WEBVTT 1 00:00:00.000 --> 00:00:04.259 Matthew Bellis: You are recording excellent. 2 00:00:06.359 --> 00:00:08.550 Julie Hogan: Okay welcome back everybody. 3 00:00:09.300 --> 00:00:22.470 Julie Hogan: So we're almost done with this kind of this this set of two lessons where we've blitz through all of the all of the cms physics objects to give you a preview of how they. 4 00:00:23.010 --> 00:00:31.830 Julie Hogan: You know how how they work in software, some of the physics concepts related to them and we've talked most recently yesterday about corrections. 5 00:00:32.850 --> 00:00:50.310 Julie Hogan: Which are which are very important in order to do to do actual research level analysis in cms, and so this this offline challenge which, I hope, several of you got to explore is really focused on those jet energy corrections. 6 00:00:51.450 --> 00:01:07.530 Julie Hogan: Which are which we talked about a bit did I lose the other tab no I didn't just hiding behind zoom so we talked quite a bit about this idea of having to correct the jets first of all because. 7 00:01:08.460 --> 00:01:18.660 Julie Hogan: Just based on the detector technology if there's only one collision going on at a given time we don't quite grab all of the energy that should be in the JET. 8 00:01:19.110 --> 00:01:39.060 Julie Hogan: And if there are multiple collisions going on at any given time, which is, of course, always the case in real life, we have some over overzealous amounts of energy going into each one of our jets, and so we have to correct, both for that and for and for other momentum and angular dependent. 9 00:01:40.260 --> 00:01:59.640 Julie Hogan: dependent things so that we can reliably reconstruct jets with the same momentum, they were they were given by a generator, and this is a process that does affect both data and the money, Carlo because this is literally correcting for detector effects. 10 00:02:00.840 --> 00:02:03.600 Julie Hogan: As well as just data Monte Carlo differences. 11 00:02:05.010 --> 00:02:17.100 Julie Hogan: And so, in the end we didn't go, we didn't go deeply over the code application here, but in the end, what we have on a show you just this final final code box. 12 00:02:19.200 --> 00:02:26.310 Julie Hogan: In our analyzers both jet analyzers that we have available for you have a corrected jet momentum. 13 00:02:27.810 --> 00:02:35.640 Julie Hogan: which takes the momentum of the JET and corrects it for both of these effects, the JET energy scale. 14 00:02:37.140 --> 00:02:43.860 Julie Hogan: which we often use this algorithm JC for energy corrections and separately, the JET energy resolution. 15 00:02:45.450 --> 00:03:04.140 Julie Hogan: You can the the mental picture of energy scale should be you know shifting the mean value of that reconstructed divided by generated momentum ratio, so that they line up and the resolution should be thought of as narrowing or widening that distribution. 16 00:03:06.540 --> 00:03:16.260 Julie Hogan: So we end up with five important momentum values to carry around with us in our analysis, first of all, we have just the momentum itself. 17 00:03:17.460 --> 00:03:28.500 Julie Hogan: Secondly, we have a set of uncertainties up and down that represent the uncertainty on the JET energy correction with the JET energy scale correction. 18 00:03:30.510 --> 00:03:42.060 Julie Hogan: And when we apply these uncertainties up and down we leave the smearing, which is the JET energy resolution that's represented by this variable PT scale we leave that guy alone. 19 00:03:43.560 --> 00:03:58.860 Julie Hogan: And then we apply that uncertainty separately, so in these last two lines, we have, we have the same old corrected general G scale corrected momentum that we used in the first line. 20 00:04:00.480 --> 00:04:07.350 Julie Hogan: And we've changed how we are smearing it to correct the JET energy resolution. 21 00:04:09.300 --> 00:04:24.000 Julie Hogan: So we end up with with these five variables, the regular momentum, and then we have two independent sets of uncertainties and that's what we were digging into in this challenge was just to kind of compare them draw some plots explore them and understand their behavior. 22 00:04:25.620 --> 00:04:26.460 Julie Hogan: So hopefully. 23 00:04:27.720 --> 00:04:31.350 Julie Hogan: Hopefully you are mostly able to see the. 24 00:04:33.360 --> 00:04:52.380 Julie Hogan: Excuse me to run to run poet over more events which is, which is nice to do we've been doing these test runs with about 200 events, but obviously there are many more than that in one file and later today even will talk about going whole hog and running multiple files. 