lynne Jones: Yay.

473
01:09:18.839 --> 01:09:21.839
Lynne Jones: Alright.

474
01:09:22.879 --> 01:09:27.659
Lynne Jones: And now we can move… oh, sorry, go ahead, Rob. We can take one more question here.

475
01:09:29.460 --> 01:09:43.779
Saurabh Jha: Yeah, I just wanted to follow up on, a little bit on Roberta's question. One of the things about the TOs, and I just checked in with Peter about this, was, some of the observations are using non-standard visits, so the exposure times might go up to 120 seconds in the TOOs.

476
01:09:44.180 --> 01:09:55.970
Saurabh Jha: And so, you know, whether those could be included in what becomes Wide Fast Deep, yeah, I mean, maybe, probably, but, you know, they are a little bit different than the other observations in the survey, or can be.

477
01:10:02.470 --> 01:10:12.859
Lynne Jones: Alright, excellent. And, let's move on to talking about the templates and the templates here. Eric…

478
01:10:13.060 --> 01:10:16.419
Lynne Jones: Are you… Were you,

479
01:10:17.580 --> 01:10:33.699
Lynne Jones: Oh, great. Okay. I… I… I didn't… I wasn't quite sure where to put in your slides. I was just going to say a little bit about our new option for the template tier, and then, hand… hand over to you for that.

480
01:10:33.700 --> 01:10:35.930
Eric C Bellm: That's fine, how much time do you want me to spend?

481
01:10:40.120 --> 01:10:48.399
Lynne Jones: You… you said you didn't want very long, so… I'm thinking you can… somewhere between 5 and 10 minutes.

482
01:10:48.720 --> 01:10:50.269
Eric C Bellm: Okay, that sounds fine.

483
01:10:51.190 --> 01:10:54.799
Lynne Jones: And some of that might be questions, if that works better.

484
01:10:55.130 --> 01:10:55.980
Lynne Jones: Okay.

485
01:10:56.100 --> 01:11:03.300
Lynne Jones: So, some of this I… Eric is gonna just say yes or no about, but basically.

486
01:11:03.450 --> 01:11:18.049
Lynne Jones: One of the issues we have is good template creation requires at least 5 to 6 visits, with the full-width half mics and transparency, required for the template co-add, and that's…

487
01:11:18.790 --> 01:11:25.789
Lynne Jones: Awesome. So they… so they… these are declination-dependent and band-dependent values.

488
01:11:26.650 --> 01:11:43.410
Lynne Jones: Now, the other thing is, why can't we just create bad templates and keep going? Well, bad templates are bad, because, replacing templates is bad for the… it adds offsets in the magnitudes of the alerts, so the template continuity over time is strongly preferred.

489
01:11:44.340 --> 01:11:59.090
Lynne Jones: And then the other thing is bad templates make more alerts, which then make even more alerts because of how, alert… the alert, prompt processing pipeline works, and this basically just loads up the prompt processing database with noise.

490
01:11:59.630 --> 01:12:00.500
Lynne Jones: And…

491
01:12:00.810 --> 01:12:09.850
Lynne Jones: So, we don't want to just make bad templates and keep going, so we want to make good templates, and then the problem is, if we don't

492
01:12:10.270 --> 01:12:20.599
Lynne Jones: set up the FBS to sort of have a specific push to acquire the 5 to 6 visits with good image quality and good transparency.

493
01:12:20.750 --> 01:12:34.199
Lynne Jones: at each point on the sky per band, there is a high likelihood that we would not acquire those necessary visits within year one. And this is not just a high likelihood, but almost certain in UNG with our… without the template here.

494
01:12:35.170 --> 01:12:39.999
Lynne Jones: So, this is why we added the template tier to the baseline surveys, so that we meet this need.

495
01:12:40.130 --> 01:12:50.109
Lynne Jones: Now, the… the thing we immediately did was we just put in templates… a template tier which… which acquired visits and singles, but because there's so many visits in this

496
01:12:50.280 --> 01:12:52.300
Lynne Jones: Tier in year one.

497
01:12:52.480 --> 01:13:00.680
Lynne Jones: Singles were not a good choice, so I keep saying we acquire the templates in pairs, because that actually was an important consideration.

498
01:13:05.190 --> 01:13:22.389
Lynne Jones: The next question is, can we improve on the template acquisition? Because there is still a risk that we would not complete the templates over the whole… the template inputs, we don't make the templates, that we would not get enough template inputs over the whole sky at the end of year one.

499
01:13:23.540 --> 01:13:37.990
Lynne Jones: So we have, a simulation called Faster Templates. This is in our V5.3 set of simulations. And the idea here is we need epics with dithering for the pixel coverage.

500
01:13:38.380 --> 01:13:54.149
Lynne Jones: Individually, they could be shallower, potentially, although they shouldn't be worse quality full-width half max, but because they are good quality full-width half max, perhaps they can just be a little bit shorter exposures, and it doesn't make them that much, shallower in depth.

501
01:13:55.390 --> 01:14:03.139
Lynne Jones: So… This simulation poses a question, what if we take the template image, the template visits.

502
01:14:03.520 --> 01:14:09.770
Lynne Jones: As 20-second images, increasing the number of epics,

503
01:14:09.970 --> 01:14:23.109
Lynne Jones: As… versus what we would have if we just took them as 30-second visits. Now, so, U-band is special, as always, because U-band is 38 seconds in a standard visit, and it would go down to 25 seconds in these shorter visits.

504
01:14:24.540 --> 01:14:39.610
Lynne Jones: Now, this is not just for us to say, because there's lots of impacts to everybody else, so we have an RFC, a request for comment, in progress within Ribbon to evaluate the possibility and impacts of making this change.

505
01:14:41.070 --> 01:14:57.969
Lynne Jones: And I will just note that 70% of the first-year images in the simulation are acquired with the template tier. So, when we say, oh, you can just take some 20-second images, we actually mean the majority of the images in year one would be these shorter images.

506
01:15:01.780 --> 01:15:19.580
Lynne Jones: the impacts on our metrics. So, if you remember this grid from yesterday, we have the high-level, metrics across a bunch of different science areas. So, these are, at the bottom, are the, SRD metrics, and we have things that go into,

507
01:15:19.710 --> 01:15:23.099
Lynne Jones: Cosmology, and then time domain.

508
01:15:23.650 --> 01:15:33.149
Lynne Jones: static science of galaxies and solar system objects. And this is all normalized to, be 5.3.

509
01:15:33.400 --> 01:15:41.200
Lynne Jones: And at the… here we have what happens if we go to these 20-second templates, so the 20-second fast template simulation.

510
01:15:44.100 --> 01:15:54.200
Lynne Jones: some things get better, some things get worse. It's not obvious we've completely captured the effect in our metrics, but one thing that is for sure is that

511
01:15:54.300 --> 01:16:01.139
Lynne Jones: On the right here, what we have is the night that you have captured a given amount of area.

512
01:16:01.410 --> 01:16:12.390
Lynne Jones: while getting, at least 4 visits per band onto, silicon. So… This is math.

513
01:16:12.550 --> 01:16:17.439
Lynne Jones: Evaluating the camera footprint on our visit bore sites.

514
01:16:18.150 --> 01:16:25.020
Lynne Jones: Which is not quite the same as how, as I mentioned yesterday, as how data management would count

515
01:16:25.640 --> 01:16:33.730
Lynne Jones: whether the… detector was usable to go into the co-add, because they have,

516
01:16:34.280 --> 01:16:41.860
Lynne Jones: probably different constraints on the edges and things like that. So, this is our estimate, but it's not the final answer.

517
01:16:42.270 --> 01:17:00.899
Lynne Jones: So this is… if we have this deck-dependent seeing limit, and we chose… we said it's a limit of 1.2 arcseconds at zenith, and how many, how much area would you have where you got 4 visits onto silicon and… per band?

518
01:17:01.500 --> 01:17:15.570
Lynne Jones: And so what you see is this is the total area of the survey, and if… with, with the Comp Survey V5.3, that's the one where we used the old downtime, so we don't have our extra downtime. It's blue.

519
01:17:15.750 --> 01:17:25.209
Lynne Jones: The green line here that's lower is our baseline 5.3. So this is the effect of not having as many visits in year one.

520
01:17:26.270 --> 01:17:40.339
Lynne Jones: And then the red line is what happens if we, take those visits and make them shorter. So we… with a shorter visit, we do have a little bit less duty cycle on the sky, but we still end up with more visits, and so…

521
01:17:41.600 --> 01:17:53.240
Lynne Jones: If we were to do that, and all of the things, like the visit gaps remain the same, then we would push back up to, the same number of visits as we would have had without the extra downtime.

522
01:17:53.670 --> 01:17:58.249
Lynne Jones: And, so this is… this is close to you. This is…

523
01:17:58.360 --> 01:18:03.329
Lynne Jones: Essentially getting templates over the sky by the end of year one.

524
01:18:04.180 --> 01:18:15.360
Lynne Jones: Now, like I said, this is uncertain. You know, you can't say, oh, if we don't do this, we won't close the sky for the templates, and if we do, then we will, but that's…

525
01:18:15.450 --> 01:18:26.259
Lynne Jones: And if you said instead of 4 per year, you said 5… sorry, 4 per band, you said 5 per band, the difference between these would scale similarly, but,

526
01:18:26.600 --> 01:18:31.800
Lynne Jones: It's true that the overall Area would be slightly different.

527
01:18:34.110 --> 01:18:39.370
Lynne Jones: Right, so from the… that's our… from the point of view of metrics. Now, is this…

528
01:18:39.720 --> 01:18:46.590
Lynne Jones: So now I'm going to hand it over to Eric to talk about actual templates, and did you have slides of your own that you wanted to talk about?

529
01:18:46.590 --> 01:18:47.649
Eric C Bellm: Yeah, I can share.

530
01:18:47.650 --> 01:18:50.250
Lynne Jones: I put this one in because I wasn't quite sure.

531
01:18:50.440 --> 01:18:56.299
Eric C Bellm: Yeah, I'll… I'll put them into the shared slide deck after I talk. I just didn't have the link in time.

532
01:18:56.710 --> 01:18:58.149
Lynne Jones: Perfect, yeah, sorry about that.

533
01:18:58.490 --> 01:19:00.269
Lynne Jones: I'm gonna stop sharing a hand to you.

534
01:19:01.590 --> 01:19:06.260
Eric C Bellm: Alright, so I'm just gonna spend a few minutes talking about,

535
01:19:06.870 --> 01:19:15.730
Eric C Bellm: templates from the data management perspective. So I, I think it's worth,

536
01:19:16.050 --> 01:19:28.930
Eric C Bellm: saying up front, you know, kind of distinguishing in our minds two separate questions. One is thinking about what we're aiming for at the end of year one, as the… for the inputs into DR1.

537
01:19:28.930 --> 01:19:52.219
Eric C Bellm: which has been kind of targeted as our source of, you know, good all-Sky templates, and so that's, I think, especially driving some of the questions around the template tier that Lynn is describing. You know, how do we get to an endpoint where we have enough inputs of good quality over the whole footprint that we can co-add. Separately, you know, of course.

538
01:19:52.220 --> 01:20:06.089
Eric C Bellm: I, like all of you, am extremely excited to get alert science going, early in the survey, in this first year, and so for that, we need to be even more, sort of, aggressive in getting templates deployed.

539
01:20:06.180 --> 01:20:19.509
Eric C Bellm: how do we manage… how do we maximize that sort of science ahead of DR1? And so there, I think it's clear, as you're gonna see in a minute, you know, we have to… we have to accept some compromises of some kind, and it's just a question of

540
01:20:19.510 --> 01:20:29.959
Eric C Bellm: It's kind of an engineering and science trade-off of where do we cut the corners, because it's… we don't have as many images as we'd like, and we have to kind of figure out what to do about that.

541
01:20:30.340 --> 01:20:50.750
Eric C Bellm: All right, there are kind of two technical constraints that really, influence what we do on template generation. The one is for alert production. There's a database we call the APDB, the Alert Production Database, which serves as sort of the core of the association and light curve generation, sort of.

542
01:20:50.880 --> 01:21:15.719
Eric C Bellm: For alerts, and it's very sensitive to, false positive rate. If we… if we pump it full of junk sources, it will… it will break down. It will… it will reach capacity and not function. We have approached that on a few occasions and, learned from the process, but, you know, we have to be thoughtful about it. And I would say, probably as science users, you also don't want just

543
01:21:15.720 --> 01:21:40.669
Eric C Bellm: cascades of bad detections either. There's lots of knobs we're learning to tune here, you know, both in things like Real Bogus and how we configure image subtraction, but I just want to put out front that this is something that we really do care about a lot, and as you'll see in a minute, templates are a major source of false positives. So, the other challenge is that it is not as easy

544
01:21:40.670 --> 01:22:03.500
Eric C Bellm: as either it sounds, or we'd like it to be to change the templates frequently. The processing takes a while, frankly, and so it's not easy to do a big template run and just update things. I also think, scientifically, most users don't actually want that, because when you change the templates, all the…

545
01:22:03.850 --> 01:22:09.870
Eric C Bellm: All the different slight curves are gonna shift. You get different amounts of contamination from

546
01:22:09.870 --> 01:22:30.859
Eric C Bellm: transients in the… in the template for supernovae and things like that, and for variables. So, all of that just says that, like, you know, we will change templates over time, but we're not gonna thrash them, especially in the first year. So, we're trying to do the best job we can, you know, put templates on and keep them there for, you know, of order a year.

547
01:22:31.760 --> 01:22:55.759
Eric C Bellm: All right, so since we can't update the templates a lot, we want to try to minimize gaps where we can, because if you have a hole in the template, you can't fill it for a while, you get no alerts from there at all. The fill factor of the camera, in terms of, when you project it into the template space, it's about 88%, and so if you just use single images, then you're gonna just have

548
01:22:55.760 --> 01:22:57.940
Eric C Bellm: 12% holes. So…

549
01:22:58.280 --> 01:23:17.839
Eric C Bellm: So we want multiple inputs, and what is the very minimum? If you think of these as just uncorrelated and ask, you know, how many, how many, what area doesn't have any pixels on it, you only need sort of 2 or 3, and you get real close to 100% fill factor. So that's, like, kind of the minimum.

550
01:23:17.920 --> 01:23:30.060
Eric C Bellm: like, baseline to if you want to avoid holes. There are some other image quality and image subtraction issues, which we'll discuss there, but in terms of, like.

551
01:23:30.170 --> 01:23:39.209
Eric C Bellm: What's… what is the bare minimum for year one if… if we accept that we don't want to just leave giant holes in the survey, then… then there's a few images to input?

552
01:23:39.750 --> 01:23:44.200
Eric C Bellm: And again, those should be good quality images, they shouldn't have clouds in them, things like that.

553
01:23:45.000 --> 01:23:48.510
Eric C Bellm: As we add images, we do better.

554
01:23:48.550 --> 01:24:06.869
Eric C Bellm: with 3 or more images, we can do something, we can do artifact rejection. We have some algorithms that, especially work nicely when you get lots of images, where we… it's, able to reject, footprint… detection footprints that only show up in a configurable fraction of the inputs.

555
01:24:06.870 --> 01:24:10.690
Eric C Bellm: As you can see, on the right here, the difference between

556
01:24:10.690 --> 01:24:28.380
Eric C Bellm: sort of, you know, mean and sigma clip coads and this, what we call compare warp rejection, which is great for getting rid of things like satellite trails and glints and so on. So, this is, what we, you know, why we push for more inputs when we have them.

557
01:24:29.360 --> 01:24:42.029
Eric C Bellm: So again, we're trying to get to that sort of greater than 3 images, and that's where we sort of talked to Lynn about, can we get to 5, 6 sort of images to, to get inputs.

558
01:24:42.410 --> 01:24:59.630
Eric C Bellm: All right, what have we done so far? We have DP2 coming, later this year, and that is, based on commissioning images from last year, which got, especially this wide trance that you see here in the ecliptic plane and crossing through, sort of, the galactic center region.

559
01:24:59.870 --> 01:25:10.660
Eric C Bellm: We… this is the, a view of the HIPS maps from the Deepco ad. The template selections are very slightly different and a little smaller, but it's useful for illustration.

560
01:25:10.660 --> 01:25:34.220
Eric C Bellm: And when you zoom down at this level, it looks great, and you're like, okay, we got templates on all that area. It's not quite that simple. When you zoom in in the middle somewhere, you see that, yes, indeed, it is pretty well covered. But at the edges, you can see, that, you know, these are sort of single epochs that have big chip gaps in them. Don't pay too much attention to the interpolation here, but you can see the…

561
01:25:34.270 --> 01:25:36.379
Eric C Bellm: The chip structure of the camera.

