WEBVTT 1 00:00:00.539 --> 00:00:02.669 Julie Hogan: Okay hands on activity. 2 00:00:04.589 --> 00:00:07.140 Julie Hogan: So we're back in the let's just. 3 00:00:07.379 --> 00:00:10.889 Julie Hogan: let's just find the schedule we're here now, this will be our last. 4 00:00:11.820 --> 00:00:17.190 Julie Hogan: synchronous session together today and then there's three boxes below this. 5 00:00:18.570 --> 00:00:26.130 Julie Hogan: That are for that or for offline work before we can meet again tomorrow sooner this afternoon or tomorrow morning, depending on your time zone. 6 00:00:27.930 --> 00:00:35.070 Julie Hogan: The The challenge for this lesson and some some reading ahead for the blue colored lessons are coming tomorrow. 7 00:00:36.330 --> 00:00:40.470 Julie Hogan: So we've got a couple things to play with right now. 8 00:00:41.490 --> 00:00:47.670 Julie Hogan: So similar similar to what we did last time this is a little bit of a choose your exercise. 9 00:00:48.420 --> 00:01:01.170 Julie Hogan: We would like you to try to do exercise one everyone, but then there are two exercises related to the be tagging so option A and option B, and you can feel free to choose whichever one works for interesting to you. 10 00:01:02.640 --> 00:01:16.080 Julie Hogan: So, so this is a great time for any questions from the previous session great time to try out some of these some of these things in the code, since we talked about them more conceptually just before. 11 00:01:16.620 --> 00:01:25.320 Julie Hogan: And and let's see how it goes i'll kind of poke around on a couple of them myself and then we'll go over the solutions here at the very end. 12 00:02:14.100 --> 00:02:25.770 Julie Hogan: Hello Julie, can you hear me yes sorry my headphones were not working for some time, so I could not hear what you were saying, can you please repeat. 13 00:02:27.480 --> 00:02:34.680 Julie Hogan: yeah sure so so right now what we're working on is the hands on exercises for the advanced object lesson. 14 00:02:36.000 --> 00:02:42.060 Julie Hogan: So we're in if you're on on this page with the schedule we're in here. 15 00:02:44.460 --> 00:02:53.100 Julie Hogan: And we'd like everyone to try to do exercise one, and then you can choose option A or option B. 16 00:02:54.870 --> 00:02:56.010 Julie Hogan: For exercise to. 17 00:03:03.720 --> 00:03:10.530 Julie Hogan: Now I didn't grab the new on correction tools so i'm going to start there and run not or which several of you did lower ninos. 18 00:19:58.500 --> 00:20:04.170 Julie Hogan: Alright, everybody about 20 minutes left in this session so take at least seven eight more and then we'll go through. 19 00:20:05.220 --> 00:20:06.690 We go through some of the solutions. 20 00:20:09.990 --> 00:20:17.670 Julie Hogan: Certainly if you're having any trouble actually running the tools, please let us know or, if you have questions about the material. 21 00:26:57.540 --> 00:27:03.780 Julie Hogan: Okay, hopefully, you guys had some success or running running these making a couple of edits. 22 00:27:05.190 --> 00:27:09.690 Julie Hogan: Counting see things looking at some things let's start with exercise one. 23 00:27:10.890 --> 00:27:13.680 Julie Hogan: So results of the new on corrections. 24 00:27:14.910 --> 00:27:32.100 Julie Hogan: check that the corrections when apply but plotting we can basically plot corrected versus on corrected this is just in the money Carla because we don't have time to run the data file, so we don't have a full comparison here, but we can see what the correction is doing. 25 00:27:33.420 --> 00:27:37.380 Julie Hogan: Just by plotting the two different masses. 26 00:27:38.730 --> 00:27:43.380 Julie Hogan: So if you happen to have been watching my screen, as I was doing this. 27 00:27:45.600 --> 00:28:02.100 Julie Hogan: slimming the X axis here is actually really helpful for finding the Z because we had a bunch of events that ended up down at zero, and I think there was even one event that ended up with a very, very large number, where we just We certainly didn't have a Z like dime you on mass. 28 00:28:03.120 --> 00:28:23.280 Julie Hogan: for whatever reason, but when we use the route little tools here which actually I have a reasonable example of in the in the window right here whenever i'm drawing a certain branch I can apply some cut by giving in a different argument work my picture. 