25 00:04:53.430 --> 00:04:58.800 Julie Hogan: But we can get nice plots with the statistics just in in one group file. 26 00:04:59.820 --> 00:05:01.560 Julie Hogan: And so, let me go forward. 27 00:05:03.000 --> 00:05:05.550 Julie Hogan: We go forward to the solutions page. 28 00:05:11.400 --> 00:05:15.690 Julie Hogan: So we can take a look at what we've gotten. 29 00:05:17.310 --> 00:05:27.660 Julie Hogan: So, since i'm since i'm down to one screen i've moved locations, if my video looks different than it did before I only have one screen so i'm not going to be able to see the chat very well. 30 00:05:28.560 --> 00:05:35.880 Julie Hogan: So i'll give you your your first moment in the workshop break and not asked make you answer things out loud. 31 00:05:38.460 --> 00:05:44.760 Julie Hogan: So the first question was is the difference between the raw and corrected momentum larger or smaller than the uncertainty. 32 00:05:45.270 --> 00:06:00.750 Julie Hogan: This question is kind of digging out, you know how significant is this correction is this a small correction just a large correction, what do we, what do we expect, what are we working with here and then which uncertainty dominates So if I draw. 33 00:06:02.340 --> 00:06:14.310 Julie Hogan: corrected versus uncorrected and then I look individually at the JET energy scale uncertainties and the energy resolution uncertainties, we can learn about their respective magnitudes. 34 00:06:16.530 --> 00:06:30.480 Julie Hogan: And so, these are the plotting commands effectively that I used here to draw the following clocks and with some variations, for which branches on drawing each time, but the format kind of stays the same. 35 00:06:34.050 --> 00:06:41.580 Julie Hogan: When you're drawing on the command line we often need to go into the folder you've noticed that your route files have many folders. 36 00:06:42.300 --> 00:06:48.540 Julie Hogan: inside and you can see the and around those folders, just as you would on the normal unix shell. 37 00:06:49.320 --> 00:07:03.750 Julie Hogan: And then we can take the name of the tree itself, and we can draw a variety of things all on top of each other, and I use that gui tool to just kind of quickly change the colors so that it's easy for me to keep track of what i'm looking at. 38 00:07:05.100 --> 00:07:08.850 Julie Hogan: So here's the plot I got, which is quite big let's see if we can. 39 00:07:10.170 --> 00:07:11.940 Julie Hogan: go down a little bit that'll be good. 40 00:07:14.550 --> 00:07:33.390 Julie Hogan: So this is the plot that I got to compare in blue and and read the raw momentum with the corrected momentum and then in the black dotted lines of one form or another, we have the uncertainty, due to the JET energy scale. 41 00:07:34.740 --> 00:07:56.940 Julie Hogan: So you can see right away a couple of differences, the red and the black lines are quite tightly grouped so this uncertainty in general and we're looking at basically all the jets is not an overwhelmingly large uncertainty in this analysis and it's certainly. 42 00:07:58.200 --> 00:08:09.240 Julie Hogan: It certainly indicates that the red line is very significantly different from the blue line, so we are making a true correction, especially down here at low momentum. 43 00:08:10.140 --> 00:08:24.540 Julie Hogan: which is what we would kind of expect if we come back up here to these to these plots based on the way this first plot looks in the low momentum region that is where we should certainly expect the largest deviation. 44 00:08:26.640 --> 00:08:36.630 Julie Hogan: You can tell also that i've manual that we've applied PT thresholds, so the fact that this red and black lines is is so sharp. 45 00:08:37.590 --> 00:08:55.350 Julie Hogan: indicates that there's probably some sort of cuts going on in our code, as opposed to that being the physics, of the situation and that's true we've applied a cut at 15 GB so the raw momentum doesn't doesn't abide by that cut, but the corrected momentum does. 46 00:08:56.610 --> 00:09:01.890 Julie Hogan: And that's really something that we analysts have to think about when we're doing an analysis. 