WEBVTT

1
00:00:05.609 --> 00:00:06.839
thanks again.

2
00:00:09.240 --> 00:00:10.290
Matthew Bellis: started recording.

3
00:00:11.700 --> 00:00:12.450
Matthew Bellis: alright.

4
00:00:13.320 --> 00:00:14.160
So.

5
00:00:15.240 --> 00:00:18.840
Julie Hogan: We are going to go through as promise.

6
00:00:20.040 --> 00:00:25.530
Julie Hogan: The higgs the towel towel example that will be using for your analysis flow.

7
00:00:26.790 --> 00:00:45.210
Julie Hogan: And then I would love to also talk about some other people's individual analyses So hopefully please me please be willing to raise your hands and we'll we'll collaborate a little bit on building building your cms analysis selection.

8
00:00:46.440 --> 00:00:59.610
Julie Hogan: So let's jump in to the first question So the first thing we have to decide whenever we're building an analysis is is what is our signal and what are we searching for.

9
00:01:00.630 --> 00:01:06.180
Julie Hogan: And so, in this case for higgs to tau tau there is a paper.

10
00:01:07.410 --> 00:01:13.050
Julie Hogan: This is the cms paper that was published searching for.

11
00:01:14.250 --> 00:01:20.670
Julie Hogan: higgs decaying to a pair of uptown leptons in this data set that we're able to consider now in open data.

12
00:01:21.840 --> 00:01:31.710
Julie Hogan: They have many, many channels, so they have each tile leptons decays hydroponically or electronically to an electron or am yuan leading to six different final states.

13
00:01:32.340 --> 00:01:40.710
Julie Hogan: we're just going to focus on one it's very probable that this paper was put together by a team of different.

14
00:01:41.550 --> 00:01:48.330
Julie Hogan: For an analyst perhaps you know you could imagine something like one graduate student per final state or something like that.

15
00:01:49.200 --> 00:02:01.290
Julie Hogan: To eventually build into this paper, so if you're if you're considering a physics process that has a lot of different possible decays often the easiest way to go forward is to break it down a little bit.

16
00:02:01.950 --> 00:02:12.780
Julie Hogan: into things that are similar to each other and then and then recombine at the end, so we can look just at the at the fineman diagrams to figure out what we expect.

17
00:02:13.410 --> 00:02:21.330
Julie Hogan: So these are the two fineman diagrams that will consider we've got either glue on fusion producing.

18
00:02:21.810 --> 00:02:31.410
Julie Hogan: This this higgs boson or we've got vector bows on fusion producing this higgs boson and they might give us different final states, so of course in.

19
00:02:31.800 --> 00:02:43.740
Julie Hogan: In the top five a diagram the higgs is produced by itself, I in the lower fineman diagram the higgs is produced, along with to corks which will probably end up appearing in the cms detector as jets.

20
00:02:45.000 --> 00:03:00.030
Julie Hogan: So, given that we were told our final state is a hydroponic towel and a towel, that the case to me yuan's We certainly expect that we should be able to find that hydroponic TAO object.

21
00:03:01.230 --> 00:03:03.930
Julie Hogan: We should be able to find a new on.

22
00:03:06.330 --> 00:03:26.850
Julie Hogan: All of the neutrinos at least the two neutrinos should give some missing missing and energy in the detector and there may be a couple of jets so there may be a couple of probably light Cork jets if this vector goes on fusion process is what happened to produce the higgs boson.

23
00:03:29.220 --> 00:03:29.790
Julie Hogan: So.

24
00:03:31.350 --> 00:03:39.660
Julie Hogan: Based on these particles which trigger or triggers would be useful to make sure we have our signal events.

25
00:03:40.380 --> 00:03:54.630
Julie Hogan: Representative data set so Is anyone willing to you know unmute or write in the chat What was your idea for a higgs to town town trigger That would be good for this final state what type of objects might you want, in your trigger.

26
00:04:05.790 --> 00:04:12.150
Julie Hogan: Okay, tell part to tell plus x is starting to come in these are what I would consider names of data sets.

27
00:04:13.230 --> 00:04:20.880
Julie Hogan: names of data sets and so that's very important and and you're you're absolutely correct will end up looking at the towel plus x data set.

28
00:04:23.310 --> 00:04:25.500
Julie Hogan: But we could be more specific than X.

29
00:04:27.270 --> 00:04:28.920
Julie Hogan: here for this final stage.

30
00:04:34.710 --> 00:04:43.890
Julie Hogan: I don't have the menu, to see if anybody's raised their hand, so I think we can be totally flexible and you can just unmute and speak, if you would like to share anything in this session.

31
00:05:00.120 --> 00:05:02.610
Julie Hogan: Okay, so the obvious, you know.

32
00:05:03.960 --> 00:05:16.560
Julie Hogan: shouldn't say it that way, the most straightforward thing we could do okay glue glue to H, to tell tell that again is the name of like a that might be our signal sample that might be the data set we look up.

33
00:05:17.370 --> 00:05:25.890
Julie Hogan: To be our signal sample in terms of the actual trigger we probably want something that's either going to give us TAOs.

34
00:05:26.970 --> 00:05:31.740
Julie Hogan: New ons or a combination of both so.