29 00:28:25.530 --> 00:28:31.440 Julie Hogan: We can see the zippy quite nicely here in the black line is the this raw value. 30 00:28:32.460 --> 00:28:44.310 Julie Hogan: And the red is what happens after we apply these corrections, so we can see that that definitely some of the the mass values have shifted. 31 00:28:44.970 --> 00:29:08.340 Julie Hogan: If I look right here in the peak we're a little bit narrower right here in the peak region, although of course there's some spreading off to the sides because i'm i'm still just cutting on the same exact events events with a wrong mass between 84 and 98, so this is a very subtle collection. 32 00:29:09.570 --> 00:29:17.310 Julie Hogan: home, you can kind of see what I was talking about that this is this is something that's of interest to the precision measures of things related to. 33 00:29:21.960 --> 00:29:24.150 Julie Hogan: cool let's see if I got the same graph is, and you know. 34 00:29:26.520 --> 00:29:30.090 Julie Hogan: looks really similar mm hmm yep awesome. 35 00:29:31.710 --> 00:29:43.440 Julie Hogan: great and so here are some of those plotting commands that you can reference for how to how to apply cuts and how to draw things on top of each other in the same panel, which is a useful feature. 36 00:29:48.000 --> 00:29:59.580 Julie Hogan: And here, actually, and you know, has provided us with the same plot for the actual data, so this is interesting, in contrast to the Monte Carlo where we're seeing most in kind of a. 37 00:30:01.320 --> 00:30:13.050 Julie Hogan: shifting of the let's say the width as opposed to where the peak lies in data, you can see that what we're doing is we're shifting a Z mass that's a little bit too low. 38 00:30:14.400 --> 00:30:25.860 Julie Hogan: In its peak position we're shifting it a little bit higher that's a really great comparison, so if you take the time to run the data file, hopefully, you will see the same thing. 39 00:30:32.700 --> 00:30:33.120 Julie Hogan: Okay. 40 00:30:34.140 --> 00:30:51.330 Julie Hogan: Be tagging had a couple of challenges here as well, so the first to add some alternate taggers So if you if you look back over the video you'll see me running through this I was just taking. 41 00:30:52.500 --> 00:30:59.310 Julie Hogan: This list that prints itself out, oh no it's I guess it's not influencing my other screen. 42 00:31:00.870 --> 00:31:12.330 Julie Hogan: There we go this list that prints itself out and I chose to add in the JET probability and the high purity track counting. 43 00:31:15.960 --> 00:31:31.500 Julie Hogan: And we need to do that in a couple of places, so this is an overview if you've never done it before if it's new for you, how do you add go through and add a branch, we always have to declare the variable. 44 00:31:32.820 --> 00:31:45.210 Julie Hogan: declare the branch and then somehow access the information and this solution is really missing one tiny elements that i'll show you in the file. 45 00:31:47.610 --> 00:31:53.580 Julie Hogan: i'm just going to search the phrase jet the tag and let's walk through everywhere, it appears. 46 00:31:54.630 --> 00:31:55.710 Julie Hogan: Declaration. 47 00:31:58.050 --> 00:31:59.010 Julie Hogan: branch. 48 00:32:01.920 --> 00:32:14.700 Julie Hogan: And then, for every event we declare out these factors, because the vector here is not saying here's the vector of all jets for all events it's the vector objects in this event. 49 00:32:15.750 --> 00:32:28.860 Julie Hogan: So that's if you ever observed that your file sizes blowing up on you one thing to check is to see if you have neglected declare some vector and you'll be end up storing many, many copies of the same thing. 50 00:32:31.170 --> 00:32:39.240 Julie Hogan: OK, and then here we go, we can grab these grab these different strings from the be discriminated function. 51 00:32:41.100 --> 00:32:48.570 Julie Hogan: So here in the solution, I plotted out what the track counting variable looks like it's completely different. 