47 00:09:02.700 --> 00:09:12.030 Julie Hogan: You know if i'm going to decide the my jet my jet momentum threshold is 30 when i'm analyzing with a shifted up or a shifted down. 48 00:09:12.990 --> 00:09:28.170 Julie Hogan: scenario I really should do things differently than i've drawn here, I really should evaluate you know all these projects that live a slightly above or below my my threshold I should consider what the uncertainty does to the momentum of those jets when i'm selecting. 49 00:09:30.780 --> 00:09:40.260 Julie Hogan: So i'm mental picture of this of the magnitude of this uncertainty it's small, but we can certainly see the the dotted black lines. 50 00:09:41.520 --> 00:09:43.500 Julie Hogan: When I come down to this. 51 00:09:45.030 --> 00:09:58.350 Julie Hogan: Corresponding plot now we're looking at the uncertainty on the JET energy resolution in the dawn of black lines, the smearing up and down and and here since we're so zoomed out. 52 00:09:59.010 --> 00:10:14.340 Julie Hogan: And we're looking at such an inclusive picture of all the jets you, you have to get to the low statistics region way over here at higher momentum before you even really start to see that the red and the black lines are not exactly on top of each other. 53 00:10:15.630 --> 00:10:20.490 Julie Hogan: partially because we're in log scale, so this uncertainty is certainly smaller. 54 00:10:21.630 --> 00:10:30.450 Julie Hogan: than the JET energy scale corrections and indeed some analyses on some analyses just go forward with the JET energy scale corrections and they don't. 55 00:10:30.930 --> 00:10:43.860 Julie Hogan: They don't really need to apply this jet energy resolution because it's just too small, of an effect to really matter for their analysis so that's something that can be studied individually, case by case. 56 00:10:48.420 --> 00:10:48.840 Julie Hogan: Okay. 57 00:10:50.910 --> 00:11:10.470 Julie Hogan: So, then, the the other interesting thing to do is to investigate how things change when we go to different regions of space and so just just so that we can anticipate the answer will get let's come back over here to the lesson page and take a look at the plots. 58 00:11:12.600 --> 00:11:28.560 Julie Hogan: So we have a couple ways on this lesson page to to see the uncertainties that will be working with here in the JET energy resolution section, we have this plot, which comes from the reference paper, which is linked at the top of the page. 59 00:11:29.970 --> 00:11:38.430 Julie Hogan: And you can see right away the scale factor values, so this is a data over Monte Carlo ratio this correction is a basic. 60 00:11:39.000 --> 00:11:55.350 Julie Hogan: scale factor that's applied only on simulation to correct differences between data and simulation and you can tell that usually in most regions of stupidity, so this is a angle in the detector going starting at the Center and going out for the edges. 61 00:11:56.790 --> 00:12:02.520 Julie Hogan: In most regions, the scale factors between about 1.1 and 1.5. 62 00:12:03.750 --> 00:12:14.190 Julie Hogan: And so that what this means is that, in general, in data if I plot, I wonder, you know what we have time let's plays zoom. 63 00:12:16.380 --> 00:12:21.930 Julie Hogan: let's play zoom because I have a lot of white space, right here to the side of my picture. 64 00:12:23.160 --> 00:12:23.700 Julie Hogan: Here we go. 65 00:12:26.490 --> 00:12:26.760 Okay. 66 00:12:27.960 --> 00:12:41.580 Julie Hogan: So let's play zoom and let's have this plot be the momentum reconstructed okay reconstructed divided by the momentum generated for a certain jet. 67 00:12:43.410 --> 00:12:55.020 Julie Hogan: And that's what we have called the response in in some of the other plots on this page, and so what you can imagine, is that one is the ideal value. 68 00:12:56.280 --> 00:13:04.830 Julie Hogan: But we are not always going to you know it's not going to be a spike at one Okay, we have, we have a gaussian ish distributed. 69 00:13:06.300 --> 00:13:08.970 Julie Hogan: range of these response values. 70 00:13:09.990 --> 00:13:15.780 Julie Hogan: And so, a scale factor that's larger than one indicates that you know if this is the data. 71 00:13:18.750 --> 00:13:20.760 Julie Hogan: Then the simulation. 72 00:13:21.870 --> 00:13:23.010 Julie Hogan: tends to be. 73 00:13:24.150 --> 00:13:24.780 Julie Hogan: narrower. 74 00:13:26.130 --> 00:13:37.830 Julie Hogan: Okay, the simulation tends to be narrower, this is this is somewhat sensible right, I mean Okay, the the simulated detector is a little more optimistic than the real detector. 