562
01:25:36.830 --> 01:25:46.859
Eric C Bellm: This is U-band for comparison, and in U-band, for instance, there's even larger holes, even in the middle of this… of this footprint.

563
01:25:47.200 --> 01:26:03.899
Eric C Bellm: So as Lynn, showed, we have deployed these DP2 templates in addition to the existing drilling templates. There's a dashboard that's been circulated, and we're keeping updated as we add new ones. We added a few more, from,

564
01:26:03.900 --> 01:26:07.310
Eric C Bellm: From a few weeks ago, and are continuing to do so.

565
01:26:07.310 --> 01:26:11.380
Eric C Bellm: And so our alert volumes have accordingly increased.

566
01:26:11.490 --> 01:26:17.780
Eric C Bellm: What I do want to say, and this, again, impacts the… sorry, darn it. Let me…

567
01:26:18.140 --> 01:26:21.070
Eric C Bellm: Rapidly edit this.

568
01:26:21.170 --> 01:26:22.609
Eric C Bellm: Meant to make that a…

569
01:26:22.870 --> 01:26:42.160
Eric C Bellm: animation. We do see that even the DP2 templates have some challenges. So this is a pattern of alerts that we've sent out in the last few days, and you can see there's a pretty strong grid structure in some regions of this, of the sky. And this is, we have tracked down due to,

570
01:26:42.280 --> 01:26:54.660
Eric C Bellm: when we have co-added even, you know, a relatively small number of images in this wide area, there's a wide enough range of PSF sizes that when you have chip gaps.

571
01:26:54.800 --> 01:27:07.820
Eric C Bellm: Sometimes the PSF that results in the COAD is printing through has some pretty strong spatial variations, which you can see on the right for a patch here. I apologize, the full width half max here is in pixel units.

572
01:27:07.820 --> 01:27:26.319
Eric C Bellm: Which is, you know, multiplied by 0.2, but you can kind of see how these structures, you know, have the imprint of the focal plane in full width, half max space. And so image differencing has a fair bit of trouble right now. It's sort of managing, you can't…

573
01:27:26.320 --> 01:27:31.030
Eric C Bellm: The kernel is not able to handle such rapid changes, and so you get

574
01:27:31.030 --> 01:27:42.500
Eric C Bellm: an excess of false positives. We have a number of ideas about how to deal with this, both at the vetting side and in template construction, image differencing, and so on.

575
01:27:42.500 --> 01:27:56.850
Eric C Bellm: But I guess I would just, you know, highlight the point that, you know, it's not quite as simple as, you know, just take, take in image, you know, take two images and add them and everything will be fine. It really does… we do see problems.

576
01:27:56.850 --> 01:28:07.429
Eric C Bellm: I mean, the good news is these are not such large, bogus volumes that it's overwhelming the APDB, but it is something where we feel like we need to take some corrective action.

577
01:28:08.180 --> 01:28:12.430
Eric C Bellm: All right, so we're looking ahead to incremental templates with,

578
01:28:12.480 --> 01:28:26.180
Eric C Bellm: with pre-LSST images, as well as the previous input inputs. Lynn hasn't seen these, because I just got them done this morning. So these are plots of basically all of the Rubin science images subject to some cuts.

579
01:28:26.200 --> 01:28:45.109
Eric C Bellm: And… and sort of looking where we… where we could start building incremental templates now, and so this is band by band. I'll call your attention mainly to the green boxes, which are tracks that have complete coverage of at least 3 input images, and so that would be kind of a good place to start.

580
01:28:45.430 --> 01:28:54.949
Eric C Bellm: So this is U-band, there's some, you know, maybe, I guess it's labeled here. 42 tracks are maybe plausible.

581
01:28:54.950 --> 01:29:14.720
Eric C Bellm: G, a little wider. The white areas already have templates deployed, either from the DDFs or from DP2. R-band, a little bit of coverage, sort of, in the south. It remains patchy. This… there's a region, here that maybe we could explore.

582
01:29:14.720 --> 01:29:23.750
Eric C Bellm: Iban, there's these big stripes that were taken during OES commissioning, which we can build templates in, so those are a natural place for us to start.

583
01:29:23.880 --> 01:29:30.220
Eric C Bellm: Zband, I think, is in some ways the best. There are some really large areas here that have quite a bit of coverage.

584
01:29:30.310 --> 01:29:47.160
Eric C Bellm: And then Y band a little less. So, I think we do have a bit of a head start building templates over a reasonable but not overwhelming area, but we need to sort of understand, some of the issues we're seeing with the DP2 templates.

585
01:29:47.160 --> 01:29:56.710
Eric C Bellm: In order to make sure that we're making the right choices there as we deploy them. So, happy to… I don't know if we're going to questions here, Lynn, but that's all I have for the moment.

586
01:29:59.770 --> 01:30:11.090
Lynne Jones: Yep, it was, just questions, and then we… we back up to the break next. So, actually, that was awesome, Eric, thank you so much. And all of that new stuff, very fascinating.

587
01:30:11.090 --> 01:30:25.110
Lynne Jones: I wanted to make the comment that when we know that we can build a template in a particular area of the sky, we will be able to load that into the template tier survey, so that they know that, like, okay, yeah, we…

588
01:30:25.290 --> 01:30:37.260
Lynne Jones: You may not know about those observations, because we're going to wipe its memory clean when we start the survey, but we can tell that, yes, there are already templates available for this part of the SCA.

589
01:30:38.240 --> 01:30:40.320
Lynne Jones: Alright, David.

590
01:30:41.100 --> 01:30:46.880
David Schlegel: Yeah, these DP2 template observations, are that all the range of rotation angles?

591
01:30:47.620 --> 01:30:58.659
Eric C Bellm: Yeah, that's right. It's getting the… maybe Lynn can speak better to it, but my understanding is those are widefast, deep-style dithers in rotation and translation.

592
01:30:59.340 --> 01:31:12.869
Lynne Jones: Yeah, the… so the… the stripe across the sky in the ecliptic area was mostly wide-fast-y, style. It was mostly coming from, our Paris survey.

593
01:31:12.970 --> 01:31:19.879
Lynne Jones: tier, but… The… because of the fact that, it's a smaller area on the sky.

594
01:31:19.990 --> 01:31:24.889
Lynne Jones: We… the repeat rate was a little bit higher, and

595
01:31:26.010 --> 01:31:32.259
Lynne Jones: Yeah, but other than that, it would be like our normal survey observations.

596
01:31:45.540 --> 01:31:46.870
Lynne Jones: Alright, wow.

597
01:31:47.170 --> 01:31:50.849
Lynne Jones: I really thought everybody would have questions for you, Eric, because of the…

598
01:31:51.130 --> 01:32:04.350
Lynne Jones: I'm really fascinated to see the, you know, where we can go with incremental templates, and what we might be able to do even before we start the survey, because these are really

599
01:32:04.570 --> 01:32:10.999
Lynne Jones: big questions, like, what are we going to try to do in the first year regarding templates?

600
01:32:17.530 --> 01:32:22.289
Lynne Jones: I… nobody has asked, but I'm gonna ask, because…

601
01:32:22.620 --> 01:32:27.700
Lynne Jones: I get asked, how soon can you have templates for a particular area of the sky?

602
01:32:27.930 --> 01:32:37.789
Lynne Jones: And I know from watching that there's a… can be a significant lag, because as you said, it's quite labor-intensive.

603
01:32:38.730 --> 01:32:45.580
Lynne Jones: Do you… do you want to say a few words on, like, the rate of, like, once we cover the sky, or, like…

604
01:32:46.000 --> 01:33:00.890
Eric C Bellm: Yeah, I had this in a slide and kind of brushed past it. My estimate right now is that the delta between, we know we have enough images here, and we could get this into production, is about a month.

605
01:33:01.090 --> 01:33:15.690
Eric C Bellm: And a lot of that is human steps, you know, either shepherding data through, or looking… vetting the data before we deploy it, or getting it deployed. I hope that can come down over time as we streamline

606
01:33:15.690 --> 01:33:22.509
Eric C Bellm: You know, our metrics and learn to trust, you know, what we're looking at.

607
01:33:22.550 --> 01:33:28.450
Eric C Bellm: you know, again, we sort of started with the DDFs, which have a very different sort of…

608
01:33:28.680 --> 01:33:46.470
Eric C Bellm: behavior when you co-add them, versus some of these wider areas. So, you know, that's… that's my estimate right now, is it takes a month, and I hope that it improves over time. It kind of needs to, but, I don't… I don't think that's… that's happening quickly, so…

609
01:33:50.560 --> 01:33:51.440
Lynne Jones: John?

610
01:33:52.950 --> 01:34:00.260
John Gizis: So I guess I… I guess I don't quite follow the big picture case for, sort of, taking a major part of the first

611
01:34:00.540 --> 01:34:06.829
John Gizis: year and turning it into a 20-second survey, right? I mean, if there's one-month delays anyway on templates.

612
01:34:07.240 --> 01:34:10.460
John Gizis: The alert, you know, volume is so large.

613
01:34:10.590 --> 01:34:22.890
John Gizis: you know, in the long run, of course, you can… you will have templates and can go backwards in time, so it's just about alert production. Doesn't it seem like there'll be enough alerts that maybe you should just stick with the 30-second? Like, what's the big picture? Like.

614
01:34:23.890 --> 01:34:27.809
John Gizis: Like, you really won't be able to generate alerts even after one year.

615
01:34:28.220 --> 01:34:29.080
Lynne Jones: So…

616
01:34:29.080 --> 01:34:33.289
John Gizis: Over the whole sky? In every filter? Is that the problem? Like…

617
01:34:33.290 --> 01:34:34.450
Lynne Jones: Oh.

618
01:34:34.450 --> 01:34:36.919
John Gizis: Understand what drives such a big change.

619
01:34:37.470 --> 01:34:39.789
Lynne Jones: Yeah, I don't… I don't… okay, so Eric, I saw Eric…

620
01:34:40.300 --> 01:34:45.819
Lynne Jones: You put up your hand, so I'm gonna say maybe you… if you have something to say about this, go ahead.

621
01:34:45.820 --> 01:34:55.250
Eric C Bellm: Okay, sure, we can take two turns. I think the advantage to 20-second exposures is that it gives you more images

622
01:34:55.730 --> 01:35:03.990
Eric C Bellm: you know, more… a larger number of images at any given position and filter in the sky at the end of year one. And…

623
01:35:03.990 --> 01:35:17.529
Eric C Bellm: If those are… especially if those are preferentially taken in better conditions, that increases the odds that you have a good set of inputs that you can, reject artifacts, you know.

624
01:35:17.570 --> 01:35:31.159
Eric C Bellm: you know, dither out chip caps, etc. So, you're sort of sacrificing a little bit of depth in exchange for kind of an absolute image quality improvement. It's less about, in my mind, it's less about

625
01:35:31.200 --> 01:35:41.139
Eric C Bellm: Getting alerts out in year one, and more about, do you have enough, you know, what is the quality and aerial coverage of templates at the end of, at the end of year one?

626
01:35:41.140 --> 01:35:59.050
Eric C Bellm: I will say, you know, Lynn didn't say, but, you know, we have been discussing in data management, like, can we support 20-second images? Are there concerns we need to worry about? Largely, the answer is no, with one exception. There is some concern that the active optics system

627
01:35:59.050 --> 01:36:22.450
Eric C Bellm: instead of being able to correct, sort of, every, you know, 2 images later would be then at 3 images later correction because of the shorter, sort of, cycle time. And that is a significant concern that they worry that the image quality will suffer, and that may be, you know, maybe a showstopper. And there's some discussion, I don't know if Bob or others want to comment, you know, we should just test this.

628
01:36:22.560 --> 01:36:31.350
Eric C Bellm: But I think that's maybe the biggest concern that I'm aware of with the 20-second exposures outside of the survey strategy question.

629
01:36:31.670 --> 01:36:41.530
John Gizis: And could I ask a related question? My metric actually seemed to survive, because the, you know, the extra images probably makes up for anything you lose, so it's, it's, you know, it's all fine.

630
01:36:41.610 --> 01:36:56.269
John Gizis: But I heard a lot, I think, in previous workshops of the need for really high-quality, G-band and R-band images, that it was important to get, you know, sort of deep, good-seeing images there.

631
01:36:56.360 --> 01:36:58.139
John Gizis: And it sort of seems like that might…

632
01:36:58.680 --> 01:37:09.480
John Gizis: be a real problem to what I heard, you know, repeatedly in a lot of documents, that that was sort of really a priority for the 10-year static case. Maybe it's all in the wash, since it's only 1 tenth.

633
01:37:09.800 --> 01:37:14.239
John Gizis: But, has that… is there any comments about that?

634
01:37:15.110 --> 01:37:31.670
Lynne Jones: This is… this is a good point, because I think that this is one of the things that's not necessarily well captured in the metrics that we have, like, how… and… and what it goes into is, how well can you de-blend things like stars and galaxies, is my understanding.

635
01:37:32.060 --> 01:37:35.360
Lynne Jones: Yeah, so…

636
01:37:36.340 --> 01:37:44.120
Lynne Jones: That might be an issue. I think it certainly is one of the things we would have to go and consider.

637
01:37:44.320 --> 01:38:00.989
Lynne Jones: It also, after 10 years, it does kind of probably come out in the wash, but the question is, do you need that at the end of year one? Which I think was actually their request, that you have those at the end of year one. That said, they also didn't have a large number of these

638
01:38:01.390 --> 01:38:09.129
Lynne Jones: you can sort of think of them as Pathfinder, like, they're the, like, that's pin to blending based on these, like, really deep, good-seeing images.

639
01:38:09.320 --> 01:38:10.430
Lynne Jones: And…

640
01:38:11.210 --> 01:38:25.630
Lynne Jones: it's not obvious you wouldn't have that anyway, even with the slightly shallower depths, because they are in good conditions, so that you naturally get a slightly deeper depth. So I think that's something to also look at, for sure.

641
01:38:25.690 --> 01:38:32.020
Lynne Jones: It's also true that it may be a showstopper based on the AOS,

642
01:38:32.110 --> 01:38:34.500
Lynne Jones: So, we'll see. There's, there's… there's…

643
01:38:35.510 --> 01:38:40.210
Lynne Jones: This is still very much in the, we're going to…

644
01:38:40.450 --> 01:38:43.899
Lynne Jones: Have to think about this a little bit more, maybe test some data on the sky.

645
01:38:44.160 --> 01:38:48.030
Lynne Jones: The other… there wasn't another…

646
01:38:50.250 --> 01:38:58.879
Lynne Jones: there's maybe some concern about U-band, and maybe we would have to leave U-band at 38 seconds, just because,

647
01:38:59.570 --> 01:39:11.410
Lynne Jones: getting even further… getting back into the read noise, which is what would start to happen with the 25-second exposure, might be bad. Alright, here's… this is the other thing I was gonna say, though.

648
01:39:12.190 --> 01:39:23.780
Lynne Jones: it may not be that we have… then… if I didn't say, oh, it's so that we can get templates over the whole sky and we can close the sky, but if instead I said, and this might hold up DR1,

649
01:39:24.120 --> 01:39:26.160
Lynne Jones: Does that change your opinion?

650
01:39:28.020 --> 01:39:39.710
Lynne Jones: Because that… not having… not being able to make templates over the whole sky might actually mean, like, oh, we're gonna wait a little bit longer until we get those images so we can make them as part of DR1.

651
01:39:40.810 --> 01:39:48.160
Lynne Jones: That is not my decision to make, obviously, and it's… it's not a for-sure thing, but, like, that would be a possibility.

652
01:39:48.470 --> 01:39:52.860
Lynne Jones: So… Damn.

653
01:39:55.340 --> 01:40:04.510
Dan Taranu: Yeah, so you showed this plot that had the number of… or the area covered versus the number of nights, or the various scenarios, and it kind of looked like

654
01:40:05.380 --> 01:40:11.429
Dan Taranu: The slope from the baseline after you reach around 15,000 square degrees was just about the same.

655
01:40:11.870 --> 01:40:21.119
Dan Taranu: I was wondering, are those the same 15,000 square degrees? Are you envisioning a scenario where you do 20 seconds for, like, 200 days and then switch over, or…

656
01:40:21.710 --> 01:40:24.040
Lynne Jones: This one… this one here, you mean?

657
01:40:24.310 --> 01:40:25.040
Dan Taranu: Yeah.