52 00:32:49.110 --> 00:33:06.840 Julie Hogan: than the completely different than the csv discriminate or which was a value from 01 this one's actually has a physics, meaning it's related to the impact parameters and I wanted to point out this reference so we've we've referenced here. 53 00:33:08.340 --> 00:33:23.220 Julie Hogan: and actual what we call physics analysis summary which is on algorithms for the JET identification and so you can find you can find all of the information about the various tigers references. 54 00:33:25.410 --> 00:33:30.660 Julie Hogan: What I was looking at here on my screen is the other one the probability. 55 00:33:32.610 --> 00:33:54.240 Julie Hogan: And I drawn i've drawn it in two halves kind of so what i've drawn here is the JET probability tags for jets that past the csv algorithm edits tightest working point so this one that's in the solid histogram should be real be tags according to the the other algorithm. 56 00:33:55.680 --> 00:34:06.600 Julie Hogan: And then the points here the ones that are in points in error bars, these are, this is the JET probability discriminate when the JET be tag. 57 00:34:07.380 --> 00:34:27.480 Julie Hogan: Which is the csv algorithm was extremely small so it's very light pork like so actually you can see, this is a good discriminate or as well we're light corks pile up near zero in this jet probability discriminate her and the Court jets have a have a higher value numerical data. 58 00:34:33.120 --> 00:34:48.990 Julie Hogan: And then the very last exercise was to count oh I you know I still have some fun, though, the very last exercise was to count the number of jets prevent that or be tagged according to this medium working point of the CSP algorithm. 59 00:34:50.310 --> 00:34:55.680 Julie Hogan: So I did that I think we can find that in here as well. 60 00:34:58.110 --> 00:35:04.290 Julie Hogan: Okay store branch called jet and CSP me, so I followed my own instructions, so that I know what's a search. 61 00:35:06.630 --> 00:35:10.110 Julie Hogan: And you can see, all the same pieces declaration. 62 00:35:12.450 --> 00:35:13.020 Julie Hogan: branch. 63 00:35:17.850 --> 00:35:24.330 Julie Hogan: initializing it in this case to zero, as opposed to clearing a vector and in all these cases. 64 00:35:25.980 --> 00:35:41.280 Julie Hogan: i'm following the template of what I see above and below me and i'm you know if if declaring route branches is new, for you, I was like i'm looking at the templates of all these other things, to give me a pointer as to how I should do that. 65 00:35:43.800 --> 00:36:00.570 Julie Hogan: Okay, and then we actually need to count, so if the combine secondary vertex rejects jet tags discriminate or is greater than 0.679 which I got from the lesson page the medium working point I will increment this counter. 66 00:36:06.450 --> 00:36:06.810 Okay. 67 00:36:08.280 --> 00:36:14.310 Julie Hogan: So let's take a look at that that should live here in my tree somewhere, here it is way up at the top. 68 00:36:15.360 --> 00:36:23.340 Julie Hogan: So in this in this top work sample I tend to have 567 jets got quite a few jets. 69 00:36:24.450 --> 00:36:27.450 Julie Hogan: And then, if I look at how many are be tagged. 70 00:36:28.680 --> 00:36:37.620 Julie Hogan: peek at one lots of them also at two so that makes perfect sense we expect there to be to bottom forks in our top court parents. 71 00:36:38.460 --> 00:36:54.210 Julie Hogan: We know we don't have a sweet, you know, this is a 65 or 70% efficiency to identify them, so a peek at one with a large component to make sense, I can turn this off of log scale as well. 72 00:36:55.440 --> 00:36:57.960 Julie Hogan: and see what it looks like in linear scale yeah. 73 00:36:59.700 --> 00:37:15.360 Julie Hogan: Very few with three even fewer would for a couple of zero So you can see, actually, you know i'm I have a significant amount of my top forks that just don't their their bottom corks don't look like on works, according to the algorithm and that's too bad, but it happens. 