35
00:05:33.480 --> 00:05:41.700
Julie Hogan: This analysis is really perfect for a trigger and our on our jargon, for this is a cross trigger that has more than one object in it.

36
00:05:42.330 --> 00:05:51.330
Julie Hogan: And so the trigger, that is, that will end up using here looks like this, so the ahl T path has an isolated new on.

37
00:05:52.290 --> 00:06:03.120
Julie Hogan: With some criteria on it that we're about to talk about and it requires a particle flow TAO so we have this combination of two objects, and so this is.

38
00:06:04.050 --> 00:06:14.550
Julie Hogan: This is what I mean by the difference between a data set name and a trigger we will certainly find this type of trigger in the towel plus x data set.

39
00:06:15.300 --> 00:06:33.360
Julie Hogan: We will certainly find that many events in the glue glue H2 tau tau Monte Carlo sample past this trigger, but this is an actual specific flag, we can ask for that will give us a muon and TAO which pretty pretty well describes our final state.

40
00:06:37.410 --> 00:06:47.850
Julie Hogan: Okay, so which objects, would we require did anybody come up with a hypothesis about what you might require for let's let's just throw it out there, for your towel.

41
00:06:48.960 --> 00:06:51.150
Julie Hogan: How am I use select a towel.

42
00:07:19.320 --> 00:07:29.940
Julie Hogan: Okay, so as soon as you said, with with them you on or an electron, and so we know how the tau is going to decay so we've already declared that one of them is going to decay hydroponically.

43
00:07:30.510 --> 00:07:41.760
Julie Hogan: And one of them is going to decay to a new one, so the towel, that the case to imbue on in terms of the cms detector shows up as a new one.

44
00:07:42.630 --> 00:07:52.620
Julie Hogan: Okay, we have we actually don't have a great way to tell that it was ever a towel, because we aren't going to detect it in the detector as a towel, it will only look like a new on.

45
00:07:53.130 --> 00:08:03.090
Julie Hogan: So when we're asking the cms software when we're asking our route files about TAOs we're getting the hydroponic we're getting the hydroponic versions.

46
00:08:03.870 --> 00:08:12.690
Julie Hogan: So what we're going to need to do is look in the is excuse me we're going to need to look in these TAO collection, so if I go, let me get my.

47
00:08:13.740 --> 00:08:15.570
Julie Hogan: index for the lesson here.

48
00:08:18.450 --> 00:08:30.000
Julie Hogan: We go back here the musicians and towers page will be a good reference for us for this for this analysis example we're going to want to open, you know this towel collection.

49
00:08:31.650 --> 00:08:39.360
Julie Hogan: particle flow town producer we're going to want to open this towel collection or we're going to want to make selections on the objects there for our hydroponic TAO.

50
00:08:41.400 --> 00:08:44.880
Julie Hogan: kind of similar idea for the new on.

51
00:08:50.850 --> 00:09:06.900
Julie Hogan: So we certainly need at least one you on and we need one hydraulic towel okay so um you on means you know, opening up are you on collection and looking through it a towel hydroponic town means opening up our town collection and looking through it.

52
00:09:08.490 --> 00:09:19.200
Julie Hogan: And if we wanted to target that vector bows on fusion let's say that's that's what I was interested in, as opposed to the other type of production, then I would probably require that there are two jet's.

53
00:09:20.370 --> 00:09:30.270
Julie Hogan: So that would be that would be something to consider you could also decide you're not interest you want to try to get rid of vector bows on fusion, you could reject jets.

54
00:09:32.190 --> 00:09:47.250
Julie Hogan: Okay, the last thing which which Cecilia pointed out here on the on the chat, we will have some neutrinos it's a little bit tricky to decide always whether we need a missing et requirement, there are neutrinos involved.

55
00:09:47.760 --> 00:10:00.600
Julie Hogan: They might be they might not not have a very large momentum in this case, so we may or may we may have to study our signal to decide if it really helps us to use that type of condition.

56
00:10:02.790 --> 00:10:08.700
Julie Hogan: Okay, so let's do this we've got let's let's have this trigger.

57
00:10:11.130 --> 00:10:16.260
Julie Hogan: which has this language in it about him, you on so this language here refers to the new one.

58
00:10:18.120 --> 00:10:35.970
Julie Hogan: And this language here refers to the towel that hydraulic TAO so what kinematic criteria might you require for each object So what did what did anyone come up with let's say for for the new on did anyone have a.

59
00:10:35.970 --> 00:10:39.990
Julie Hogan: kinematic set of kinematic criteria that they were thinking of for the meal on.

60
00:10:40.830 --> 00:10:42.540
Franz Glessgen: would be nice to have an isolated mural.

61
00:10:43.740 --> 00:10:44.760
Julie Hogan: Yes, I agree.

62
00:10:45.030 --> 00:10:47.490
Julie Hogan: that's not let that's kind of the next question.

63
00:10:47.700 --> 00:10:52.710
Julie Hogan: So what about what about things like literally kinematics.

64
00:10:54.030 --> 00:10:57.060
Franz Glessgen: Big into the gap between the two howdy case.

65
00:10:59.040 --> 00:11:04.890
Julie Hogan: Okay big angular gap between the two towers Okay, so it should be separated from the from the hydroponic tell.