658
01:40:27.820 --> 01:40:44.339
Lynne Jones: Yes, sorry, after… so this would only be 20-second visits coming from the template tier. Any other tier would just do normal 30-second visits. However, like I said, in the first year, and the template tier switches off after the first year, right? So in the first year,

659
01:40:44.740 --> 01:40:50.420
Lynne Jones: On the order of 70% of the visits would be the shorter visits coming from the template tier.

660
01:40:53.910 --> 01:40:57.210
Lynne Jones: So yeah, after this point, it looks…

661
01:40:57.620 --> 01:41:04.880
Lynne Jones: It will become fairly similar, but that's also… the template here isn't trying to acquire any visits after that point.

662
01:41:06.160 --> 01:41:08.700
Dan Taranu: Right. Well, what I meant was after about

663
01:41:08.840 --> 01:41:14.680
Dan Taranu: Knight 300, the slope of the green line looks basically the same as red, right?

664
01:41:16.310 --> 01:41:20.779
Dan Taranu: So, if you were to switch over from 20 seconds to 30 seconds after, like.

665
01:41:20.960 --> 01:41:23.009
Dan Taranu: I don't know, 250 nights?

666
01:41:25.010 --> 01:41:25.830
Lynne Jones: I'm…

667
01:41:27.690 --> 01:41:35.690
Lynne Jones: So, okay, Peter can say something more about this if he wants, because he made this plot, but I'm gonna guess that that's…

668
01:41:36.030 --> 01:41:38.989
Lynne Jones: Not just because of,

669
01:41:40.190 --> 01:41:49.729
Lynne Jones: A lot of that has to do with how much area is new and appearing on the sky, maybe rather than quite so much as,

670
01:41:50.770 --> 01:41:58.429
Lynne Jones: the rate of acquiring just enough. I'm not sure, yeah. Peter, do you have anything to say about that?

671
01:41:58.430 --> 01:42:08.540
Peter Yoachim: I bet that rapid rise is that we observed some area at the start of the survey, but didn't get enough to get templates, and so it's come back around, and so now we're able to rapidly, you know.

672
01:42:08.750 --> 01:42:13.240
Peter Yoachim: One image gets us to the threshold, and so that's why there's that rapid rise.

673
01:42:13.630 --> 01:42:14.010
Lynne Jones: Yeah.

674
01:42:14.010 --> 01:42:18.419
Peter Yoachim: Yeah, it's… it's area that's coming back around that… that now we can complete.

675
01:42:21.370 --> 01:42:31.539
Lynne Jones: Yeah, that makes sense. So, it's… it's because we did… we did part of the sky that we're observing here, we did it back here, but we didn't reach 4.

676
01:42:32.460 --> 01:42:33.010
Peter Yoachim: Yeah.

677
01:42:36.150 --> 01:42:36.900
Lynne Jones: Okay.

678
01:42:41.330 --> 01:42:48.230
Lynne Jones: I'll pause for a minute if there's any more questions. We are into our break, so… But also I think that…

679
01:42:49.090 --> 01:42:57.080
Lynne Jones: these are all really good questions, and I really appreciate hearing about the templates and the current plans. So thank you, Eric.

680
01:43:03.260 --> 01:43:03.920
Lynne Jones: Alright.

681
01:43:04.800 --> 01:43:14.999
Lynne Jones: I don't see any more questions. Remember, you can ask questions on Slack, or on Community if you have a more general question. That's great, a great place as well.

682
01:43:15.560 --> 01:43:16.400
Lynne Jones: Alright.

683
01:43:17.450 --> 01:43:25.289
Lynne Jones: Actually, this is the time for our break, so we'll come back at, in 20 minutes at 10 o'clock.

684
02:03:01.450 --> 02:03:02.480
Lynne Jones: All right.

685
02:03:02.830 --> 02:03:05.600
Lynne Jones: Here we are, 10 o'clock,

686
02:03:06.960 --> 02:03:12.029
Lynne Jones: Hopefully everybody comes back from the break now. Okay, so…

687
02:03:12.260 --> 02:03:24.060
Lynne Jones: Next, we have, two more specific areas to cover, rolling cadence and options around rolling cadence, and Leanne Guy's gonna talk a little bit, with us about

688
02:03:24.490 --> 02:03:30.029
Lynne Jones: uniformity in coads and data management processing for DRP.

689
02:03:30.590 --> 02:03:38.039
Lynne Jones: As well, to go with that. And then we're going to talk about the DESE survey options. And then we have some time for general discussion.

690
02:03:38.190 --> 02:03:39.620
Lynne Jones: As, as we need it.

691
02:03:40.390 --> 02:03:41.340
Lynne Jones: Okay.

692
02:03:43.010 --> 02:03:44.040
Lynne Jones: So…

693
02:03:44.830 --> 02:04:00.340
Lynne Jones: rolling cadence. So, if you remember, this is what we do so that we can boost the cadence, for time domain in certain areas of the sky, at the expense of in other seasons that has slightly lower cadence. So.

694
02:04:00.730 --> 02:04:12.719
Lynne Jones: When a part of the sky is in active rolling cadence, the… the gap between, nights of visits is… is sort of 2.

695
02:04:12.910 --> 02:04:24.829
Lynne Jones: On average. And in a low cadence season, the gap between coming back to the same, part of the sky, the same pointing on the sky, is about 7 nights.

696
02:04:25.200 --> 02:04:27.270
Lynne Jones: So it does vary, and then…

697
02:04:27.430 --> 02:04:37.909
Lynne Jones: In between, in our seasons, which are just, like, normal, uniform cadence, like in the first year, the gap between revisits is actually,

698
02:04:38.380 --> 02:04:45.310
Lynne Jones: Medium value of 3, you know, obviously there's a little bit more scatter in there, so…

699
02:04:45.560 --> 02:04:55.819
Lynne Jones: So, if you look at the whole survey, and don't look at individual seasons of rolling, you mostly… and you're just measuring something like the

700
02:04:55.940 --> 02:05:08.009
Lynne Jones: the gap between visits. You mostly see the uniform seasons, and it sort of washes out that increase from rolling and the decrease from the, low activity rolling errors. But.

701
02:05:08.330 --> 02:05:13.650
Lynne Jones: If you are looking at something like how much… how well you're recovering particular light curves.

702
02:05:13.760 --> 02:05:21.620
Lynne Jones: That will be influenced by the details of the observations in a particular season or year, and so you'll notice that a little bit more.

703
02:05:21.740 --> 02:05:26.830
Lynne Jones: So that's one of the things that, depending on how you're…

704
02:05:27.230 --> 02:05:33.380
Lynne Jones: Looking at what you're looking at when you're trying to evaluate impacts of

705
02:05:34.170 --> 02:05:38.120
Lynne Jones: just the visit spacing on time domain.

706
02:05:38.310 --> 02:05:42.839
Lynne Jones: metrics. You do have to consider whether you're

707
02:05:43.180 --> 02:05:50.730
Lynne Jones: Sensitive to the changes in a given year of rolling, or if you're sort of washing out over the whole survey.

708
02:05:51.850 --> 02:05:53.799
Lynne Jones: So, that said…

709
02:05:54.070 --> 02:05:59.180
Lynne Jones: That makes some of the metrics a little tough to interpret when you're looking at 10-year metrics.

710
02:05:59.900 --> 02:06:00.690
Lynne Jones: And rip.

711
02:06:01.610 --> 02:06:15.940
Lynne Jones: what we have is, in baseline V5.3, we start rolling after year one. This is the standard. We don't want to roll inside of year one, because we want fairly uniform coverage at the end of year one, so that we can do

712
02:06:15.970 --> 02:06:23.400
Lynne Jones: Calibration and photometric and astrometric catalogs over the whole sky with something that's,

713
02:06:23.800 --> 02:06:26.530
Lynne Jones: That has the number of visits and is fairly uniform.

714
02:06:26.950 --> 02:06:29.759
Lynne Jones: Now, if we're still acquiring visits.

715
02:06:30.090 --> 02:06:49.720
Lynne Jones: needed for the input templates, instead of starting rolling right after year 2, or right at the start of year two, at the end of year one, we could delay the start of rolling until after year two, because there is, a reasonable argument to be made that you should not start rolling when you don't have templates yet.

716
02:06:49.730 --> 02:06:58.649
Lynne Jones: Because, the templates are needed to fully leverage the time domain, and the time domain is the reason for going into the rolling cadence.

717
02:07:00.120 --> 02:07:04.690
Lynne Jones: Now, it is also too early yet to know if this is a problem.

718
02:07:04.970 --> 02:07:22.660
Lynne Jones: And the decision can wait while we're investigating options on whether we should start rolling at the end of year one, or whether we should start rolling at the end of year two. So, some of the things we're talking about here in the various simulations that we have are questions of

719
02:07:23.460 --> 02:07:28.419
Lynne Jones: things to think about, we don't… we won't be making the decision today.

720
02:07:28.720 --> 02:07:31.220
Lynne Jones: Or… or immediately, anyway.

721
02:07:32.760 --> 02:07:46.780
Lynne Jones: This series of plots over here on the right is the number of visits in IBAND, so I chose just a single band, after year one. So, year one is a time…

722
02:07:47.030 --> 02:07:54.279
Lynne Jones: deadline, not a season. So remember, this is just the… the end of year one. We would have this many visits in IBAN.

723
02:07:54.710 --> 02:08:01.329
Lynne Jones: At the end of year two, our coverage would look like this, because we've started rolling during year two.

724
02:08:02.210 --> 02:08:11.439
Lynne Jones: At the end of year three, it looks like this, because we've started rolling, but also we've started the opposite rolling, but you still see all of this signature rolling.

725
02:08:11.550 --> 02:08:21.949
Lynne Jones: And then we have a season for the rolling cadence, where instead of doing a high activity or low activity, it has an average, so the medium.

726
02:08:22.180 --> 02:08:31.730
Lynne Jones: Or normal uniform activity. And so that lets the sky catch up and, equalize, and so at the end of year four.

727
02:08:31.840 --> 02:08:40.570
Lynne Jones: we have what looks like a fairly uniform coverage on the sky. Now, I did make these plots, removing any of the special surveys.

728
02:08:41.120 --> 02:08:46.420
Lynne Jones: Because that… then this lets us see exactly how uniform,

729
02:08:46.780 --> 02:08:53.139
Lynne Jones: Things like the… the pair survey, and our normal wide, fast, deep kind of observations would look.

730
02:08:54.120 --> 02:09:12.439
Lynne Jones: If we delay the start of rolling until after year two, what this basically means is we're shuffling the high and low activity and uniform seasons so that the end of year two is uniform, so instead of being here at the end of year four, year 2 would be uniform.

731
02:09:12.680 --> 02:09:18.539
Lynne Jones: And then 3 and 4 would be non-uniform, because we've just shuffled that order.

732
02:09:19.920 --> 02:09:20.780
Lynne Jones: Okay.

733
02:09:22.060 --> 02:09:24.320
Lynne Jones: So, we have some simulations.

734
02:09:24.500 --> 02:09:27.840
Lynne Jones: That illustrate

735
02:09:28.450 --> 02:09:36.280
Lynne Jones: different options. So if we go with the baseline, normally at the end of year four, we have a uniform data release.

736
02:09:36.520 --> 02:09:42.449
Lynne Jones: And I'm gonna keep saying uniform with quotes because, this is part of the things that

737
02:09:42.570 --> 02:09:44.589
Lynne Jones: that Leon is going to talk about.

738
02:09:45.070 --> 02:09:49.410
Lynne Jones: Well, we'll talk about a little bit more in a minute.

739
02:09:50.100 --> 02:10:03.499
Lynne Jones: So at the end of year four, normally we would have a uniform data release, but if we reshuffle and delay the start of rolling, that year 4 uniform data release disappears, and it moves into year 2. So these are…

740
02:10:03.520 --> 02:10:22.969
Lynne Jones: Two other options. This is if we just delay the start of rolling, or we don't change anything else. So you can see year four, so across in this plot is all year four, for different simulations. Down in these series of images is different years, so it's 4, 5, and 7, not 6. I skipped a year, but…

741
02:10:23.370 --> 02:10:24.470
Lynne Jones: There.

742
02:10:24.760 --> 02:10:35.669
Lynne Jones: So this is… on the left is the baseline, year 4, 5, and 7. And the second row is just a simple… we'll delay the start of rolling for one year.

743
02:10:35.960 --> 02:10:46.420
Lynne Jones: So this is what we… the simulation we're calling Roll Mash, 5.3. So it goes… this is, again, your 4, 5, and 7, and you can see your 7…

744
02:10:46.570 --> 02:10:55.279
Lynne Jones: Because we only shuffled some of these rolling seasons early in the survey, year 7 looks very similar. So you have year 7 in the baseline, year 7 in Roll Mash.

745
02:10:55.450 --> 02:10:58.370
Lynne Jones: That's our uniform data release year.

746
02:10:59.450 --> 02:11:06.660
Lynne Jones: And then, if we not only reshuffle the, seasons of uniform rolling,

747
02:11:06.830 --> 02:11:17.310
Lynne Jones: in the first year… sorry, in year 2 and 3 and 4, if we also shuffle them in later years, we can move that uniform data release that was in year 7

748
02:11:17.640 --> 02:11:22.939
Lynne Jones: into your 5. And so this is roll U5, V5.3.

749
02:11:24.060 --> 02:11:29.629
Lynne Jones: So this one… in this U5 simulation, we also delayed the start of rolling for a year.

750
02:11:30.210 --> 02:11:42.150
Lynne Jones: But instead of taking the uniform season, right before the year 7 data release, we've moved it up in time, and so we have a uniform data release at year 5 instead.

751
02:11:42.450 --> 02:11:53.210
Lynne Jones: So in both of these simulations, because we wait, we have a uniform data release in year 1, year 2, and then this would be year 7, and year 10, and this would be year 5, and year 10.

752
02:11:54.760 --> 02:11:59.089
Lynne Jones: So, if we delay… This started rolling.

753
02:11:59.350 --> 02:12:08.189
Lynne Jones: We inevitably lose uniform data release at year 4, but then we can have a choice about if it's year 7 or year 5.

754
02:12:11.060 --> 02:12:15.999
Lynne Jones: And of course, year 10 in all of these simulations is the same.

755
02:12:18.550 --> 02:12:24.199
Lynne Jones: So… Because at the end of 10 years, you've done the same…

756
02:12:24.330 --> 02:12:35.239
Lynne Jones: number of visits, and the simulation is generally the same. Just reshuffling the seasons when you're doing rolling cadence and when you're not should be…

757
02:12:35.700 --> 02:12:38.989
Lynne Jones: really have a minimal impact on those Year 10 metrics.

758
02:12:39.220 --> 02:12:50.870
Lynne Jones: And so this is our metric grid again. It's, here's V5.1, the baseline, and so this is what everything is normalized to, and these are these high-level summary metrics.

759
02:12:51.640 --> 02:12:54.990
Lynne Jones: Now… Role…

760
02:12:55.170 --> 02:13:06.689
Lynne Jones: Year 2, U7 is, like, we have… wait until year 2, this is the roll mash, sorry, it's renamed. This is one reason why, in general, we discourage people from renaming, but I did it anyway.

761
02:13:08.580 --> 02:13:15.429
Lynne Jones: And this is the one where it was… the uniform data release was shuffled into year 5 instead of year 7.

762
02:13:15.880 --> 02:13:18.530
Lynne Jones: And you can see it's not exactly the same.

763
02:13:19.100 --> 02:13:23.760
Lynne Jones: The changes are pretty minimal, so this… note the scaling here, this is,

764
02:13:24.930 --> 02:13:30.829
Lynne Jones: 10% into the… into the red, or 10% into the… into the blue.

765
02:13:31.040 --> 02:13:39.009
Lynne Jones: And these are all generally a few percent, except for some. There's… there are some of these time domain metrics which seem a little bit more sensitive.

766
02:13:39.040 --> 02:13:56.500
Lynne Jones: And it may be that what we're actually seeing here is noise, rather than actually something, real in the changes of the metrics, because really, these impacts should be minimal, but it is possible that as we change things around.

767
02:13:56.790 --> 02:13:59.630
Lynne Jones: There's some unintended effects.

768
02:14:00.090 --> 02:14:06.229
Lynne Jones: So it's worth looking at, but it is also possible that… Not that significant.

769
02:14:08.090 --> 02:14:20.369
Lynne Jones: So this is the thing we should think about, like, what do we do about, if we did delay the start of rolling, which we don't know we would do yet, but it is a possibility, especially if we can't.

770
02:14:20.520 --> 02:14:24.710
Lynne Jones: Complete templates around the sky at the end of year one.