75 00:13:38.640 --> 00:13:54.420 Julie Hogan: that's that's an okay space to live in, we can always make things worse it's actually quite it would be difficult to correct things to make them to make them better if our simulated detect are somehow worse reconstructing jets and then the real detector. 76 00:13:56.580 --> 00:14:04.830 Julie Hogan: Okay, so that's what a scale factor larger than one indicates, and we can see the uncertainties here just shown in the shaded regions on the plot. 77 00:14:06.030 --> 00:14:12.690 Julie Hogan: So my my question in the challenge was to repeat those plots for jets that live in this region. 78 00:14:14.580 --> 00:14:31.590 Julie Hogan: Which is going to be many, many fewer there's gonna be a lower statistics plot by far most jets will live in a more central region, and so you can expect right away that the uncertainty, we observed for this jet energy resolution will probably get larger. 79 00:14:33.450 --> 00:14:33.720 Okay. 80 00:14:39.270 --> 00:14:43.320 Julie Hogan: And that's the problem with playing zoom is that now, I have to clear everything. 81 00:14:44.730 --> 00:14:57.090 Julie Hogan: Because it doesn't follow it's not locked my browser okay down at the bottom of the lesson page, we have more information about the uncertainties. 82 00:14:57.840 --> 00:15:09.990 Julie Hogan: specifically for the JET energy scale for the journey corrections has an end here, they actually have you know their own individual separated out plots in the published paper. 83 00:15:11.160 --> 00:15:20.760 Julie Hogan: And so, these plots are showing you the different contributions to the uncertainty from the different levels of correction. 84 00:15:21.270 --> 00:15:37.680 Julie Hogan: So if I zip way back up to the top of the page, we were looking at this graphic that shows us different in sequentially applied corrections going across and they each carry some individual uncertainty which we combine together. 85 00:15:39.180 --> 00:15:40.800 Julie Hogan: to form the total uncertainty. 86 00:15:42.150 --> 00:15:51.780 Julie Hogan: So, most of the time, and this is true of what we've implemented in poets, most of the time analysts are working with the yellow uncertainty. 87 00:15:53.160 --> 00:16:02.940 Julie Hogan: Rarely do we need to explicitly apply the flavor based corrections or any time stability related corrections. 88 00:16:03.540 --> 00:16:18.120 Julie Hogan: Those would come into some people's specific analysis, depending on what you're studying most analyses and this is what we've implemented for you apply basically the red the black and the blue. 89 00:16:19.500 --> 00:16:22.980 Julie Hogan: Corrections, and so we grab the yellow uncertainty. 90 00:16:25.230 --> 00:16:33.870 Julie Hogan: And you can see how its distributed were more uncertain at very low momentum or more uncertain at very high stupidities. 91 00:16:35.160 --> 00:16:44.520 Julie Hogan: So I did the typical you know the typical cms search analysis, you know some new physics, usually massive whatever. 92 00:16:45.840 --> 00:16:52.470 Julie Hogan: The typical analysis of that sort, will consider jets starting from about. 93 00:16:53.610 --> 00:16:55.530 Julie Hogan: Starting from maybe about 30. 94 00:16:56.760 --> 00:16:59.340 Julie Hogan: ggv and moving upward. 95 00:17:01.380 --> 00:17:23.940 Julie Hogan: Some even choose 40 or 50 and most analyses, unless you unless your specific physics points you to look very forward toward the edges of the detector if your if your physics scenario doesn't instruct you to do that most analysts will stick to this region in here. 96 00:17:27.420 --> 00:17:40.020 Julie Hogan: And by doing that you know we do restrict the the acceptance, a little bit, obviously we may lose some signal events with objects with jets that are outside those ranges but we reduce the uncertainties significantly. 97 00:17:42.330 --> 00:17:53.520 Julie Hogan: Okay, so same story here when I asked you to plot in the region of greater than or equal to three in pseudo rapidity absolute value we expect the uncertainties to get a bit more. 98 00:17:54.870 --> 00:17:57.330 Julie Hogan: So let me clear again. 99 00:18:01.200 --> 00:18:01.410 Okay. 100 00:18:07.980 --> 00:18:12.870 Julie Hogan: So let's uncover let's uncover the high Ada. 101 00:18:17.730 --> 00:18:22.950 Julie Hogan: i've done exactly the same type of plotting command here but i've added. 102 00:18:24.150 --> 00:18:27.540 Julie Hogan: a cut in this second region. 103 00:18:28.890 --> 00:18:33.180 Julie Hogan: Excuse me second parameter of the draw command so i've added my cup there. 