66
00:11:08.340 --> 00:11:11.580
Julie Hogan: Any other kinematic criteria such as momentum.

67
00:11:14.130 --> 00:11:14.640
Julie Hogan: places and.

68
00:11:14.880 --> 00:11:17.640
Franz Glessgen: They come from the higgs you need to rest most to be the higgs.

69
00:11:19.980 --> 00:11:20.400
Julie Hogan: yep.

70
00:11:20.640 --> 00:11:27.450
Julie Hogan: Okay yep so when we when we are looking at combining the different objects so we're word leaping ahead a little bit.

71
00:11:28.530 --> 00:11:32.160
Julie Hogan: let's stay way back down on just just the new on itself.

72
00:11:34.680 --> 00:11:36.030
Julie Hogan: or just the Tower itself.

73
00:11:46.440 --> 00:11:49.440
Franz Glessgen: Maybe a high high momentum for both object object.

74
00:11:49.860 --> 00:11:52.170
Julie Hogan: A highest momentum for both objects.

75
00:11:52.530 --> 00:11:55.920
Julie Hogan: Absolutely absolutely a good idea, and what the real.

76
00:11:55.950 --> 00:12:01.410
Julie Hogan: The real reason I ask this question is because I want to point out, though a connection here to this trigger path.

77
00:12:02.220 --> 00:12:11.490
Julie Hogan: So we actually have some criteria that are already imposed on you, by the fact that we have chosen a trigger, and this will happen.

78
00:12:12.840 --> 00:12:19.710
Julie Hogan: almost always when you're selecting triggers for yours, for your analysis, there are some constraints.

79
00:12:20.940 --> 00:12:34.290
Julie Hogan: Of the trade off in cms is that of course we would love to just be able to save everything right but we can't so we have to determine what our physics choices are in order to decide what events we save.

80
00:12:35.580 --> 00:12:44.490
Julie Hogan: And so it's easier to save all the events from a certain physics process if that physics process involves high energies high momentum.

81
00:12:45.150 --> 00:12:49.680
Julie Hogan: And so the lower you try to go in your in your.

82
00:12:50.460 --> 00:13:00.030
Julie Hogan: let's say energy requirements in a trigger the more likely it is that you will end up with a trigger that is pre scaled which you learned about in your pre exercise.

83
00:13:00.360 --> 00:13:09.090
Julie Hogan: And when you're doing an analysis you tend to want to avoid that you want to choose a trigger where you are going to actually select all of the data.

84
00:13:10.620 --> 00:13:20.490
Julie Hogan: That came through that trigger as opposed to some random subset of it, so this trigger language tells us that the new on.

85
00:13:21.630 --> 00:13:29.040
Julie Hogan: The new on selected by this trigger will have at least 17 GB of energy or excuse me transverse momentum.

86
00:13:31.140 --> 00:13:46.800
Julie Hogan: And this language here tells us that the new on will live in the central region of the detector so if pseudo rapidity starts at zero here's my cms detector pseudo rapidity starts at zero and goes towards infinity.

87
00:13:47.910 --> 00:13:55.620
Julie Hogan: 2.1 is is basically this this wedge let's say in the detector but, of course, both top and bottom.

88
00:13:57.510 --> 00:14:11.100
Julie Hogan: So we automatically have some kinematic requirements that have been placed on our new on we will lose all the new ones below 17 gv, we will lose all of the nuances that are at large pseudo rapidity.

89
00:14:12.900 --> 00:14:16.290
Julie Hogan: we've got the same type of thing going on on our town this.

90
00:14:16.350 --> 00:14:19.140
Julie Hogan: Is legal here 20 yep question.

91
00:14:20.250 --> 00:14:27.360
Dillon Fitzgerald: Do you necessarily lose him isn't this just what you trigger on so couldn't there be other than the events that are below that creature.

92
00:14:27.630 --> 00:14:33.300
Julie Hogan: that's true there could be other it so so thanks for correcting me the the myth that the trigger me one.

93
00:14:34.380 --> 00:14:34.950
Julie Hogan: here.

94
00:14:35.130 --> 00:14:47.580
Julie Hogan: Is 17 GB or more, and in this central region so you have no guarantee of any new on that's a lower momentum and higher pseudo rapidity, although of course they might exist.

95
00:14:48.960 --> 00:14:49.140
yeah.

96
00:14:51.240 --> 00:14:51.480
Dillon Fitzgerald: yeah.

97
00:14:52.770 --> 00:14:55.350
Edgar Fernando Carrera Jarrin: And same sorry sorry to.

98
00:14:55.560 --> 00:14:59.850
Edgar Fernando Carrera Jarrin: interrupt Julie, and this is where the trigger.

99
00:15:00.990 --> 00:15:16.140
Edgar Fernando Carrera Jarrin: trigger object analyzer could come handy because, with that type of analyzer you can actually make sure that the new on you, you trigger done is what you think it is.

100
00:15:17.850 --> 00:15:18.090
Julie Hogan: yeah.

101
00:15:18.180 --> 00:15:18.780
Absolutely.