771
02:14:25.010 --> 02:14:31.500
Lynne Jones: Then what do we do about later years, and down the road, do we shift some of these uniform data releases around?

772
02:14:35.030 --> 02:14:46.150
Lynne Jones: Now, I also wanted to talk a little bit about uniformity and what that means, because we talk a lot about, hey, let's have a uniform data release.

773
02:14:46.640 --> 02:14:51.779
Lynne Jones: But, that's not really very well defined.

774
02:14:52.490 --> 02:14:58.619
Lynne Jones: And… It's not obvious, what's the best way to support that?

775
02:14:58.770 --> 02:15:04.890
Lynne Jones: There's some uniformity we can enforce, we can try to gain from survey strategy.

776
02:15:05.210 --> 02:15:13.859
Lynne Jones: By not letting our… our number of epochs or depths get too uneven across the sky.

777
02:15:13.980 --> 02:15:22.019
Lynne Jones: And an example of trying to manage that uniformity is with this uniform season in the rolling cadence.

778
02:15:23.470 --> 02:15:42.339
Lynne Jones: There's also an aspect, like, you could enforce uniformity later in the processing chain at data management, although if we do that there, it necessarily means dropping visits to gain uniformity, so you're not dropping visits completely, but not including them in the coads.

779
02:15:43.030 --> 02:15:49.270
Lynne Jones: So… This raises questions, how could or should we achieve

780
02:15:49.400 --> 02:15:54.490
Lynne Jones: uniformity in the coads and the catalogs that come from the coads.

781
02:15:54.570 --> 02:16:14.260
Lynne Jones: What does that mean? There is some natural variation due to weather, that will lead to variations in the coated depth across the low dust, wide, fast, deep region, but is it that we want less than the magnitude of variation in the estimated coated depth, the number of epochs?

782
02:16:14.360 --> 02:16:20.140
Lynne Jones: Is it a different number on the… on the RMS of the coated depth?

783
02:16:20.580 --> 02:16:25.720
Lynne Jones: And what does it mean for visits which are not included in the co-ads?

784
02:16:25.970 --> 02:16:44.380
Lynne Jones: So these are all good questions, that I had, and then the plots here show us the co-ed depth and I-band, if you do include all of the special surveys, versus if you don't. So even if we don't have a question about, like, what do we do with rolling cadence, what do we do with DESI,

785
02:16:44.629 --> 02:16:52.719
Lynne Jones: anything, we already have non-uniformity due to things like the TOs, and other special surveys.

786
02:16:53.790 --> 02:16:57.799
Lynne Jones: So… Leanne, are you there?

787
02:16:59.260 --> 02:17:00.490
Leanne Guy (Rubin): I am here.

788
02:17:00.620 --> 02:17:12.929
Lynne Jones: Excellent, excellent. I was going to hand this over to you to talk a little bit about, some of the things we chatted about the other day, like, what happens to visits that don't get included in co-ads and things like that?

789
02:17:13.850 --> 02:17:30.650
Leanne Guy (Rubin): Yeah, okay, thanks, Lynn, thanks for introducing us. So, I've sort of just come along at the last minute at Lynn's request to support this meeting. I have nothing prepared, I have no answers for you, but I am here to take a lot of questions and to think about how we can address some of the open topics and questions.

790
02:17:30.969 --> 02:17:46.130
Leanne Guy (Rubin): What happens to visits that are not included in COADs? They will still be perfectly valid single visit images, and there will be catalogues of detections in those single visit images, and they will be included in a data release. They just won't have gone into a COAD.

791
02:17:46.580 --> 02:17:58.249
Leanne Guy (Rubin): You'll already see this, in fact, with Data Preview 2, which is coming up sometime in the July to September timeframe. We have a region of DP2 which has co-ads.

792
02:17:58.250 --> 02:18:12.709
Leanne Guy (Rubin): And then we have a lot of other visits scattered across a very large area of the sky. So the co-added region is roughly 3,000 square degrees, although there's variation of co-added depth within that 3,000 square degrees. And then you can get out to another sort of, I think it's about 12,000 or so

793
02:18:12.830 --> 02:18:23.079
Leanne Guy (Rubin): degrees, with just single visit images taken at various positions on the sky. So, some of the positions on the sky will only actually have one visit, maybe some will have two, but none of these were co-added.

794
02:18:23.709 --> 02:18:28.149
Leanne Guy (Rubin): So, you'll already see how that looks in the Data Preview 2 release.

795
02:18:29.730 --> 02:18:38.829
Leanne Guy (Rubin): So the term throw out is a bit strong, maybe. We will throw them out of the co-add, but we will not throw them out of the release, provided they meet other quality criteria.

796
02:18:44.620 --> 02:18:48.419
Lynne Jones: Sorry for my thrashing there as I tried to unmute myself.

797
02:18:49.190 --> 02:18:59.980
Lynne Jones: Yeah, and those… the catalogs from the single visit images are calibrated in the same way, like, against, the full astrometric and photometric calibration, etc.

798
02:18:59.980 --> 02:19:19.579
Leanne Guy (Rubin): In a data release, yes. Like, single visit images in the prompt, nightly processing, no, have a, I think, more fast approximate calibration, but in data release, yes, they are fully calibrated, single-visit images. It's… they're processed in exactly the same way as every other single visit image, irrespective of whether it goes into a co-add or not.

799
02:19:20.690 --> 02:19:32.559
Lynne Jones: So if we were able to make a goal on the uniformity of the co-edit depth, so that when we make catalogs from the co-edit images.

800
02:19:32.709 --> 02:19:34.230
Lynne Jones: that the…

801
02:19:34.360 --> 02:19:41.939
Lynne Jones: Depth in those catalogs is fairly uniform across the sky, and obviously there's a limit to how much you can do there.

802
02:19:42.170 --> 02:19:53.410
Lynne Jones: But if we were to do that, you would still have all of the single visit measurements well-calibrated, so you could do light curve analysis,

803
02:19:53.740 --> 02:19:59.599
Lynne Jones: On all of the visits, and it would just be that your, co-edited visit

804
02:19:59.880 --> 02:20:04.090
Lynne Jones: Co-edited catalog would be more uniform and more shallow.

805
02:20:04.850 --> 02:20:16.130
Leanne Guy (Rubin): Yeah, I mean, the co-added images that are created would be shallower, they would be missing some images, and that would, of course, have an impact on detection in co-ads, which is what goes into the object catalogue, so…

806
02:20:16.170 --> 02:20:32.349
Leanne Guy (Rubin): let's say, quite simply, they wouldn't have the same depth. Let's… if we… if we remove some, you'd lose some depth, so then you might not have, you might have fewer objects detected in those coads, or their parameters might not be measured with the same sort of accuracy or characterization. There might be more uncertainty around them.

807
02:20:32.610 --> 02:20:37.640
Leanne Guy (Rubin): Yep, I think that answers the question.

808
02:20:39.800 --> 02:20:42.740
Lynne Jones: Let's see if, there are questions from…

809
02:20:42.990 --> 02:20:49.410
Lynne Jones: the people listening. Or comments on, like, what the… the impact of this uniformity or not is.

810
02:20:49.840 --> 02:20:50.880
Lynne Jones: Roberta?

811
02:20:52.740 --> 02:20:55.980
Roberto Assef: Yeah, one question, is it…

812
02:20:56.350 --> 02:21:14.589
Roberto Assef: Would it be feasible in any way to sort of think of doing the full code, and then produce a good enough error model, so that at the catalog space, you could add those sort of additional uncertainties to homogenize your catalog at whatever level you want to.

813
02:21:15.220 --> 02:21:16.679
Roberto Assef: To do that.

814
02:21:17.170 --> 02:21:28.880
Leanne Guy (Rubin): I think so, yes, but as Dan's on there, and Dan's more deeply embedded in the pipelines than me, he might have a response to that, but I don't think, in principle, that would be a problem. I'm not sure if it's part of our current processing

815
02:21:29.020 --> 02:21:30.140
Leanne Guy (Rubin): Pipelines.

816
02:21:30.810 --> 02:21:31.460
Roberto Assef: Okay.

817
02:21:32.290 --> 02:21:33.739
Leanne Guy (Rubin): Dan, do you want to throw something in?

818
02:21:34.600 --> 02:21:36.160
Dan Taranu: Well, I was gonna say that…

819
02:21:36.270 --> 02:21:43.549
Dan Taranu: the image we do detection on doesn't necessarily have to be the same as the coeds we do measurement on, and

820
02:21:44.940 --> 02:21:51.240
Dan Taranu: The simplest example is we could do detection on a multivand image, and that has been tested, but not

821
02:21:51.690 --> 02:21:57.879
Dan Taranu: Fully implemented and… The impact hasn't really been measured yet, but it is a possibility.

822
02:21:58.810 --> 02:22:12.139
Leanne Guy (Rubin): Can I throw… can I add something just quickly to that? If we were to do this, though, consistently across the entire widefasted for a data release, that would involve producing two coads, essentially, for every position on the sky, right? To do detection on…

823
02:22:12.290 --> 02:22:15.579
Leanne Guy (Rubin): The deepest, and then measurement on the uniform.

824
02:22:17.990 --> 02:22:27.579
Dan Taranu: Yeah, but I mean, we are already making template coads, for DP2. We have also added the pretty co-ads that we're making images out of, so…

825
02:22:28.360 --> 02:22:38.229
Dan Taranu: I mean, yeah, there is a storage question. If this is… the thing we do detection on is something we persist as a product separately from the other coads, but I don't think it's out of the question.

826
02:22:39.110 --> 02:22:45.709
Leanne Guy (Rubin): So, so technically it's not out of the question, but there is a non-zero question of storage cost plus processing time plus

827
02:22:45.920 --> 02:22:47.200
Leanne Guy (Rubin): etc.

828
02:22:47.460 --> 02:22:59.190
Leanne Guy (Rubin): The short of it is, I can't tell you we can do that right now, because it's got budget involved, and I'd have to go back and talk to other people as well, but it's good to understand that the scientific and technical side doesn't necessarily pose any problems.

829
02:22:59.190 --> 02:23:12.549
Lynne Jones: And my understanding from previous questions about this was that it was… it probably did… was out of scope, especially once you included all the catalogs that you would generate from that, as well as the co-ed itself.

830
02:23:12.720 --> 02:23:14.980
Lynne Jones: But yeah, that, I mean…

831
02:23:15.460 --> 02:23:17.610
Lynne Jones: Those are really good questions, like.

832
02:23:17.610 --> 02:23:26.539
Leanne Guy (Rubin): I mean, there's also going to be provenance questions as well, and keeping track of all this, and what was done on what, so there's all sorts of, you know, data wrangling that has to go in as well.

833
02:23:26.880 --> 02:23:29.300
Leanne Guy (Rubin): Yeah.

834
02:23:31.110 --> 02:23:33.109
Leanne Guy (Rubin): Like, it's not just a flip-the-switch thing.

835
02:23:33.110 --> 02:23:33.430
Roberto Assef: Good.

836
02:23:33.430 --> 02:23:35.950
Leanne Guy (Rubin): There is work involved, but it's not,

837
02:23:36.240 --> 02:23:38.840
Leanne Guy (Rubin): It's… it's not, what's the word?

838
02:23:39.540 --> 02:23:42.410
Leanne Guy (Rubin): Well, we would know how to do it, like, yeah.

839
02:23:42.410 --> 02:23:45.959
Lynne Jones: But this does raise the point that somebody else could do it, too.

840
02:23:46.520 --> 02:23:54.429
Leanne Guy (Rubin): Indeed. But then it… that also raises the point of, you know, how do we define our data releases as our, sort of, flagship product?

841
02:23:54.920 --> 02:23:58.410
Leanne Guy (Rubin): That's… yeah, sorry, we're interrupting Rachel.

842
02:24:02.280 --> 02:24:05.860
Rachel Mandelbaum: You can't interrupt someone who hasn't started talking.

843
02:24:06.930 --> 02:24:10.010
Rachel Mandelbaum: so, I just…

844
02:24:10.640 --> 02:24:16.770
Rachel Mandelbaum: I… it could be that I… that I… that I missed something, but the way this slide and some of the discussion was framed.

845
02:24:16.880 --> 02:24:36.310
Rachel Mandelbaum: It seems like there may be a lack of awareness that there's been a lot of work on desk in, you know, defining metrics of sufficient uniformity in COAGs, and, quite a few people, you know, Boris Leistat, Renee Losec, me, a few others, put a lot of time and effort into getting metrics into MAF.

846
02:24:36.640 --> 02:24:39.089
Rachel Mandelbaum: So that they could be

847
02:24:39.290 --> 02:24:52.780
Rachel Mandelbaum: you know, run on the simulations and use to assess them. They cover, a subset of the questions one cares about with uniformity, so if there's a strategy that, for example, the metrics currently assume that if

848
02:24:52.890 --> 02:25:07.320
Rachel Mandelbaum: you know, RI and Z band are uniform-ish than the others are, so if there's a simulation for which that assumption is false, they won't catch that. But, you know, a lot of work went into that, and I wanted to advocate for those being included by default.

849
02:25:07.410 --> 02:25:20.160
Rachel Mandelbaum: and used, at least in part, to address this. I also want to say that I recognize that, you know, there are other science cases besides the Cosmology ones and desk for which uniformity matters, and it could be that

850
02:25:20.220 --> 02:25:32.069
Rachel Mandelbaum: for example, for galaxies, maybe the criteria are a bit different, but… but I do think that having some metrics is better than having no metrics, so I wanted to flag that, and I'll put a link in the, in the Slack

851
02:25:32.270 --> 02:25:35.660
Rachel Mandelbaum: Channel for this workshop, as well, to the, to the paper.

852
02:25:36.030 --> 02:25:51.680
Lynne Jones: Yeah, thank you, Rachel. I mean, you have a very good point, and I will apologize, because we did not run these metrics on the current set of simulations. We had missed including this in the automatic processing, and it turns out

853
02:25:51.770 --> 02:26:01.029
Lynne Jones: when we went back to look at it, there… we… we actually… there's a, NumPy upgrade-related issue with… with…

854
02:26:01.200 --> 02:26:02.430
Lynne Jones: Running it.

855
02:26:02.740 --> 02:26:16.980
Lynne Jones: So, we'll… we will look at that, we will work with you guys to make sure we're fixing it in a way that is consistent with your, outputs. I think what I'm missing, and maybe this is already in the metric, and I'm just missing it, is…

856
02:26:17.810 --> 02:26:29.679
Lynne Jones: I thought that the metrics indicated which simulations were better than others, and not necessarily gave the guidance that would be, if data management was setting out.

857
02:26:29.680 --> 02:26:40.239
Lynne Jones: to make a uniform co-add, what is their specific goal? And maybe I just missed it, and that's, like, on my part for not understanding what the metric is telling me.

858
02:26:41.660 --> 02:26:49.019
Lynne Jones: But I didn't know if there was a number on, like, this is the uniformity requirement at the end.

859
02:26:51.020 --> 02:26:56.840
Rachel Mandelbaum: So, we might want to talk offline about that. I think the short answer is that

860
02:26:57.280 --> 02:27:12.600
Rachel Mandelbaum: For one of the metrics, there is essentially a criterion one can apply, and, you know, we used it that way in the paper. But I do want to say that, you know, the reason that people have found it kind of slippery is, you know, is this uniform enough?

861
02:27:12.820 --> 02:27:16.020
Rachel Mandelbaum: Because the answer is that when it's not uniform.

862
02:27:16.240 --> 02:27:33.189
Rachel Mandelbaum: one has to try to mitigate that with, you know, like, how you do the analysis. So, you know, in principle, if it's really non-uniform, but maybe, you know, people spend, like, years putting together new analysis methods, technically, yes, they… maybe they could do cosmology, but, like.

863
02:27:33.320 --> 02:27:37.839
Rachel Mandelbaum: I'm not sure that's what we should be aiming for.

864
02:27:37.960 --> 02:27:42.710
Rachel Mandelbaum: So, the metrics generally are about, you know, with existing analysis methods.

865
02:27:42.920 --> 02:27:59.620
Rachel Mandelbaum: you know, which still take quite a while, lots of work and validation to apply to, you know, a large imaging survey, you know, is this, is this, uniform enough? And we can talk details offline, maybe, but, I wanted to just share the…

866
02:28:00.190 --> 02:28:05.400
Rachel Mandelbaum: the… the basic thought process behind, those metrics.

867
02:28:12.080 --> 02:28:13.120
Lynne Jones: Sora?