104 00:18:36.300 --> 00:18:47.190 Julie Hogan: And we can see a couple of things, this is the one this is the plot for jet energy scale, just like we saw before we're seeing that there's a bigger deviation. 105 00:18:47.850 --> 00:19:01.140 Julie Hogan: Between corrected and uncorrected between blue and red, especially in the low momentum region we've seen a larger discrepancy there, so the correction is doing more work and the uncertainty has gotten a bit larger. 106 00:19:04.740 --> 00:19:07.350 Julie Hogan: And then we actually have the same story. 107 00:19:08.640 --> 00:19:10.800 Julie Hogan: In the JET energy resolution. 108 00:19:11.940 --> 00:19:18.480 Julie Hogan: Larger discrepancy well it's exactly the same discrepancy between blue and red in this plot, as it is in the other cloth. 109 00:19:19.920 --> 00:19:26.190 Julie Hogan: minus the fact that my my y axis has gotten a little bit mucked up but. 110 00:19:27.990 --> 00:19:30.180 Julie Hogan: We have also much larger. 111 00:19:31.260 --> 00:19:43.410 Julie Hogan: difference between the red lines between the black the different types of black shaded lines, so I was flipping back and forth between these two tiny going how different are they. 112 00:19:44.640 --> 00:19:45.750 Julie Hogan: Different are they. 113 00:19:46.830 --> 00:19:47.790 Julie Hogan: they're. 114 00:19:48.960 --> 00:19:59.790 Julie Hogan: pretty much the same by in magnitude so we've we've significantly changed from this scenario where the JET energy resolution uncertainty was small. 115 00:20:00.450 --> 00:20:10.980 Julie Hogan: The JET energy correction uncertainty was not overwhelming but certainly larger, more dominant to a region where they're both quite large and approximately on par with each other. 116 00:20:12.480 --> 00:20:19.620 Julie Hogan: So it's so that's something that can influence your choices when you're when you're making your analysis choices, if you have a physics case. 117 00:20:19.980 --> 00:20:34.380 Julie Hogan: That puts you out toward the edges of the detector to look for jets you certainly are not in a situation where it would be where you want to neglect the second source of uncertainty and correction, but it's important to apply them both. 118 00:20:37.110 --> 00:20:45.060 Julie Hogan: Okay, so our key points to close out this to close out this part of the lesson. 119 00:20:46.170 --> 00:20:54.750 Julie Hogan: Is that, generally, these are very significant corrections I don't know of any physics analysis and cms that does not apply the energy corrections. 120 00:20:55.620 --> 00:21:13.770 Julie Hogan: There you know in fact built into our to our processing that's what we've done for you, with the the physics analysis tool kit type jets you're actually getting the same processing that we use for real and cms these days, which brings those corrections, along with it for free. 121 00:21:15.450 --> 00:21:32.970 Julie Hogan: So they typically are significant corrections they took to lower the momentum because typically we are dealing with scenarios, where we have multiple collisions going on, simultaneously, and we need to, and we need to reduce, on average, the energy project. 122 00:21:34.470 --> 00:21:48.540 Julie Hogan: For most jets the jets energy scale is the larger effect, but in the end CAP region of the detector both of them are very important and the JET energy resolution uncertainty becomes basically equivalent with the other. 123 00:21:51.000 --> 00:22:13.080 Julie Hogan: So that wraps up our advanced object challenge, and we have plenty of time, I think we have 14 minutes for any discussion about any of these physics object lessons for questions about this anything that you would like to talk about before we move into analysis flow and really walk through. 124 00:23:18.930 --> 00:23:29.970 Julie Hogan: Okay, well, it sounds like people are at least for the moment content with their knowledge so Edgar we're going to give you 10 extra minutes. 125 00:23:30.570 --> 00:23:38.100 Julie Hogan: I hope you guys will will bookmark these pages, I know that we have a lot of information on these lesson pages that we didn't literally all speak through. 126 00:23:38.850 --> 00:23:47.070 Julie Hogan: But, but we hope they're good resources for you and points you even to other resources from cms as your as you're getting started in your work. 127 00:23:49.290 --> 00:23:49.740 Well, thanks. 128 00:23:52.110 --> 00:23:56.640 Matthew Bellis: Thanks Julie i'm going to stop the recording now for for that section.