102
00:15:20.730 --> 00:15:33.840
Julie Hogan: So Okay, so we talked about them you on same idea with our TAO here we're being told that this this trigger is selecting events that have a hydraulic towel, with a momentum of 20 GB.

103
00:15:35.190 --> 00:15:44.250
Julie Hogan: And what's important to know is that whenever you're whenever you're relying on a trigger to select objects, for you.

104
00:15:44.940 --> 00:15:55.350
Julie Hogan: When it says 17 gv we would love for that to be a step function like this new on if the highest momentum you add in the event is between zero and 17 we get nothing.

105
00:15:56.040 --> 00:16:15.030
Julie Hogan: And then, as soon as there's a 17 gv new on boom we detect it and we trigger on it, that of course is to perfect of a scenario, so you should always imagine a an error function in your mind sorry i'm, at least on my screen i'm drawing this backwards, so if i'm inverted.

106
00:16:16.230 --> 00:16:17.340
Julie Hogan: invert me in your brain.

107
00:16:19.230 --> 00:16:35.490
Julie Hogan: You should always imagine some sort of some sort of curvy turn on function in momentum when you're thinking about this, so what this usually means is that it as analysts, it is safer to add a little momentum buffer.

108
00:16:36.510 --> 00:16:37.680
Julie Hogan: On our selection.

109
00:16:38.850 --> 00:16:46.590
Julie Hogan: So if you go to this paper and you look at their table and you look at their table for how they are selecting.

110
00:16:47.460 --> 00:17:04.710
Julie Hogan: The the leptons it will probably take me too long to find it right now, if you go to this paper and you look at what they're doing in this channel they require that the nuance have the leading new on has a momentum of at least 20 gv instead of 17.

111
00:17:06.120 --> 00:17:10.620
Julie Hogan: And that the hydraulic TAO has momentum of at least 30 GB instead of 20.

112
00:17:11.730 --> 00:17:23.010
Julie Hogan: So they've added a buffer here so that we have no we're trying to avoid the uncertainty of how we select or don't select.

113
00:17:24.720 --> 00:17:26.910
Julie Hogan: leptons kind of in that turn on region.

114
00:17:30.780 --> 00:17:43.590
Julie Hogan: Any questions about this part this is this is some one of the things that's always difficult, especially, you know as students and cms is figuring out how do we decode what these things, what these trigger freezes mean.

115
00:17:47.220 --> 00:17:49.710
yeah so maybe just two quick question.

116
00:17:51.030 --> 00:18:02.730
Dillon Fitzgerald: We probably also want crump leptons but I don't know if that's something we apply at this step, or when we're correlating objects together, or we could just use like the impact grander.

117
00:18:04.380 --> 00:18:08.820
Julie Hogan: yeah so that's part of that's part of the next box here okay.

118
00:18:11.340 --> 00:18:18.330
Julie Hogan: here's a question in the chat, why do we select different absolute Ada from you on and for tau that's a great question.

119
00:18:20.370 --> 00:18:26.790
Julie Hogan: It sometimes these decisions are driven by the detector itself so, for instance.

120
00:18:28.110 --> 00:18:40.770
Julie Hogan: jets have a different jets have a different pseudo rapidity acceptance region than nuance do because the new on detectors have a different pseudo rapidity cut off than the than the calorie amateurs do.

121
00:18:41.760 --> 00:18:47.880
Julie Hogan: So there are considerations like that jets will often be accepted at wider pseudo rapidity ease.

122
00:18:50.310 --> 00:18:59.400
Julie Hogan: In this case, the new on detectors we can we can actually select milan's out to 2.4 this trigger has tightened that in.

123
00:19:04.350 --> 00:19:10.440
Julie Hogan: If you if you were unhappy with that you could go you could go searching for some different trigger that had a wider acceptance.

124
00:19:12.000 --> 00:19:13.410
Edgar Fernando Carrera Jarrin: You might have so.

125
00:19:13.500 --> 00:19:16.320
Julie Hogan: You might have to trade off a higher momentum in order to get them.

126
00:19:18.660 --> 00:19:25.110
Edgar Fernando Carrera Jarrin: and also the detectors have different technologies which span different sub regions in Ada.

127
00:19:26.250 --> 00:19:29.550
Edgar Fernando Carrera Jarrin: And so those things are taking into account.

128
00:19:33.510 --> 00:19:51.960
Julie Hogan: Okay, so let's know think that's our kinematic criteria let's now think about the quality criteria, so we actually already had some some discussion about this, the suggestion was that are you on in particular should be isolated, it should be separated from the other TAO.

129
00:19:53.040 --> 00:20:00.090
Julie Hogan: Which is absolutely correct So how do we code those up into the identification and isolation criteria.

130
00:20:01.440 --> 00:20:05.850
Julie Hogan: Okay, we can think about them separately tell selection.

131
00:20:06.990 --> 00:20:13.800
Julie Hogan: So if we come back over here to new ons and towels and we go to the wiki page.

132
00:20:16.050 --> 00:20:18.600
Julie Hogan: We go to the wiki page that has the big list.

133
00:20:20.070 --> 00:20:21.450
Julie Hogan: Of all the ids.

134
00:20:22.950 --> 00:20:36.690
Julie Hogan: um there's there's this great selection that's it's it's a flag called by decay mode finding you will always want to use this okay super I trust you to our blog so we're always going to use that.