868
02:28:14.100 --> 02:28:22.109
Saurabh Jha: Yeah, I think, you know, I second everything Rachel just said. Just one part of it, which, and I'm really glad to hear from Leanne and Dan that, you know.

869
02:28:22.110 --> 02:28:38.540
Saurabh Jha: possibilities are, you know, not really off the table. It's not exactly the message that we'd sort of internalized a while ago, and so I think that's a very positive development. I think part of it was this philosophical idea, right? If you have, you know, quote-unquote, good visits in some part of the sky.

870
02:28:38.540 --> 02:29:01.049
Saurabh Jha: in time for a data release, are you going to deliberately not include them in a co-add? And I think the… there was at least a philosophical bent, at least coming from, you know, some portions of the community that said, no, you would have to include them in a data release. And so then, you know, we'd be talking about two separate coads, a uniform one and a full one.

871
02:29:01.050 --> 02:29:19.679
Saurabh Jha: But, you know, I think from DESC, I think also Galaxies, as I recall, had this as their highest, priority scientifically, is that for the coads and the catalogs that are made from those coads to be as uniform as possible. And of course, you know, those first two questions, you know, what does it actually mean? What does uniformity mean? I think that's a totally valid question.

872
02:29:19.680 --> 02:29:26.019
Saurabh Jha: But if it is possible to do something like that, where we know that those visits, you know, even if they're there.

873
02:29:26.250 --> 02:29:33.549
Saurabh Jha: They've been sitting on disk for a few months, and now is the time for the new data release, and they're gonna be treated just as single

874
02:29:33.550 --> 02:29:58.460
Saurabh Jha: visits, and they will show up in the COAD in the next data release, or in the final data release. I think that would be a huge scientific win if we could have some sort of, you know, definition of what a uniform co-ad and catalog means. That's up to the scientific community. But as that, as the main product of a data release, is something closer to uniform, I think that would be a huge scientific win. So I'm all in support

875
02:29:58.460 --> 02:30:07.600
Saurabh Jha: of trying to make that, you know, technically feasible and have the science drive the, you know, technical and budgetary requirements.

876
02:30:09.420 --> 02:30:10.930
Lynne Jones: Yeah, and, and…

877
02:30:11.520 --> 02:30:28.700
Lynne Jones: I know Desk did do this analysis of, like, how much depth you would lose if you were just correcting for rolling cadence all the time. And it was a significant amount of depth, so it is also possible that what we end up looking at is, some kind of…

878
02:30:28.820 --> 02:30:47.089
Lynne Jones: hybrid, where we do decide, okay, these will be uniform data releases, these may not, just based on the rolling cadence, but then there is this aspect of, like, we will have non-uniformity from the TOs, from the special surveys, from other things, like, if we include those stripes that Eric showed earlier.

879
02:30:47.190 --> 02:30:49.600
Lynne Jones: Right, that were the engineering data.

880
02:30:49.930 --> 02:30:58.460
Lynne Jones: That is very non-uniform. So, what do we do with all of this, and when do we want to,

881
02:30:58.560 --> 02:31:03.860
Lynne Jones: ensure we have uniform data releases. So I think these are all really good questions, and it's…

882
02:31:03.990 --> 02:31:09.280
Lynne Jones: there's… there's things we can do at Survey Strategy to make it,

883
02:31:09.430 --> 02:31:17.690
Lynne Jones: make it a good place to start from, but I do think that we can't make it all uniform from survey strategy at the end of the day.

884
02:31:18.800 --> 02:31:32.849
Saurabh Jha: Yeah, I agree, and I was focusing specifically on the uniform checkpoints that are, you know, in either one, you know, 14710 or 12510, you know, whatever it ends up being, those four checkpoints where we think it's going to be roughly uniform.

885
02:31:33.260 --> 02:31:36.989
Lynne Jones: Yeah, I think that's a great approach. Leanne?

886
02:31:38.480 --> 02:31:54.869
Leanne Guy (Rubin): Yeah, so I put my hand up to say something, and I'm going to respond, though, to a few of the points that came up. So, Saurabh, yes, you're right. In the past, there was sort of more of a… I wouldn't say it was a message, but it was maybe a bit of a reluctance for us to pivot our strategy to make uniform coeds during the construction period.

887
02:31:54.870 --> 02:32:00.430
Leanne Guy (Rubin): We had, the DM Science team, had a lot of discussions with Rachel and other people on the desk about this.

888
02:32:00.830 --> 02:32:03.250
Leanne Guy (Rubin): At the time, though, it wasn't so much that

889
02:32:03.250 --> 02:32:28.250
Leanne Guy (Rubin): we'd rejected the idea, but just that there were other… there were other avenues to explore, with the survey, and with trying to, you know, with trying to work this into the survey and not having to make duplicate sets of data products, and not having to make two sets of data products. So we wanted to explore all those avenues first, before we actually got to the point of doing work, which would, you know, have to go into our system to make these extra data products, to curate them, to have provenance, to search

890
02:32:28.250 --> 02:32:35.389
Leanne Guy (Rubin): of them, etc. And to pay the, you know, it is a non-zero cost increase to serve up an extra set of coads and do this work, so…

891
02:32:35.790 --> 02:32:46.180
Leanne Guy (Rubin): it wasn't off the table, it's just we wanted to explore all the other easier options first about, about, you know, getting this into the baseline survey cadence. So that's more what that was about.

892
02:32:49.960 --> 02:33:02.670
Leanne Guy (Rubin): let's see, what did you… you said it was… it would be a scientific win if we could have something like this, but it… doesn't that really depend on what science people are doing? Because uniform coads does sacrifice depth.

893
02:33:02.830 --> 02:33:17.189
Leanne Guy (Rubin): And my recollection from… I forget who did the paper, someone on the desk, was it was something like 10% over 10 years, was it? Or maybe it was a bit less than that on that sort of order, I can't remember, but… I mean, you do sacrifice depth, and it is, like, not including data, so…

894
02:33:18.430 --> 02:33:25.750
Leanne Guy (Rubin): I guess it doesn't… it just doesn't it depend on what science people are trying to do as to whether it's a scientific gain or not?

895
02:33:26.710 --> 02:33:51.410
Saurabh Jha: Yeah, yeah, for sure. I mean, I was sort of, maybe responding maybe too eagerly to the idea that there would be two sets. One would be a full, and one would be a uniform, and so you wouldn't have… it wouldn't be such a trade-off. I think, you know, for a lot of time domain, that's going to be a lot of single-visit science anyway. We have this separate issue about nightly co-ads or monthly co-ads, and you know, like, there are all these other time scales you could consider for other science cases.

896
02:33:51.410 --> 02:33:54.609
Saurabh Jha: So I didn't want to open that discussion.

897
02:33:54.610 --> 02:33:56.430
Leanne Guy (Rubin): Yeah, I…

898
02:33:56.430 --> 02:34:07.589
Saurabh Jha: I fully agree, yeah, different science cases will have different requirements, but, you know, for Desk and Galaxy, certainly having catalogs as uniform as possible across the sky is a big scientific…

899
02:34:07.590 --> 02:34:15.059
Leanne Guy (Rubin): So if we could set up the two sets of data products, the depth-optimized COAD and the, uniform-optimized COAD.

900
02:34:15.300 --> 02:34:19.120
Leanne Guy (Rubin): together. Yes, that would be a science game for everyone.

901
02:34:19.980 --> 02:34:32.190
Leanne Guy (Rubin): I guess I should probably go back and try to cost that now, given that we've gotten to this point. Cost in terms of actual dollars and cost in terms of people, time, effort, you know, all the rest of it.

902
02:34:33.750 --> 02:34:38.699
Leanne Guy (Rubin): Okay, so what I did put my hand up to ask about, unless, Rachel, you want to say something about what I just mentioned?

903
02:34:39.130 --> 02:34:40.610
Leanne Guy (Rubin): I have a different question.

904
02:34:41.370 --> 02:34:44.329
Rachel Mandelbaum: It is… it is kind of about what you… you just mentioned.

905
02:34:44.330 --> 02:34:45.000
Leanne Guy (Rubin): Go ahead.

906
02:34:45.000 --> 02:34:51.380
Rachel Mandelbaum: kind of the costs and the trade-offs in these different things you were just talking about. So,

907
02:34:51.620 --> 02:34:54.130
Rachel Mandelbaum: It could be this isn't quite what you meant, but…

908
02:34:54.530 --> 02:34:55.100
Rachel Mandelbaum: like.

909
02:34:55.610 --> 02:35:07.419
Rachel Mandelbaum: when we see, some of the depth maps where, for example, the TO observations are included together with everything else, and leading to some splotchiness, you know, at year 10,

910
02:35:07.580 --> 02:35:09.560
Rachel Mandelbaum: I had thought that that…

911
02:35:09.930 --> 02:35:23.429
Rachel Mandelbaum: could actually be more effort than a typical co-add, because, like, some of the exposures might have different exposure times. Like, it might actually be harder to throw the TOOs in than it is to, like, exclude them, and so it's felt like an…

912
02:35:23.580 --> 02:35:28.000
Rachel Mandelbaum: like, the cost trade-off isn't just that the uniform co-add is, you know.

913
02:35:28.000 --> 02:35:28.350
Leanne Guy (Rubin): Yeah.

914
02:35:28.350 --> 02:35:32.070
Rachel Mandelbaum: extra effort, but it could be I'm not totally…

915
02:35:32.370 --> 02:35:32.920
Lynne Jones: Yeah.

916
02:35:32.920 --> 02:35:34.249
Leanne Guy (Rubin): Lynn wants to say something.

917
02:35:34.250 --> 02:35:39.240
Lynne Jones: A little hard to find all the things and not hit my slides, sorry.

918
02:35:40.280 --> 02:35:58.140
Lynne Jones: Yes, and then there's also the question that, like, the TO observations may be at higher air mass, and they could be under, like, non-optimal conditions because the timing is the constraint. There will be visits that are not included in the coads because they don't meet quality.

919
02:35:58.200 --> 02:36:14.220
Lynne Jones: constraints. So we're already going to be not including some visits into the co-ads, and so this is maybe, like, adding an extra level of intentionally excluding visits, which could otherwise have been included, I think is one way to look at it.

920
02:36:15.610 --> 02:36:26.289
Lynne Jones: I am gonna say, we should move along to Desi, but we do have extra discussion time after DESI to cover any questions that have come up, throughout…

921
02:36:26.310 --> 02:36:36.030
Lynne Jones: all of this area. And I also do think this is very good preparation for when we talk about the DESI stuff, because that does also introduce some non-uniformity.

922
02:36:36.200 --> 02:36:41.259
Lynne Jones: But I'm gonna… if you have something to wrap up here real quick, Leanne.

923
02:36:41.260 --> 02:36:49.499
Leanne Guy (Rubin): Yeah, very, very quickly, but we can come back to this discussion. You've got a good starting set of questions there. A couple of others that I'd throw into it are.

924
02:36:49.500 --> 02:37:13.229
Leanne Guy (Rubin): And very importantly, how do we select which images to not include in the co-ad? And there's a whole lot of criteria. You know, we already have quality cuts on images to include in coads like seeing, for example. But, you know, does, for example, the time distribution of images throughout a co-add have any impact on the science people want to do? So that's also a really important one that I think data management cannot decide.

925
02:37:13.550 --> 02:37:21.440
Leanne Guy (Rubin): And we need some sort of algorithm, we need some sort of selection process agreed by the community across all sciences as to how

926
02:37:21.540 --> 02:37:23.230
Leanne Guy (Rubin): Optimally, to decide.

927
02:37:23.430 --> 02:37:34.189
Leanne Guy (Rubin): you know, what are the criteria that you want to cut on, and how do you combine those criteria to satisfy all the different sciences? That's just another important question I want to throw in there.

928
02:37:36.010 --> 02:37:37.040
Lynne Jones: Yeah, that…

929
02:37:37.390 --> 02:37:41.450
Lynne Jones: I didn't… I didn't write that, but that is definitely a good one. How do you pick the visits?

930
02:37:41.950 --> 02:37:42.850
Lynne Jones: Okay.

931
02:37:43.630 --> 02:37:45.330
Lynne Jones: So,

932
02:37:45.670 --> 02:38:03.450
Lynne Jones: I'm going to give a really brief overview on, what we did for the DESI simulations and how that contrasts to how we would be acquiring visits, otherwise. And then I'm going to hand this over to David, who will tell you all about the great, synergies that come out of,

933
02:38:03.810 --> 02:38:09.820
Lynne Jones: What could be enabled by the… this change in the survey strategy?

934
02:38:10.550 --> 02:38:13.729
Lynne Jones: So the basic idea in the request is,

935
02:38:14.140 --> 02:38:31.020
Lynne Jones: to take this about 5,000 square degree area that is shown here in the yellow and purple areas, and sort of accelerate the coverage of that area to get to a depth of 30 to 40 visits in each of you, GRIZY, by the end of year 4.

936
02:38:33.360 --> 02:38:40.399
Lynne Jones: And the initial, like, sort of question was, could we get to the yellow portion of that footprint by the end of year two?

937
02:38:40.650 --> 02:38:49.669
Lynne Jones: On the left, what I have is the median number of visits over the sky in the baseline survey. So this is a median, and it's over…

938
02:38:50.090 --> 02:38:58.229
Lynne Jones: Just without any, any, influence, and as you've seen before, rolling cadence does start to influence things after

939
02:38:59.610 --> 02:39:01.500
Lynne Jones: After you… when you get to year two.

940
02:39:01.810 --> 02:39:07.980
Lynne Jones: But this is… this is sort of median number of visits that we would have over the sky, and you can see

941
02:39:08.500 --> 02:39:17.209
Lynne Jones: in RISY in the redder bands, we would get to that, or close to that depth at the end of year 3.

942
02:39:17.490 --> 02:39:28.369
Lynne Jones: And then UNG are where we have the problem with this goal, because at year 3, we're still very far back at, like, 10 and 16 visits.

943
02:39:28.860 --> 02:39:39.230
Lynne Jones: Note that our template tier has already accelerated U and G in year one, to get to the number of visits we need for the templates.

944
02:39:39.970 --> 02:39:49.819
Lynne Jones: And we don't actually reach 30 visits in these bands until year 6 or year 8, if we go with a natural cadence.

945
02:39:50.410 --> 02:40:06.659
Lynne Jones: So that's the primary problem. What you see, if you look at, like, our band, is that year 4, so after the… year four would be the end of the, the request, can you cover this footprint to your

946
02:40:06.760 --> 02:40:20.230
Lynne Jones: to this number of visits by year 4. At year 4 in G-Band, the dizzy footprint is very obvious, because it's ahead of the rest of the sky, and at year four, in our band, it's not really very obvious.

947
02:40:21.430 --> 02:40:25.360
Lynne Jones: Which just matches what we see from the number of visits per year.

948
02:40:26.730 --> 02:40:32.720
Lynne Jones: The other thing to consider with this is that… within the DESE footprint.

949
02:40:32.750 --> 02:40:52.390
Lynne Jones: So this is the number of… the night, and the number of visits we've acquired to that… to… up to that night. And so, in year one, we really… because we have this template tier, because we're pushing hard to get all the visits for the template, there isn't really that much of a change. Whether you're in the DESI footprint or outside the DESE footprint, we're doing the same thing in year one.

950
02:40:52.950 --> 02:40:58.110
Lynne Jones: But… After… after year one,

951
02:40:58.280 --> 02:41:06.020
Lynne Jones: if in the DESE simulations, we sort of switch over to, like, okay, let's try and get the, the visits for this

952
02:41:06.180 --> 02:41:10.290
Lynne Jones: footprint. You can see that it, accelerates.

953
02:41:10.460 --> 02:41:20.629
Lynne Jones: And so we get up to, this is a part in the year 2 footprint, so the yellow area. So you can see we get up to,

954
02:41:22.270 --> 02:41:25.260
Lynne Jones: Maybe not. Okay, sorry, it's not vibe.

955
02:41:25.820 --> 02:41:32.280
Lynne Jones: So we get, we get a lot of visits in, UGRIZ,

956
02:41:32.820 --> 02:41:50.679
Lynne Jones: because of the being in the Desi footprint, but then after you get that number of visits, because now you're very far ahead in this part of the sky, you get very few visits for a very long time, and then eventually the rest of the sky catches up to you, and so you say… so the scheduler

957
02:41:50.930 --> 02:41:57.749
Lynne Jones: Says, okay, great, now we're… we're balanced in the footprint, and we can… we should start taking visits here again.

958
02:41:57.870 --> 02:41:59.480
Lynne Jones: And so you see this…

959
02:41:59.720 --> 02:42:10.049
Lynne Jones: acceleration, and then a weight until you take more visits. And that's not as strong in the rider bands, because those are more on par with the rest of the sky.

960
02:42:10.280 --> 02:42:13.600
Lynne Jones: And then outside of the DESI footprint.

961
02:42:13.900 --> 02:42:27.130
Lynne Jones: So this is a much lower declination that's outside of that footprint. You see that you get a little bit of suppression in that period where it's… the scheduler is concentrating on DESI, and then you… you start

962
02:42:27.410 --> 02:42:29.619
Lynne Jones: Recovering visits after that.

963
02:42:33.440 --> 02:42:42.619
Lynne Jones: And of course, we can look at our metrics. The metrics here… so this is DESI, and we have two simulations for the DESI footprint. We…

964
02:42:43.170 --> 02:43:00.480
Lynne Jones: oops, sorry, we have two simulations to date for the DESI footprint. DESI yearly does that yellow band and blue band completing by year 2 and Year 4, and the DESI 3040, actually tries to complete the whole thing by year 2, which I'm sure Desi loves.

965
02:43:00.480 --> 02:43:04.739
Lynne Jones: But this does have more of an impact on,

966
02:43:04.760 --> 02:43:08.909
Lynne Jones: When you'll see that delay in the footprint.

967
02:43:09.070 --> 02:43:21.200
Lynne Jones: It's kind of a much stronger effect than we actually intended. So, the DESI yearly completes the request by year 2 and Year 4 on the relevant parts of that footprint.

968
02:43:21.630 --> 02:43:22.480
Lynne Jones: And…

969
02:43:23.770 --> 02:43:33.110
Lynne Jones: It's not clear to me that our current metrics, especially when we evaluate at the end of year 10, that they really capture, any…

970
02:43:33.660 --> 02:43:48.050
Lynne Jones: potential benefits, or the, you know, full impact for this. So, I think you need, you know, we take this with a grain of salt. There certainly are some metrics which don't seem to be impacted by the end of year 10, which, as you would expect, the co-added depth

971
02:43:48.160 --> 02:43:55.709
Lynne Jones: If… if you have static science that only cares about co-added depth, at the end of year 10, the co-added depths are very similar.

972
02:43:55.950 --> 02:43:59.089
Lynne Jones: You didn't gain or lose by having

973
02:43:59.200 --> 02:44:01.969
Lynne Jones: This reshuffling of the observation cadence.

974
02:44:02.580 --> 02:44:14.050
Lynne Jones: And for some time domain, this will be an impact, and it's not obvious that when we look at the impact at the end of year 10, that we've… we've captured this properly or not.

975
02:44:15.850 --> 02:44:29.919
Lynne Jones: And then I was gonna say, we do have one simulation that we've not made yet, but that we will look at, which… it's not clear it will make a huge difference, but, the idea is instead of covering

976
02:44:30.140 --> 02:44:41.899
Lynne Jones: the whole footprint by year 4. Instead, we split it up into smaller chunks and let the last chunk sort of hang off the end, so it's at the end of year 5. And so.

977
02:44:42.080 --> 02:44:59.900
Lynne Jones: this is not necessarily the order you would do this, but I made a little plot to give the idea. So the different colors on the right are, like, doing 1,000… about 1,000 square degrees at the end of year two, another by the end of year 3, and so then you end up with…

978
02:45:00.110 --> 02:45:05.870
Lynne Jones: the final… 1,000 square degrees being completed at the end of year 10.

979
02:45:06.000 --> 02:45:06.880
Lynne Jones: Five.

980
02:45:07.420 --> 02:45:14.930
Lynne Jones: And… It's… it's worth noting, like, this… one thing that would happen here is that after year 3,

981
02:45:15.320 --> 02:45:21.830
Lynne Jones: And maybe even during year 3, you wouldn't be accelerating any of the redder bands. So…

982
02:45:22.080 --> 02:45:35.230
Lynne Jones: the redder band cadence would only be disturbed in two of these bands, but the bluer visits, like U and G, would still be disturbed in all of the band passes, and it's just…

983
02:45:35.800 --> 02:45:41.430
Lynne Jones: It might not be quite as strong of an effect, but it would still be there, and you would still…

984
02:45:41.660 --> 02:45:47.929
Lynne Jones: if you completed to 30 visits here, you're 5. You… oop, sorry.

985
02:45:48.040 --> 02:45:51.759
Lynne Jones: You would still be sort of waiting until about,

986
02:45:52.440 --> 02:46:03.649
Lynne Jones: your 7 or 8 until you started picking up more visits in this area. Unless we do something else again, like, add more visits in UNG to this part of the sky.

987
02:46:04.560 --> 02:46:09.229
Lynne Jones: Which is yet another variation on the survey strategy, and it's not…

988
02:46:09.450 --> 02:46:21.659
Lynne Jones: super clear, given that we've lost 10% of… lost. Since we… since we've updated our forecasts to be slightly fewer visits overall, what exactly the impact of that is?

989
02:46:22.830 --> 02:46:32.589
Lynne Jones: So now I'm going to turn it over to David to tell you about all of the good things that come out of this. Oh, sorry, Kathy, you have a quick question?

990
02:46:33.760 --> 02:46:38.810
Kathy Vivas: Yeah, my question, Lynn, is, how do you see…

991
02:46:38.960 --> 02:46:43.730
Kathy Vivas: This strategy will combine with running cadence.

992
02:46:45.530 --> 02:46:54.210
Lynne Jones: Yeah, so… so the original request was actually quite smart, because in… it doesn't…

993
02:46:54.340 --> 02:47:11.060
Lynne Jones: quite line up, now we've changed the start of the survey, but in the original version of the survey strategy, which… where we had started this simulation on November, this year 2 and Year 4 did align with the rolling cadence, so it was kind of like.

994
02:47:11.300 --> 02:47:12.970
Lynne Jones: Building on the rolling cadence.

995
02:47:13.110 --> 02:47:21.409
Lynne Jones: When we changed the start of the survey, we changed which part, like, the part of the sky, and so there's some changes about whether we're in a…

996
02:47:21.560 --> 02:47:34.739
Lynne Jones: up deck band or a down deck band, and so it doesn't quite align anymore, but that's something we could reinstate, and certainly, when you look at something like this, we could arrange these to be more smartly aligned with the rolling cadence.

997
02:47:36.160 --> 02:47:47.790
Lynne Jones: All right, I do want to make sure that David gets to tell you about the good aspects here. So, I'm actually going to cut off questions, and we can come back and discuss this more afterwards.

998
02:47:48.730 --> 02:47:50.550
Lynne Jones: David, is that okay? Are you…

999
02:47:50.550 --> 02:47:54.649
David Schlegel: Yeah, no, that's good, yeah. If I could share also.

1000
02:47:54.650 --> 02:47:55.480
Lynne Jones: Oh, sorry.

1001
02:47:55.480 --> 02:48:10.980
David Schlegel: Yeah, let me start with Lynn. It's… yeah, it's been wonderful working with this SCOC group, very responsive, and we are trying to, you know, approach this in a way that we're making minimal impact on, group and observing.

1002
02:48:13.810 --> 02:48:15.550
David Schlegel: Okay,

1003
02:48:16.490 --> 02:48:33.330
David Schlegel: This is my one slide about what is DESI. Hopefully you've heard of DESI before. It's really the most wonderful multi-object spectrograph on the planet right now. Over the past 5 years, it's mapped 47 million galaxies, 20 million stars.

1004
02:48:33.510 --> 02:48:45.359
David Schlegel: This is a fully, 100% DOE-funded, operation, so we're making use of the Noir Lab, telescope at Kit Peak.

1005
02:48:45.890 --> 02:48:54.690
David Schlegel: And then here's just the, the basic stats on this, where it's, it's 5,000, robotic fibers.

1006
02:48:54.910 --> 02:49:02.180
David Schlegel: Broad wavelength range, and the, data quality is just wonderful.

1007
02:49:02.670 --> 02:49:15.229
David Schlegel: Partly because every observation, we get simultaneously 100 standard stars, 600 skies, so we can really produce, excellent Poisson Limited spectra.

1008
02:49:16.360 --> 02:49:29.879
David Schlegel: Okay, these are just, representative spectrum. These are truly representative of what the DESI main survey has been doing, because we do these dynamic exposures where we adjust

1009
02:49:29.970 --> 02:49:45.470
David Schlegel: the open shutter time according to seeing, sky brightness, transparency. And I do want to know what you're talking about, uniformity. We're now operating DeckCam in exactly this mode.

1010
02:49:45.690 --> 02:50:05.579
David Schlegel: It's not too late for Ruben, and here's some quotes here, like, what do people say about dynamic exposures? All good things. From my perspective, one of the best things is we actually get a faster survey, because on good nights, we collect about twice as much data as you might expect.

1011
02:50:05.830 --> 02:50:11.010
David Schlegel: So we really do a lot of catching up on the Prime nights.

1012
02:50:12.040 --> 02:50:17.500
David Schlegel: So, you know, you're talking about, you want to get 10% back? You could easily get 10% back.

1013
02:50:20.250 --> 02:50:27.010
David Schlegel: Okay, starting… actually, I think it's tonight, Desi begins mapping

1014
02:50:27.240 --> 02:50:33.320
David Schlegel: Further south, so we've expanded the, DESI Run 1 footprint,

1015
02:50:33.430 --> 02:50:46.220
David Schlegel: this is the expanded footprint for DESE, that we'll be doing over the next couple years, and I just want to note this because this new area fully overlaps, the Rubin

1016
02:50:46.750 --> 02:50:47.780
David Schlegel: footprint.

1017
02:50:50.900 --> 02:50:58.250
David Schlegel: Okay, what we're interested in enabling here, though, is what we're calling DESI Run 2.

1018
02:50:58.480 --> 02:51:08.290
David Schlegel: So this will be the next phase of the DESI project starting in January 2029. Most of the dark time is focused on a cosmology

1019
02:51:08.500 --> 02:51:21.949
David Schlegel: experiment at redshifts 2 to 3.5 over this, 5,000 square degree equatorial footprint that Lynn was just showing you, and we chose this footprint,

1020
02:51:22.040 --> 02:51:37.279
David Schlegel: Not because this is necessarily our best footprint, it's a little too far south for us, but again, we're trying to minimize our impact on Rubin, so we're fitting this in on the northernmost good sky that Ruben was already planning on observing.

1021
02:51:40.390 --> 02:51:51.480
David Schlegel: Okay, so we have a proposed data exchange between Desi and Ruben, so this is following the policies of both collaborations.

1022
02:51:51.630 --> 02:52:09.770
David Schlegel: Where what we're asking from Ruben is, this, early imaging when necessary over this equatorial footprint, and that's what we'll be using for target selection, for DESIRUN2. So, you know, obviously we need this

1023
02:52:09.890 --> 02:52:16.160
David Schlegel: before Desi run 2, just so we can select the targets. And then…

1024
02:52:16.280 --> 02:52:29.270
David Schlegel: In return on this data exchange, what we've put on the table is, what we call 250 effective dark hours. What that corresponds to is four and a half months

1025
02:52:29.360 --> 02:52:45.150
David Schlegel: of DESI time? And, you know, what can be done with four and a half months of DESI time? Well, if you're talking about the kind of data we've been collecting typically for the last five years, that would be 4 million spectra.

1026
02:52:46.170 --> 02:52:56.930
David Schlegel: Or one can do deeper observations, so, you know, hundreds of thousands of spectra deeper, or some combination of all of these.

1027
02:52:57.040 --> 02:53:12.549
David Schlegel: Another point here is that the data are fully reduced, typically by the time I'm having coffee in the morning, and I get up early, so these data, they're spooled over to NERSC during the night, and

1028
02:53:12.550 --> 02:53:18.700
David Schlegel: you know, we'd say typically available within 12 hours. If you're impatient, they're actually available faster than that.

1029
02:53:22.040 --> 02:53:36.400
David Schlegel: Okay, these observations would be coordinated, between the two projects so that we make, you know, best use of this time. Examples of what could be done that,

1030
02:53:36.600 --> 02:53:50.820
David Schlegel: you know, this, DESI Rubin Working Group has been looking at so far are primarily based on observations in the two deep drilling fields that are available to both telescopes.

1031
02:53:50.820 --> 02:54:01.779
David Schlegel: So we could go quite deep in these deep drilling fields for, PhotoZee catalogs in these footprints.

1032
02:54:01.910 --> 02:54:09.029
David Schlegel: And also, we could be observing transients, within a day of discovery.

1033
02:54:09.030 --> 02:54:25.079
David Schlegel: And in practice, we can do these in such a way that we do observations where you're doing both simultaneously. So you're observing these deep fields, you're picking up new transients, and then you accumulate signal-to-noise on fainter objects at the same time.

1034
02:54:27.960 --> 02:54:42.250
David Schlegel: Desi's already measured about a quarter million redshifts in these two deep drilling fields already. These footprints, they're essentially the same size as the Rubin footprints. The field of view of both instruments is about the same.

1035
02:54:42.250 --> 02:54:55.589
David Schlegel: This just shows you the density of targets that we've already published, and just, you know, you can see the scale here, 50 arcseconds. So very high density of targets that we've already observed.

1036
02:54:55.780 --> 02:55:00.900
David Schlegel: So it's kind of a down payment of what's available in these deep drilling fields.

1037
02:55:02.360 --> 02:55:07.609
David Schlegel: We've also demonstrated, that we can go,

1038
02:55:07.730 --> 02:55:25.380
David Schlegel: quite a bit deeper, so we've taken data up to about 5 hours and demonstrated that we can get Poisson-level spectroscopy. So these are objects where we've observed

1039
02:55:25.570 --> 02:55:31.550
David Schlegel: Well, 3 or 4 hours, 24th, 25th magnitude, spectra.

1040
02:55:32.180 --> 02:55:33.829
David Schlegel: So this is all published.

1041
02:55:35.620 --> 02:55:53.799
David Schlegel: We also have a pretty good understanding of, redshift success rate as a function of integration time, so this is also something that was just published a couple months back, so this is looking at that success rate versus exposure time. It's actually very dependent on the

1042
02:55:53.820 --> 02:56:08.180
David Schlegel: you know, magnitude and colors of objects. So there are parts of color space where, you know, we're already doing this right now. We're observing, for example, luminous red galaxies at redshift of 1 in 15 minutes.

1043
02:56:08.180 --> 02:56:14.139
David Schlegel: Just because it's a color space where we know we can very quickly get a successful redshift.

1044
02:56:17.160 --> 02:56:28.100
David Schlegel: Okay, timeline, so we really need the first bit of, the Rubin imaging.

1045
02:56:28.150 --> 02:56:36.070
David Schlegel: available prior to the start of this Desi Run 2 in January 2029. And we could…

1046
02:56:36.110 --> 02:56:49.869
David Schlegel: you know, it'd be great if we could have all of this prior to 2029, but we're, you know, we're willing to work with Rubin and space this out over a few years, just so that the Rubin imaging is…

1047
02:56:50.000 --> 02:56:54.109
David Schlegel: You know, still prior to each footprint that we're observing with DESE.

1048
02:56:54.890 --> 02:57:03.630
David Schlegel: And then for this, data trade, so the DESI spectroscopy, this could be provided early.

1049
02:57:03.690 --> 02:57:17.379
David Schlegel: And or it could be cadence throughout the DESI Run 2 going through 2034. So, in other words, what we're imagining is, early in DESI Run 2, we would do

1050
02:57:17.380 --> 02:57:25.300
David Schlegel: Some deep observations in these deep drilling fields, and then over the course of, you know, maybe the next 4 or 5 years.

1051
02:57:25.380 --> 02:57:30.660
David Schlegel: You know, do, transient follow-up in those, in those fields.

1052
02:57:30.930 --> 02:57:38.270
David Schlegel: Okay, and then let me just end with DESI has no plans to coordinate with any other surveys, so it really is just…

1053
02:57:38.450 --> 02:57:42.599
David Schlegel: The Ruben that we're trying to coordinate with us on this.

1054
02:57:43.270 --> 02:57:47.109
David Schlegel: Okay, so that's… That's actually all I had to show.

1055
02:57:48.460 --> 02:58:00.409
David Schlegel: And I should note, there are some other DESE folks on as well. We tried to not overwhelm your call, but, but who could answer questions here?

1056
02:58:02.600 --> 02:58:03.410
Lynne Jones: Alright.

1057
02:58:03.810 --> 02:58:06.000
Lynne Jones: Thank you, David. That's,

1058
02:58:06.860 --> 02:58:13.489
Lynne Jones: That's a… that's a great demonstration of what… what we gain from the… from… from doing this.