135
00:20:38.820 --> 00:20:48.870
Julie Hogan: So you'll find that we have, if you look at poet we've stored this this branch in our tree and let's just take a peek let's take a look, I have the.

136
00:20:49.920 --> 00:20:58.530
Julie Hogan: I have the top Cork pair file I don't have the higgs a higgs to TAO file, but we can look at the.

137
00:21:00.750 --> 00:21:01.950
Julie Hogan: We can look at the TAOs.

138
00:21:05.100 --> 00:21:11.160
Julie Hogan: know what i'm probably not going to remember the names of the branches off the top of my head so i'll grab the actual.

139
00:21:12.690 --> 00:21:13.380
Oh come on.

140
00:21:17.280 --> 00:21:17.730
Here we go.

141
00:21:18.930 --> 00:21:23.340
Julie Hogan: let's grab the actual file let's see how many of these towers actually pass that.

142
00:21:24.870 --> 00:21:28.740
Julie Hogan: Actually pass that requirement, I think we should have some towels in this sample.

143
00:21:31.710 --> 00:21:40.110
Julie Hogan: Here we go Okay, I do decay mode so actually we have a we have a we have a chunk of towels that we would just throw away so they don't pass this by decay mode finding.

144
00:21:42.210 --> 00:21:55.440
Julie Hogan: And then we have another requirement here again from our trigger so the trigger says loose ISO PF tell a loose ISO PF TAO suggest that we should be looking at.

145
00:21:56.250 --> 00:22:04.290
Julie Hogan: This tower ID that's based on isolation isolation again meaning how separated, is it from other things.

146
00:22:04.800 --> 00:22:20.640
Julie Hogan: And we should require that it that it is indeed separated from other things, and at least a loose level, so this is the kind of high efficiency, maybe, maybe lower purity but but but somewhat reasonable efficiency, a flag.

147
00:22:23.520 --> 00:22:34.620
Julie Hogan: So in fact what we're you know we talked about a buffer on momentum, you can also put a buffer on the identification from the trigger although that's a little bit less.

148
00:22:35.130 --> 00:22:48.960
Julie Hogan: little bit less important, so, so the the version of this analysis that you'll you could find on the open data portal will actually even require that the the tight flag is is fired even beyond just being loose.

149
00:22:51.930 --> 00:22:57.990
Julie Hogan: And so, then the other thing we want to know for sure, in this specific analysis, because we have.

150
00:22:58.710 --> 00:23:07.530
Julie Hogan: One hydroponic town one Munich towel, we want to know that this was a real TAO as opposed to being an electron or new on that looks like a towel.

151
00:23:08.460 --> 00:23:20.460
Julie Hogan: And so to get rid of electrons are new ones that look like towers, we have these other ideas called anti electron or electron rejection if you're looking at the tricky.

152
00:23:20.910 --> 00:23:30.360
Julie Hogan: And anti new on or muon rejection if we're looking at the wiki, and this is interesting, even to click here in the in the top quark sample.

153
00:23:32.520 --> 00:23:38.790
Julie Hogan: it's it's actually much more likely, you can see all the zeros here for electron rejection.

154
00:23:39.330 --> 00:23:52.710
Julie Hogan: If we compare that to the new ones, my take on this with that is that it's much easier for electrons to mimic a dramatic towel, which has some sense to it because they're a color and that are object, as opposed to a meal on.

155
00:23:55.620 --> 00:23:56.100
Okay.

156
00:23:58.320 --> 00:24:08.340
Julie Hogan: New on selection we don't have any information here in the trigger about what the ID should be for this new on, but we do see this phrase ISO.

157
00:24:09.240 --> 00:24:22.680
Julie Hogan: yeah we do see that it's isolated and so we should we should choose some criteria on the isolation which we have stored.

158
00:24:23.520 --> 00:24:35.640
Julie Hogan: In one of our branches Okay, so we have some we have some isolation variable here, and you can actually see, we have a dummy column, where we weren't able to calculate it for certain nuance.

159
00:24:37.230 --> 00:24:53.580
Julie Hogan: So how could we decide what criteria will use home how did this analysis decide, it should be less than 0.1 well again the first place, I would go back is to the new on wiki page.

160
00:24:54.930 --> 00:25:06.390
Julie Hogan: And I would look up isolation and they've given some recommendations so actually here, the recommendation is even that the tight cut is 0.1 to.

161
00:25:06.990 --> 00:25:15.990
Julie Hogan: This analysis use a slightly smaller slightly tighter requirement, perhaps because the the recommendations were changing when they were doing their work.

162
00:25:20.790 --> 00:25:33.690
Julie Hogan: Okay, so we can make a choice on isolation, based on what we learned from the trigger the ID we're not given any prompting so here's my rule of thumb for you in any analysis, where you expect a prompt.

163
00:25:34.950 --> 00:25:40.890
Julie Hogan: New on from your signal from your decay of your heavy particle tight is a really good way to go.

164
00:25:42.030 --> 00:25:43.320
Julie Hogan: tight is a super way to go.

165
00:25:45.960 --> 00:25:47.550
Julie Hogan: And that's what we'll do in this example.