1059
02:58:13.980 --> 02:58:17.919
Lynne Jones: Yeah, let's go with questions now. Dan?

1060
02:58:20.270 --> 02:58:25.359
Dan Taranu: When you say you don't have plans to coordinate with other surveys, do you mean imaging surveys, or also…

1061
02:58:25.560 --> 02:58:33.440
Dan Taranu: upcoming spectroscopics, I'm thinking of… Subaru PFS is definitely gonna target deep fields.

1062
02:58:35.470 --> 02:58:39.910
Dan Taranu: Foremost, I guess, is not really overlapping, but,

1063
02:58:40.410 --> 02:58:47.899
Dan Taranu: And I don't think they've asked for anything specific, but could you not also make the argument, well, we should also prioritize covering, like, the waves?

1064
02:58:48.090 --> 02:58:49.380
Dan Taranu: foremost.

1065
02:58:49.790 --> 02:58:53.110
Dan Taranu: Fields in a similar fashion.

1066
02:58:55.170 --> 02:58:59.569
David Schlegel: Sorry, that Desi should be coordinating with Waves, or what…

1067
02:59:00.180 --> 02:59:04.019
Dan Taranu: No, so, for the first part.

1068
02:59:04.540 --> 02:59:08.600
Dan Taranu: Are you coordinating at all with Subaru PFS?

1069
02:59:09.260 --> 02:59:11.690
Dan Taranu: In the deep drilling fields, for example.

1070
02:59:12.180 --> 02:59:31.329
David Schlegel: Yeah, so we don't have any coordination. So the only other coordination we have is we are obtaining other imaging that we need for DESI Run 2 from several different telescopes. We're running those imaging surveys within DESE. These are medium-band imaging surveys.

1071
02:59:31.400 --> 02:59:39.119
David Schlegel: So that is our other coordination. That's just internal, as For other spectroscopic projects.

1072
02:59:39.310 --> 02:59:43.300
David Schlegel: we're really far ahead on DESI, and it's… it's kind of…

1073
02:59:43.440 --> 02:59:51.520
David Schlegel: the question gets turned on its head of, you know, these other projects shouldn't repeat what DESI has already done.

1074
02:59:57.430 --> 02:59:59.290
Lynne Jones: Rachel?

1075
03:00:01.760 --> 03:00:05.569
Rachel Mandelbaum: Yeah, I have a comment and a question, if that's okay?

1076
03:00:05.680 --> 03:00:08.649
Rachel Mandelbaum: So the, the comment,

1077
03:00:09.190 --> 03:00:15.710
Rachel Mandelbaum: is more on the earlier part of the presentation. There was a statement that the metrics… you know.

1078
03:00:15.900 --> 03:00:17.660
Rachel Mandelbaum: For impact of… of…

1079
03:00:17.850 --> 03:00:31.410
Rachel Mandelbaum: these strategies on the science, may not capture, you know, the full impact. They may not be designed for that. I want to actually strengthen that to say that several of the desk metrics definitely

1080
03:00:31.510 --> 03:00:35.729
Rachel Mandelbaum: Do not capture the impact of this change in strategy, because they make assumptions about

1081
03:00:35.910 --> 03:00:41.780
Rachel Mandelbaum: the coverage in each band tracking each other in a way that these simulations violate, so I think it's a stronger statement that

1082
03:00:41.980 --> 03:00:48.740
Rachel Mandelbaum: you know, we'll likely need to update the metrics. But then, to ask my actual question,

1083
03:00:49.210 --> 03:01:07.900
Rachel Mandelbaum: One other timeline question that I wanted to understand is, when would a decision have to be made about doing this? Like, at what point would the Rubin strategy need to start to deviate from what it would otherwise be? Because that feels like a pretty important and relevant factor to understand.

1084
03:01:07.900 --> 03:01:12.030
Lynne Jones: Yeah, that is a great question.

1085
03:01:12.210 --> 03:01:16.970
Lynne Jones: I don't actually have a super good answer for you, except for the fact that I would say.

1086
03:01:17.600 --> 03:01:25.449
Lynne Jones: when we did run these simulations with the DESI survey, like, the DESI 3040 and the DESI yearly.

1087
03:01:25.620 --> 03:01:32.990
Lynne Jones: There was no change inside of year one, because the template tier, was put in at a higher priority.

1088
03:01:33.090 --> 03:01:35.440
Lynne Jones: I'm assuming, that's right. Is that right, Peter?

1089
03:01:37.690 --> 03:01:39.040
Peter Yoachim: Yep, that's right.

1090
03:01:39.040 --> 03:01:49.059
Lynne Jones: Yeah, so the templates being a higher priority meant that the templates were trying to get their visits, and because the templates take up most of year one, it just basically means there's no…

1091
03:01:49.250 --> 03:02:01.060
Lynne Jones: actual impact. So yes, we should decide as soon as possible, in practice, if our priority is actually first getting templates, which I think it should be.

1092
03:02:01.060 --> 03:02:09.669
Lynne Jones: But that's my opinion, of course. But then, then… then we have, then we have some time in practice.

1093
03:02:10.040 --> 03:02:20.489
Lynne Jones: we should definitely decide before… I mean, ideally, we decide as soon as possible, because it makes a difference for funding, for various things, right? But,

1094
03:02:21.140 --> 03:02:23.900
Lynne Jones: In practice, if we…

1095
03:02:24.530 --> 03:02:34.180
Lynne Jones: had a strategy in place, like, a decision in place 6 months from now, I think that that would… it would not really impact what we're actually doing on Sky.

1096
03:02:35.140 --> 03:02:50.069
Rachel Mandelbaum: Yeah, and to be clear, it's not like I wanted to push to wait until the latest possible moment. It was more trying to understand, is this, like, a right now decision, or, you know, we should do this within a couple of months, and the time will fly, so,

1097
03:02:50.230 --> 03:02:51.660
Rachel Mandelbaum: Thank you for clarifying.

1098
03:02:52.340 --> 03:02:59.119
David Schlegel: Yeah, I mean, let me just say, from the DESI side, though, we're eager to get to a decision point

1099
03:02:59.380 --> 03:03:10.789
David Schlegel: quickly, even if this isn't perturbing Rubin observations for another year. And that's just, you know, you can imagine that this could blow up the DESE Run 2

1100
03:03:10.920 --> 03:03:11.710
David Schlegel: Planning.

1101
03:03:15.620 --> 03:03:18.250
Lynne Jones: Yeah, definitely. That's good.

1102
03:03:18.680 --> 03:03:19.810
Lynne Jones: Roberto?

1103
03:03:22.470 --> 03:03:33.750
Roberto Assef: Yeah, so, there were… we discussed this a little bit on the impact on the AGN science collaboration in the latest telecom, and there were a couple of questions that, so I…

1104
03:03:34.160 --> 03:03:41.639
Roberto Assef: that I think were very interesting that came up. So, one was, is the DESE2 footprint fixed?

1105
03:03:41.760 --> 03:03:49.139
Roberto Assef: And if so, do you know how much does it overlap with Euro Cedar, at least the German part of Eurosida?

1106
03:03:51.510 --> 03:03:56.129
David Schlegel: Oh, that's a good question. I don't have that right in front of me, but maybe…

1107
03:03:56.870 --> 03:04:05.299
David Schlegel: Maybe one of my collaborators can figure this out quickly, but it's, I mean, you can look at it, it must be a fair amount of overlap, because we're…

1108
03:04:05.900 --> 03:04:09.829
David Schlegel: you know, we're overlapping LSST, which has a fair amount of overlap.

1109
03:04:12.870 --> 03:04:14.649
David Schlegel: So sorry, I don't pull that offhand.

1110
03:04:14.650 --> 03:04:20.480
Roberto Assef: That's fine. And the other question was,

1111
03:04:21.170 --> 03:04:40.270
Roberto Assef: So, right, currently the offer, right, is for the exchange to 250 dark time hours. One of the questions was, would it be instead, or, you know, with some partition, access to all the DES spectra taking over the…

1112
03:04:40.270 --> 03:04:42.280
Roberto Assef: LSSD footprint.

1113
03:04:43.250 --> 03:05:02.750
David Schlegel: So all of the DESI spectra, they are made public on some timescale, and in fact, the DESI DR1 release, which goes back a year now, most of that is actually over the Rubin footprint, in fact, just because we started

1114
03:05:03.000 --> 03:05:15.010
David Schlegel: Or it must be more than half, because we started kind of equatorial with, DESI. So we… we've… we've not discussed that. I… I…

1115
03:05:15.330 --> 03:05:21.220
David Schlegel: The DESE director's on, maybe he would like to respond to that.

1116
03:05:24.210 --> 03:05:25.140
David Schlegel: Michael?

1117
03:05:27.500 --> 03:05:29.330
Michael Levi: I'm sorry, what was the question?

1118
03:05:29.330 --> 03:05:30.150
David Schlegel: Oh.

1119
03:05:30.150 --> 03:05:31.080
Michael Levi: No, sorry.

1120
03:05:31.610 --> 03:05:33.029
David Schlegel: Do you want to just repeat your question?

1121
03:05:33.030 --> 03:05:37.860
Roberto Assef: Sure, sure. So, the question that came up was, would it be, would…

1122
03:05:38.050 --> 03:05:48.039
Roberto Assef: could there be access to the private DESE data, DESE2 data, on the regions that overlap the LSST footprint?

1123
03:05:48.040 --> 03:05:55.260
Michael Levi: Well, the 250 hours that we're talking about, I think the answer is yes, because you would be defining what those are.

1124
03:05:57.050 --> 03:05:59.020
Roberto Assef: Oh, right, but not… okay.

1125
03:05:59.350 --> 03:06:05.140
Roberto Assef: Okay, so I think the question was, instead of the 250 hours, access to all the spectra.

1126
03:06:06.980 --> 03:06:11.899
Roberto Assef: But, I mean, I see what you mean, right? So we could coordinate to make sure that it's reconciled.

1127
03:06:11.900 --> 03:06:14.429
Michael Levi: So, no, this is not an offer to open up the…

1128
03:06:14.580 --> 03:06:20.220
Michael Levi: proprietary, all of the proprietary, DESI data, I mean, that would be…

1129
03:06:20.380 --> 03:06:24.040
Michael Levi: It's possible for, you know, you as an individual

1130
03:06:24.320 --> 03:06:38.309
Michael Levi: to come in to DESE as an external collaborator with a particular scientific project to work with one of our other collaborators within DESE, to work with us on a topic where then you would have access to some of that information.

1131
03:06:38.500 --> 03:06:46.580
Michael Levi: So that's technically possible. You know, a broader… if you're asking a question, well, what about a collaboration between

1132
03:06:47.040 --> 03:06:52.209
Michael Levi: Desi and E. Rosita, you know, that's probably out of context for this meeting, but…

1133
03:06:52.210 --> 03:06:53.020
Roberto Assef: Yeah, absolutely.

1134
03:06:53.140 --> 03:06:57.370
Michael Levi: So, I think that's, you know, it would be a different format than what we're talking about here.

1135
03:06:57.880 --> 03:06:59.250
Roberto Assef: Okay. Thank you.

1136
03:07:00.950 --> 03:07:01.830
Lynne Jones: Toronto?

1137
03:07:03.070 --> 03:07:27.179
Saurabh Jha: Yeah, David, I just wanted to understand a little better, the DESI Run 2 needs, and maybe this is probably in some document, and I apologize if I have not read the appropriate document. But the example spectra you showed, like, even the long exposure ones, were getting down to, you know, 23rd, 24th magnitude, and I guess I understand that the DESI Run 2 is targeting somewhat higher redshift, so maybe, maybe slightly fainter is the goal, but…

1138
03:07:27.180 --> 03:07:50.389
Saurabh Jha: That was a surprising number a little bit to me, because if you're requesting 4-year-ish Rubin depth, that seems a lot deeper. So, you know, even in one visit, you're getting to close to 24th magnitude in U, 24.5 in G, so it feels like in a few visits, I know maybe you need signal and noise of 10 or 50 in the images, so maybe you just need very high,

1139
03:07:50.390 --> 03:08:02.159
Saurabh Jha: fidelity photometry, but it seems like in 4 years of Rubin, you get down to, like, 26th magnitude, and so is it possible that you can do the target selection with less data from Rubin?

1140
03:08:02.800 --> 03:08:14.469
David Schlegel: Yeah, we started with a request that was even deeper, and then dialed that back after some studies. The, the reason is, you know, these are the high-rich

1141
03:08:14.730 --> 03:08:25.899
David Schlegel: galaxies for which we can most efficiently and quickly, get a redshift. So they're not typical redshift 3 galaxies, they're the ones that we can

1142
03:08:26.030 --> 03:08:45.749
David Schlegel: get a redshift up. So the, primarily, you should think of these as the U-dropout galaxies, where if you want an efficient selection of… it's just the brightest U-dropout galaxies, you need about these 30 visit depths.

1143
03:08:45.890 --> 03:08:50.000
David Schlegel: The other objects that we're looking at, it's,

1144
03:08:50.300 --> 03:09:03.870
David Schlegel: It's these Lyman Alpha emitters, where their broadband photometry is really quite faint, because they're objects where we've picked them up as a 25th magnitude object.

1145
03:09:05.140 --> 03:09:15.510
David Schlegel: In medium band imaging, and then the broadband depth is deeper than that, but we still want the broadband color to distinguish those from interlopers.

1146
03:09:15.690 --> 03:09:20.489
David Schlegel: So, you know, we've looked pretty carefully at these depth requirements.

1147
03:09:22.470 --> 03:09:31.099
Julien Guy: Maybe to complement that, we, we have run target selection with random forest, assuming different depths of LSST.

1148
03:09:31.480 --> 03:09:35.910
Julien Guy: And, for four-year depths, which is equivalent to what we're discussing here.

1149
03:09:36.490 --> 03:09:43.740
Julien Guy: we get a purity of only, 55%. So, we really need that depth to,

1150
03:09:44.010 --> 03:09:46.700
Julien Guy: To get an efficient survey afterwards.

1151
03:09:48.400 --> 03:09:49.250
Saurabh Jha: Thanks.

1152
03:09:54.190 --> 03:09:55.390
Lynne Jones: John?

1153
03:09:58.860 --> 03:10:01.509
John Gizis: Oh, sorry, I didn't realize my camera was off.

1154
03:10:01.680 --> 03:10:20.289
John Gizis: So I guess, so my science case, it doesn't make any difference in the long run, you know, when you move the filters around, particularly the blue filters. But I do want to ask, because I've spent a lot of time lately hearing about Roman, it seems like this preferentially moves U and G observations out of Roman overlap.

1155
03:10:20.720 --> 03:10:25.929
John Gizis: In the upcoming years, and since we now believe it will really launch in August.

1156
03:10:26.180 --> 03:10:34.250
John Gizis: It seems like this is a very high impact on the American astronomical community, to lose

1157
03:10:34.570 --> 03:10:40.860
John Gizis: that potential science case, where presumably Roman can't observe U and G, so it was really adding something to the…

1158
03:10:41.030 --> 03:10:53.300
John Gizis: How will the SCOC, I don't think it's a question for David, address, you know, those sorts of concerns? What the views of other, communities that aren't necessarily LSST-centric?

1159
03:10:53.410 --> 03:10:56.420
John Gizis: dude.

1160
03:10:56.760 --> 03:10:59.819
John Gizis: So it's a question to raise a big issue, I guess.

1161
03:11:03.650 --> 03:11:05.100
Lynne Jones: So…

1162
03:11:05.860 --> 03:11:16.829
Lynne Jones: I don't think this is a question for me to answer, as I'm a ex-officio member of the SEOC. However, Fed is not actually, with us today, unfortunately, due to personal information.

1163
03:11:16.830 --> 03:11:20.539
Zeljko: in Ubers, so there may be too much background noise, I apologize.

1164
03:11:20.850 --> 03:11:21.360
Lynne Jones: There we go.

1165
03:11:21.360 --> 03:11:28.939
Zeljko: But, as Lynn said, we want SEOC to focus on scientific impact and cadence change.

1166
03:11:29.430 --> 03:11:37.070
Zeljko: And other aspects of this proposal will be considered elsewhere. There are, of course, many other aspects to it.

1167
03:11:37.520 --> 03:11:50.960
Zeljko: SCOC is in the best position to assess what would be impacts on MSSD science cases if we followed the strategy that we already simulated in response to this.

1168
03:11:51.770 --> 03:11:55.029
Zeljko: That is the main goal for this committee.

1169
03:11:55.310 --> 03:12:00.599
Zeljko: Those other aspects that are valid and that we need to consider, they will be addressed elsewhere.