166
00:25:49.380 --> 00:26:00.900
Julie Hogan: So Questions about this type of decision making How did we make these decisions, how would you replicate these decisions in your own analysis.

167
00:26:03.510 --> 00:26:05.280
Johan Andr?s Colorado Caicedo: I have a question in regards.

168
00:26:06.300 --> 00:26:06.690
Johan Andr?s Colorado Caicedo: to that.

169
00:26:07.890 --> 00:26:11.700
Johan Andr?s Colorado Caicedo: These cuts or these requirements are available.

170
00:26:13.410 --> 00:26:19.140
Johan Andr?s Colorado Caicedo: For data sets or real collisions or only the simulate that I said.

171
00:26:20.070 --> 00:26:25.890
Julie Hogan: None of these decisions rely on us knowing anything about the simulation truth.

172
00:26:26.790 --> 00:26:37.860
Julie Hogan: So anytime you find yourself tempted to make a requirement that that actually requires you to know to look up something about the generator status of any of your particles.

173
00:26:38.580 --> 00:26:48.900
Julie Hogan: You want to avoid that because you won't be able to make that same selection in data, so all of these things can be applied in exactly the same way on real data and simulated data.

174
00:26:49.800 --> 00:27:05.610
Johan Andr?s Colorado Caicedo: Okay, because I think in the in this possibility when when I have, for example, tell decay into one other, one may be a Python and neutrino okay.

175
00:27:06.870 --> 00:27:08.610
Johan Andr?s Colorado Caicedo: For this decay.

176
00:27:10.200 --> 00:27:12.870
Johan Andr?s Colorado Caicedo: In MC that is that.

177
00:27:14.100 --> 00:27:15.210
Johan Andr?s Colorado Caicedo: We can impose.

178
00:27:16.470 --> 00:27:34.290
Johan Andr?s Colorado Caicedo: For example, a one requirement that we respect to the MC mode, for example, the the one from or two prong or tree trunk, but in this case a I can see that so for the reason is my my question.

179
00:27:34.710 --> 00:27:39.450
Julie Hogan: yeah so, so I would avoid that type of way of thinking about selecting your events.

180
00:27:40.560 --> 00:27:52.050
Julie Hogan: You would want to recast the way you think about selecting hydroponic towels two things that can be two to two things that could be pulled out, using all of the discriminates that they have trained.

181
00:27:53.700 --> 00:28:10.620
Julie Hogan: Which don't tend to separate the different types of prongs hydroponically if that makes sense, so we have things based on isolation and we have ways to tell against electrons and nuance, but I think it's not common to try to split.

182
00:28:11.940 --> 00:28:17.460
Julie Hogan: detector based how analysis into different number of hydroponic prongs.

183
00:28:18.660 --> 00:28:19.590
Johan Andr?s Colorado Caicedo: Okay, thanks.

184
00:28:20.520 --> 00:28:21.630
Julie Hogan: that's a really good question.

185
00:28:32.130 --> 00:28:32.580
Julie Hogan: Okay.

186
00:28:34.290 --> 00:28:41.220
Julie Hogan: So the last question i've asked her about signal, are there any correlations between your objects that you might exploit.

187
00:28:42.270 --> 00:28:50.820
Julie Hogan: And one got posited earlier was that we should expect that, when we combine this new on in this town, we get a higgs boson.

188
00:28:52.650 --> 00:29:01.890
Julie Hogan: And that's a great idea and it's hard in this case it's hard to do, because we have those neutrinos.

189
00:29:03.210 --> 00:29:13.650
Julie Hogan: So in this specific case, you know I would say absolutely let's just make a higgs mass window if we were doing leptons all charge leptons, but in this case.

190
00:29:14.070 --> 00:29:24.480
Julie Hogan: With the TAO decaying to neutrinos we have we've lost some information, because we have only the visible part of the decay.

191
00:29:24.990 --> 00:29:41.970
Julie Hogan: available to us in a usable format, so we are probably not going to find a very clean higgs mass peak in this analysis, but are there any other correlations that anyone was able to think of that you would expect, between these objects.

192
00:30:12.450 --> 00:30:25.800
Julie Hogan: there's a kind of sneaky one hiding in there, that you actually might might not think of immediately, but we act, even though we have trouble accessing all of the mass we do.

193
00:30:26.460 --> 00:30:35.940
Julie Hogan: have the ability to measure the charge of both of these objects, we can measure them you on charge and the tau charge, and so we can ask that they have opposite charges.

194
00:30:36.960 --> 00:30:49.110
Julie Hogan: So that's one thing that will help us reduce some background and target our signal, because we can at least expect that the the two objects should make a neutral parent particles.

195
00:30:50.310 --> 00:30:54.390
Julie Hogan: So that's that's one correlation between now objects that we can absolutely exploit.

196
00:30:56.370 --> 00:30:56.760
Okay.

197
00:30:58.560 --> 00:31:07.560
Julie Hogan: All right, what time is it, it is 945 and we're going oh so we're like out of time okay.

198
00:31:08.670 --> 00:31:24.270
Julie Hogan: So let's keep going, then I apologize, I really wanted to take the time to talk through someone out someone's individual analysis, but maybe we'll we'll do that offline if anyone wants to talk through their analysis selection, we can we can do that together.

199
00:31:26.340 --> 00:31:27.570
Julie Hogan: So backgrounds.

200
00:31:28.650 --> 00:31:37.410
Julie Hogan: What will mimic our signal is the question, so you can obviously start with like just considered the entire standard model and try to figure out.

201
00:31:37.890 --> 00:31:48.300
Julie Hogan: What will what will sneak in so good rule of thumb is to say what would give me the same final state, maybe with a few extra things, or maybe missing a few things.

202
00:31:49.320 --> 00:31:54.240
Julie Hogan: Because it's always possible, that you get extras for free from Q CD.

203
00:31:56.160 --> 00:32:16.890
Julie Hogan: And it's always possible, that you just miss things in your selection so Cecilia has said Z to tau tau tau tau 100% is the biggest background for this process because it's identical, the only thing that's different is the fact that we've got this you know, probably some different production.

204
00:32:18.540 --> 00:32:27.630
Julie Hogan: going on, but the Z will have different properties, so what I wanted to do in this solution is link you to the open data portal page.

205
00:32:28.140 --> 00:32:39.990
Julie Hogan: For this analysis that you can look at more later, and they have listed out what the background processes are and Bingo the most dominant one is the to tau tau.

206
00:32:42.000 --> 00:32:50.130
Julie Hogan: Z to leptons will probably also sneak in there, because we could by mistake called some of these leptons of tau.

207
00:32:50.760 --> 00:33:00.990
Julie Hogan: w plus jets will probably sneak in there, it certainly can produce me once i'm one of the one of the jets could be mistaken as a towel all of those type of things.

208
00:33:01.590 --> 00:33:11.190
Julie Hogan: So we've got the big hitters here we've got the big hitters here Z w top court pairs and, of course, just multijet production multiple jets.

209
00:33:11.550 --> 00:33:26.100
Julie Hogan: By accident, we get a new on and by accident, we get something that looks like a cow, but this process has such a humongous cross section in proton collisions that all of those tiny rare accidents, will certainly show up.

210
00:33:29.460 --> 00:33:29.850
Julie Hogan: Okay.

211
00:33:30.870 --> 00:33:47.730
Julie Hogan: So, based on these processes which simulations should we include we talked about this a little bit in maths in maths challenge so similar logic once you've decided on a standard model process to include you can go finding the the simulated data sets for it.

212
00:33:48.750 --> 00:33:51.630
Julie Hogan: The one holdout is this Q CD.

213
00:33:52.740 --> 00:33:57.090
Julie Hogan: So when we say QC we just mean jets jets showing up.

214
00:33:59.190 --> 00:34:11.970
Julie Hogan: jets showing up randomly and that one is difficult to model is simulation so in cms physics analyses we rarely rely on the simulation.

215
00:34:12.510 --> 00:34:25.050
Julie Hogan: To model that background it is much preferable to try to find some way to use the data itself to model that type of background, which we'll talk about more in later lessons.

216
00:34:26.700 --> 00:34:39.240
Julie Hogan: Okay, so should we apply any upper or lower limits on the number of certain physics objects so that's one way, you can try to get a handle on your backgrounds, is to actually think out is it worth it for me to say.

217
00:34:39.930 --> 00:34:48.000
Julie Hogan: i'm not going to allow there to be five there must only be two or not going to allow there to be three there must only be one.

218
00:34:49.860 --> 00:34:59.280
Julie Hogan: That that's hard to give a rule of thumb but you should certainly study your signal samples, and your background samples to determine if that would help you.

219
00:35:01.200 --> 00:35:13.230
Julie Hogan: In this, it can be helpful, but you would it would be analysis based in this case when we do higgs to tau tau with one hydroponic and one you want to TAO.

220
00:35:13.740 --> 00:35:29.130
Julie Hogan: we're not going to specifically veto which is the jargon word veto we're not going to veto on additional leptons we're just going to look for events that have at least one good new on at least one of these good towels and work on combining them.

221
00:35:32.400 --> 00:35:35.700
Julie Hogan: Okay, are there any objects, we should veto.

222
00:35:38.340 --> 00:35:46.470
Julie Hogan: So, based on this list of backgrounds we've got a couple more minutes left, I think we have enough time for this question so based on our list of backgrounds here.

223
00:35:47.520 --> 00:35:55.530
Julie Hogan: Is there anything that that stands out as something we might be able to get rid of by specifically rejecting certain objects.

224
00:35:56.490 --> 00:36:02.250
Dillon Fitzgerald: yeah constructs invariant mess of a Z well me to church shirts.

225
00:36:03.900 --> 00:36:18.570
Julie Hogan: Anything that reconstructs invariant massive the Z, that is, that is certainly something we could explore one question you might want to answer first is look at your signal and see how much of your signal overlaps down into that mass range.

226
00:36:21.210 --> 00:36:22.290
Franz Glessgen: of too many jets.

227
00:36:23.430 --> 00:36:27.960
Julie Hogan: And events with too many jets yep we if we have some of that with 10 jets.

228
00:36:28.140 --> 00:36:31.950
Julie Hogan: it's probably not higgs the gel we could do we could do a cut.