1170
03:12:01.330 --> 03:12:04.299
John Gizis: Yeah, so overlap with external,

1171
03:12:04.780 --> 03:12:10.939
John Gizis: surveys is out of bounds, but a purely LSST consideration of the impact.

1172
03:12:11.420 --> 03:12:12.440
John Gizis: Is… Right.

1173
03:12:12.590 --> 03:12:15.679
John Gizis: That's very helpful for us to understand. Thank you so much.

1174
03:12:16.750 --> 03:12:18.540
Zeljko: Okay, sorry for the noise.

1175
03:12:21.570 --> 03:12:30.640
Lynne Jones: I think it is… it is certainly a goal in general of the SELC to consider as many aspects as possible of

1176
03:12:30.750 --> 03:12:37.360
Lynne Jones: of survey strategy changes as they come up. There is always, of course, the limitation that,

1177
03:12:38.700 --> 03:12:50.830
Lynne Jones: our knowledge within the project and our knowledge from… coming from the science collaborations will be imperfect. And so, if this comes up through the science collaborations, that they say, oh, you know, one of the…

1178
03:12:50.830 --> 03:13:02.140
Lynne Jones: things that I'm now concerned about is, like, how LSST survey coverage is… is going to impact my other science. It's true, that's a little bit,

1179
03:13:03.790 --> 03:13:16.390
Lynne Jones: It's a bit of a question that's, like, out of bounds for what the LSSD is, because the priority of the SEOC has to remain, the LSSD science, but it certainly is something that they can… can think about.

1180
03:13:16.970 --> 03:13:20.720
Lynne Jones: Nobody's going to say that you… they couldn't consider something.

1181
03:13:23.070 --> 03:13:27.340
Lynne Jones: But yeah, it… there are always…

1182
03:13:27.770 --> 03:13:33.850
Lynne Jones: There are always, unfortunately, things that we're gonna miss, of course, but…

1183
03:13:34.150 --> 03:13:38.359
Lynne Jones: at the end of the day, we are here to try to make the best LSST.

1184
03:13:39.000 --> 03:13:42.180
Lynne Jones: And, let's see.

1185
03:13:43.470 --> 03:13:59.600
Lynne Jones: We have… we do have some more time here for questions. We can talk about, Desi, and we can talk about other questions from other… other parts of the… the session. I do want to make sure, if we have things we didn't want to circle back to from the rest of the…

1186
03:14:00.030 --> 03:14:05.810
Lynne Jones: the workshop, that we… that we cover those as well. But… so damn.

1187
03:14:07.890 --> 03:14:11.130
Dan Taranu: Yeah, sorry, I had a lingering question from before, and I…

1188
03:14:11.430 --> 03:14:15.569
Dan Taranu: Missed the discussion about the DDFs earlier, but looked at… looked at the slides.

1189
03:14:18.210 --> 03:14:36.439
Dan Taranu: Do you have a lot of metrics for specifically optimizing the radial dithering in the… or translational dithers in the DDFs? Is it basically trading off the area that gets full coverage versus the larger area that might get less coverage on the outer skirts, and…

1190
03:14:37.100 --> 03:14:43.630
Dan Taranu: Does that one failed detector that's very near the edge of the focal plane, but not quite, does that change things at all, or…

1191
03:14:44.610 --> 03:14:51.890
Lynne Jones: Okay, so this is… This is a good question that touches along a lot of,

1192
03:14:52.000 --> 03:14:55.619
Lynne Jones: On a particular math aspect as well.

1193
03:14:55.850 --> 03:14:59.409
Lynne Jones: With math, what we do is,

1194
03:15:00.550 --> 03:15:04.880
Lynne Jones: We do imprint the camera footprint when we're calculating our metrics.

1195
03:15:05.000 --> 03:15:09.540
Lynne Jones: We don't necessarily have the resolution to really say this is the…

1196
03:15:10.420 --> 03:15:21.119
Lynne Jones: We don't usually run our metrics at super high resolution, so we don't… we represent the chip gap statistically, but not necessarily exactly, for example.

1197
03:15:21.120 --> 03:15:34.770
Lynne Jones: That said, we should take our camera footprint and take out the… because we do have the chip gaps in our camera footprint. We haven't taken out the missing detector, and we should do that, and that's a very good point.

1198
03:15:34.820 --> 03:15:39.160
Lynne Jones: So I don't know if the missing detector makes a difference or not.

1199
03:15:39.320 --> 03:15:49.559
Lynne Jones: The metrics we have for the deep drilling fields and the dithering really do just, as you say, look at the trades between as you get wider.

1200
03:15:49.640 --> 03:16:03.390
Lynne Jones: bigger dithers, you decrease the cadence, so we do have things that are looking at, like, what is the… essentially, what is the cadence within the deep drilling fields, and you do, decrease the coated depth.

1201
03:16:03.480 --> 03:16:13.849
Lynne Jones: Because you're spreading that over more area. I don't know specifically we look at, like, what happens with the roll-off, but as you… towards the edges of the dinner, but we,

1202
03:16:14.540 --> 03:16:23.100
Lynne Jones: We definitely… that is the primary driver for having smaller jitters, is you get a better cadence, and you get deeper depths.

1203
03:16:25.570 --> 03:16:26.970
Lynne Jones: That's a problem.

1204
03:16:27.400 --> 03:16:51.980
Saurabh Jha: Yeah, just to follow up on that, so yeah, it's true that the metrics don't always capture that for the DDFs, but this is something we put in the DDF task force of the SCOC we posed to the science collaborations, and basically, you know, you can imagine you have this umbra of the DDF, and then you have this penumbra, which gets bigger the more you dither, and you could imagine some science cases would like, you know, basically you're making your effective area of a deep drilling field just bigger.

1205
03:16:51.980 --> 03:17:01.210
Saurabh Jha: But it turns out, in terms of, like, external synergies for the DDFs and, like, you know, space-based imaging and stuff, that the Rubin footprint is already big.

1206
03:17:01.210 --> 03:17:22.599
Saurabh Jha: Right? Compared to the size of these fields. And so, increasing that area was not generally a science win across the board for the science collaborations. And so, from the science collaborations point of view, keeping that as small as possible, like maybe just dithering over chip gaps, that would be the ideal. And then that would improve the cadence, just as Lynn said, and get the highest co-added depth.

1207
03:17:22.600 --> 03:17:29.029
Saurabh Jha: So basically, the dithering we would want to do with EDFs is as little as possible that DM says is okay.

1208
03:17:33.650 --> 03:17:46.570
Lynne Jones: And it does seem like we've, moved on to other questions, so I'm gonna make sure we thank all of the DESI people for joining us. Of course, like, please stay on so that in case we come back to more DESI questions, but thank you for…

1209
03:17:47.030 --> 03:17:49.589
Lynne Jones: For your presentation and joining us.

1210
03:17:49.690 --> 03:17:51.920
Lynne Jones: Neal?

1211
03:17:53.570 --> 03:18:09.800
William Nielsen Brandt: Yeah, sorry, I just did want to ask one other just point of clarification, just to be sure I didn't miss something along the way. So, if the DESE plan were implemented, what would you anticipate the effects upon the deep drilling fields, if anything, material, to be?

1212
03:18:09.960 --> 03:18:18.960
William Nielsen Brandt: I presume, we would still maintain truly excellent time domain coverage in any of the deep drilling fields that are also in that footprint.

1213
03:18:19.220 --> 03:18:25.959
William Nielsen Brandt: That would… that's obviously very important, and it's key to the nature of the deep drilling field, you know, in terms of the U and G bands.

1214
03:18:26.170 --> 03:18:31.440
William Nielsen Brandt: But are there other effects you could anticipate there that we should be thinking about?

1215
03:18:32.630 --> 03:18:35.750
Lynne Jones: I have.

1216
03:18:35.750 --> 03:18:49.980
David Schlegel: Let me jump in for a sec. I mean, I don't see why this would affect Ruben at all, but we do have the opportunity on the DESI side. We don't have any other cadencing requirements or anything like that, so if, you know, if the request

1217
03:18:50.380 --> 03:18:59.299
David Schlegel: for this data exchange from DESE said, hey, you know, we really want to do the DESI follow-up on the following days, or, you know, just

1218
03:18:59.460 --> 03:19:07.069
David Schlegel: you know, more of a triggered response, you know, we could do that, so that… I think we'd certainly see that on the DESE side.

1219
03:19:08.060 --> 03:19:12.360
William Nielsen Brandt: Thanks. Yes, I'm mostly interested in the Rubin side here.

1220
03:19:12.740 --> 03:19:16.369
William Nielsen Brandt: So, Lynn, did you have any insights there? I mean.

1221
03:19:17.120 --> 03:19:31.950
Lynne Jones: So it's true, I haven't specifically looked to see in the Desi Sims, did we change the deep drilling coverage? However, I would guess we did not, and so Rob is shaking his head.

1222
03:19:31.950 --> 03:19:39.460
Lynne Jones: Because of the fact that, the deep drilling fields are programmed separately, and they're in a higher tier, so they should…

1223
03:19:39.710 --> 03:19:49.050
Lynne Jones: continue as, as previously planned, and we didn't, like, decrease the DDF coverage because we were getting this other coverage as well.

1224
03:19:50.550 --> 03:19:51.250
Lynne Jones: That's wrong?

1225
03:19:51.250 --> 03:19:51.930
William Nielsen Brandt: Okay.

1226
03:19:53.330 --> 03:20:01.560
Saurabh Jha: Yeah, I was gonna say the same thing, that the DFs are not affected. It's really the WFD visits in that northern part of the sky that get shipped.

1227
03:20:01.790 --> 03:20:06.979
William Nielsen Brandt: Very good. I just wanted to be sure there wasn't some second-order effect that I wasn't appreciating, but thank you. Okay.

1228
03:20:08.010 --> 03:20:17.150
Lynne Jones: No, in fact, it's usually the other way around, that the DDFs affect the coverage of the rest of the sky, because they are in this higher priority tier.

1229
03:20:17.460 --> 03:20:29.790
Lynne Jones: And this was, for example, one of the reasons we had to move the ultra-deep season of EDFS out of year one, because we were already a little stretched thin in that part of the sky.

1230
03:20:29.990 --> 03:20:39.310
Lynne Jones: And so… and with, like, assuming a higher fault rate and lower efficiency at the start of the survey, it made things a little bit too tight.

1231
03:20:39.870 --> 03:20:48.920
Lynne Jones: So… Yeah, the deep drilling fields are… Or… Very strongly,

1232
03:20:49.650 --> 03:20:52.280
Lynne Jones: defined, I guess, is one way to put it.

1233
03:20:53.030 --> 03:20:59.489
Lynne Jones: All right, do we have other questions? We have about… 10 more minutes.

1234
03:20:59.990 --> 03:21:01.060
Lynne Jones: Yeah, Rachel.

1235
03:21:02.840 --> 03:21:05.250
Rachel Mandelbaum: Yeah, so this is a question for…

1236
03:21:05.370 --> 03:21:15.650
Rachel Mandelbaum: Rubin Leadership, so looking at who's on the call, maybe Phil or Jelco? So… For the,

1237
03:21:16.130 --> 03:21:28.929
Rachel Mandelbaum: Does the observing time that would come out of this exchange? Like, does Rubin leadership have an idea for a model for how to allocate that to maximize the benefit to the Rubin community?

1238
03:21:32.290 --> 03:21:37.550
Phil Marshall (SLAC/Rubin): I would say… Not yet, but we have some…

1239
03:21:37.780 --> 03:21:40.239
Phil Marshall (SLAC/Rubin): I've already done some thinking about,

1240
03:21:41.160 --> 03:21:47.399
Phil Marshall (SLAC/Rubin): other in-kind contributions of telescope time and so on. So…

1241
03:21:47.710 --> 03:21:56.119
Phil Marshall (SLAC/Rubin): we would look at what we've done there with the, you know, contributions from Subaru and other places, and think about whether we could

1242
03:21:56.270 --> 03:22:05.479
Phil Marshall (SLAC/Rubin): treat this, you know, the DESI time or the DESI data in a similar way. I mean, we're trying to be…

1243
03:22:06.400 --> 03:22:19.159
Phil Marshall (SLAC/Rubin): even-handed in our support of the RSST science community, and YLAB has, you know, open skies policies that we're trying to observe as well. So, we would try and make sure that the…

1244
03:22:19.390 --> 03:22:23.729
Phil Marshall (SLAC/Rubin): that the benefits of any data exchange with DESI were… were shared around.

1245
03:22:25.680 --> 03:22:32.399
Phil Marshall (SLAC/Rubin): it may well be that some parts of the RSSD science community would benefit more from this than others.

1246
03:22:32.530 --> 03:22:34.660
Phil Marshall (SLAC/Rubin): and… You know, we would…

1247
03:22:34.820 --> 03:22:46.539
Phil Marshall (SLAC/Rubin): we would try and take that into account as well, but, I mean, there would… there would be something significant to work out here in terms of how best to use these data. I mean, the…

1248
03:22:46.740 --> 03:22:50.600
Phil Marshall (SLAC/Rubin): I think… I suppose there's some hope that,

1249
03:22:51.390 --> 03:23:11.149
Phil Marshall (SLAC/Rubin): that there could be simple ways, and I think David was highlighting these in his slides, simple ways for… for using DESI spectra, kind of synoptically, similar to the LSST survey itself, so looking for applications that already serve multiple science cases, and PhotoZ training is one of those, for example.

1250
03:23:12.810 --> 03:23:30.839
Nathalie Palanque-Delabrouille: Yeah, so I'm just reiterating what David showed, but with just a one-slight summary of topics that we've discussed for collaboration between DESI, Spectra, and LSST, and in the deep billing fields in particular, we can really contribute to the DSST science, the Rubin science.

1251
03:23:30.840 --> 03:23:35.040
Nathalie Palanque-Delabrouille: By providing the spectra of live transients, or a supernovae host.

1252
03:23:35.040 --> 03:24:00.030
Nathalie Palanque-Delabrouille: And for the weak lensing science, we can also significantly support that science by providing the very deep spectra for, photocalibration. We've, tested, up to a few hours, but we've shown that with the instrumentation upgrade that we're planning for this second phase of DESI, we could actually, stack up to 80

1253
03:24:00.030 --> 03:24:04.040
Nathalie Palanque-Delabrouille: hours of Spectra, and so that would really provide very deep,

1254
03:24:04.040 --> 03:24:21.400
Nathalie Palanque-Delabrouille: spectra for calibration of photos Zs, and so this is an example of what we've been discussing so far, but of course, this is not set in stone, and we can definitely continue to identify what would be the best, outcome of this data exchange.

1255
03:24:23.920 --> 03:24:24.890
Phil Marshall (SLAC/Rubin): Thanks, Natalie.

1256
03:24:39.940 --> 03:24:40.600
Lynne Jones: Wow.

1257
03:24:41.180 --> 03:24:45.409
Lynne Jones: I think we've, we've… Covered a lot, and we've had a lot to think about.

1258
03:24:45.630 --> 03:24:48.520
Lynne Jones: I don't see any more questions or hands coming up.

1259
03:24:49.120 --> 03:24:50.180
Lynne Jones: So…

1260
03:24:50.390 --> 03:24:56.609
Lynne Jones: I'm going to say thank you all very much for coming and sharing and, talking with us today.

1261
03:24:57.170 --> 03:25:12.679
Lynne Jones: And all of the… all of you community members and all of the SCOC members, thank you, as well. It's been, a lot to go through and to talk about, and I hope we all have some good ideas of how the survey will start.

1262
03:25:13.180 --> 03:25:21.790
Lynne Jones: What we're doing in terms of the strategy, and where we might be going for the next year or two, as well as over the 10 years of the survey.

1263
03:25:23.730 --> 03:25:33.620
Lynne Jones: you know what, I think we can be done. We can have a few minutes back out of your day, and thank you all for coming. We do have Monday coming up, and, I'm…

1264
03:25:34.250 --> 03:25:45.920
Lynne Jones: hoping that Fed will be back to, help us coordinate for that, but otherwise, we'll be reaching out to the science collaborations to organize feedback. Monday is a little bit more for

1265
03:25:46.180 --> 03:25:50.619
Lynne Jones: feedback from you in the science collaborations to the SEOC.

1266
03:25:51.870 --> 03:25:53.410
Lynne Jones: So, thanks very much.

1267
03:25:53.530 --> 03:25:55.910
Lynne Jones: We'll talk to you on Monday.

1268
03:25:58.240 --> 03:25:58.970
Phil Marshall (SLAC/Rubin): Excellent.