229
00:36:34.290 --> 00:36:39.450
Julie Hogan: towels with the same charge yep so we did we did talk about that will use opposite charge requirement.

230
00:36:40.350 --> 00:36:51.840
Julie Hogan: The one that I pulled out, I think you guys are, I think you guys are all correct because there's not a there's not a strict right or wrong answer here, I think you guys are all correct one that stands out is this Tiki bar.

231
00:36:53.490 --> 00:36:56.610
Julie Hogan: which has mostly likely.

232
00:36:57.900 --> 00:37:03.900
Julie Hogan: Most likely has be tagged jets in it so jets that look as though they come from bottom corks.

233
00:37:04.890 --> 00:37:20.790
Julie Hogan: Is is a good handle we could have to get rid of those top work background events and that's what the actual cms paper did they rejected events with with top Cork, excuse me, they rejected events with jets that look like bottom works.

234
00:37:23.430 --> 00:37:39.180
Julie Hogan: OK, and then actually in the published analysis, the categories that we're doing the BF used it and use the JET Vito as was just suggested they look specifically in the Center region of the detector and said there should not be many jets there.

235
00:37:42.150 --> 00:37:47.700
Julie Hogan: Other than other than anything that might be overlapping the leptons or the town.

236
00:37:50.790 --> 00:37:59.910
Julie Hogan: Okay, and then correlations between objects in the background that might be useful for rejection so, given the time i'm just going to click this open.

237
00:38:01.050 --> 00:38:06.240
Julie Hogan: there's one other background that we can get a good handle on and it's w.

238
00:38:07.590 --> 00:38:19.200
Julie Hogan: When we have this Tower decay we're getting to neutrinos out and so that kind of anytime you have multiple disappearing objects, the missing et kind of gets scrambled and it and it.

239
00:38:20.070 --> 00:38:29.970
Julie Hogan: It doesn't represent just one object so it's difficult to use it straight away to analyze any specific let's say mass.

240
00:38:31.620 --> 00:38:43.650
Julie Hogan: In the case of the w it's a little bit different because we expect that we have our laptop let's say in this case the Milan and the missing energy should form that w.

241
00:38:44.820 --> 00:38:57.690
Julie Hogan: which would not be the case in a higgs event, so we can actually form what we call the transverse mass and so you can look that up in the paper, the transverse mass between the Milan in the missing et.

242
00:38:58.920 --> 00:39:08.040
Julie Hogan: tends to give you something that's quite near the w bows on mass, even though we're missing the Z information, the longitudinal information from the neutrino.

243
00:39:08.520 --> 00:39:30.600
Julie Hogan: It tends to be quite close and so we could we could actually say let's let's make sure we're down low let's make sure that the transverse mass here between our new on and our neutrinos looks looks very small, because it should not, it should not give us anything that's w like.

244
00:39:32.460 --> 00:39:41.520
Julie Hogan: So that's the last analysis, requirement that was put on this channel by the paper we have the individual selections from yuan's for TAOs.

245
00:39:42.630 --> 00:40:03.720
Julie Hogan: require them to be the same charge require that this transverse mass of the new on and missing et is small, to get rid of w background reject be tagged jets to get rid of top court background, and in the case of vector bows on fusion reject extra jets in the central region of the detector.

246
00:40:04.800 --> 00:40:06.390
Julie Hogan: And that's how they put things together.

247
00:40:07.980 --> 00:40:16.890
Julie Hogan: So the summary here is to pay attention to your triggers when you choose a trigger you usually have imposed on yourself some kinematic requirements.

248
00:40:18.360 --> 00:40:27.330
Julie Hogan: pay attention to which final state objects should be prompt impact parameter is usually wrapped into those identification criteria.

249
00:40:29.370 --> 00:40:38.370
Julie Hogan: A typical prompt final state particle would have some significant momentum be tight qualities and it may or may not be isolated based on your physics process.

250
00:40:39.360 --> 00:40:47.550
Julie Hogan: If you're going to veto on objects to to get rid of them you tend to want that process that condition.

251
00:40:48.060 --> 00:40:55.440
Julie Hogan: To get rid of background usually want it to be a very high efficiency criteria, so will often veto on things with a loose.

252
00:40:56.070 --> 00:41:05.550
Julie Hogan: Identification criteria, and we can always look for correlations between objects to select or reject based on whatever we want to do.

253
00:41:06.000 --> 00:41:16.950
Julie Hogan: And then, all these for all these conditions, you can study your signal you can study your backgrounds to try to balance that trade off of keeping your signal events and discarding your background events.

254
00:41:19.710 --> 00:41:20.310
Julie Hogan: Okay.

255
00:41:22.140 --> 00:41:30.630
Julie Hogan: Any last questions, based on this decision making procedure before we move on to the next lesson.

256
00:41:55.650 --> 00:41:56.400
Julie Hogan: Okay.

257
00:41:57.510 --> 00:41:58.680
Edgar Fernando Carrera Jarrin: I don't hear Oh, let.

258
00:41:59.340 --> 00:42:00.960
Edgar Fernando Carrera Jarrin: me just stop the recording.

259
00:42:01.260 --> 00:42:04.170
Edgar Fernando Carrera Jarrin: Before we begin the next session.