74 00:37:17.580 --> 00:37:30.060 Julie Hogan: Okay, so what I really want to look at our what happens when I apply the uncertainty so let's do this let's apply, excuse me, the uncertainty is the wrong word, the scale factor. 75 00:37:31.590 --> 00:37:34.140 Julie Hogan: This branch is called jet and CSP. 76 00:37:35.610 --> 00:37:39.630 Julie Hogan: let's put on the be tag wait. 77 00:37:40.770 --> 00:37:43.470 Julie Hogan: Oh, are these spelled wrong in the solutions. 78 00:37:44.760 --> 00:37:45.870 Julie Hogan: You bet they are. 79 00:37:47.190 --> 00:37:49.050 Julie Hogan: You bet they are there's an underscore. 80 00:37:51.090 --> 00:37:55.320 Julie Hogan: Okay, a question for Option a sorry if I missed that should we clear the vectors that we're filling. 81 00:37:55.770 --> 00:38:07.260 Julie Hogan: In the JET loop each event, yes, so you should you do not need to clear them in the Jet Blue you just it's sufficient to clear them at the beginning of the analyze function, because the analyze function is called for everyone. 82 00:38:08.700 --> 00:38:10.350 Julie Hogan: So that is what you can do. 83 00:38:12.600 --> 00:38:19.380 Julie Hogan: Okay, so I apologize to anyone who's copying and pasting because of the wrong spelling of these weights. 84 00:38:20.910 --> 00:38:26.130 Julie Hogan: So let's do this, let me go ahead and draw this one with. 85 00:38:27.180 --> 00:38:29.730 Julie Hogan: dots i'm going to draw dots. 86 00:38:31.680 --> 00:38:38.700 Julie Hogan: But I don't really want aero bars right now so i'm going to say P history and that we're getting the dots without airburst. 87 00:38:39.930 --> 00:38:44.850 Julie Hogan: So what we're going to see the solid line is the raw number of. 88 00:38:45.930 --> 00:38:49.860 Julie Hogan: The tech jets and now we're going to see what happens when we apply the scale factor. 89 00:38:51.600 --> 00:38:55.740 Julie Hogan: Oh, we get a super tiny dot okay not helpful let's make the finger. 90 00:38:59.670 --> 00:39:02.460 Julie Hogan: By using some of our gui tools. 91 00:39:04.110 --> 00:39:06.720 Julie Hogan: i'm going to make it bigger and i'm going to turn it back. 92 00:39:11.010 --> 00:39:12.480 Julie Hogan: Bigger there we go. 93 00:39:14.610 --> 00:39:21.780 Julie Hogan: Okay, so what is applying the scale factor do applying the scale factor has shifted the mean of this distribution down. 94 00:39:23.190 --> 00:39:35.760 Julie Hogan: That makes sense, we saw in the lesson that Monte Carlo over predicts the number of B corp jets, and so we expect applying the scale factor to reduce that number that's what we're seeing. 95 00:39:37.500 --> 00:39:44.460 Julie Hogan: What about uncertainties, what about uncertainties, we have other wait function so means my editor. 96 00:39:45.510 --> 00:39:47.640 Julie Hogan: We have wait wait up and wait down. 97 00:39:49.980 --> 00:39:54.570 Julie Hogan: So let's draw weight up and we're gonna have to watch roots teeny tiny dot. 98 00:39:55.620 --> 00:39:56.160 Julie Hogan: Here we go. 99 00:39:59.790 --> 00:40:01.530 Julie Hogan: Okay now somewhere is my editor. 100 00:40:12.060 --> 00:40:15.360 Julie Hogan: There it is okay let's make it blue. 101 00:40:17.700 --> 00:40:19.770 Julie Hogan: squares and they are. 102 00:40:22.080 --> 00:40:36.060 Julie Hogan: Okay, do we shift the uncertainty up, you can see, we actually get a different a different answer, and we would see the same thing if we shift the uncertainty down so i'll just go to the solution where i've already got this plot. 103 00:40:38.250 --> 00:40:45.270 Julie Hogan: With a nice legend here's how I made that legend I right clicked somewhere in the space and I chose build legend. 104 00:40:46.380 --> 00:40:51.570 Julie Hogan: And I accepted their default options and then you can like drag it around. 105 00:40:53.670 --> 00:41:07.260 Julie Hogan: Okay, so you can see we've got a shifting lower of the of the Nina the distribution and then we've got an uncertainty i'm in this case it's I think I did a quick calculator it's about 6%. 106 00:41:07.680 --> 00:41:24.060 Julie Hogan: Uncertainty on this distribution do this source so that's a look at some of the other some of the techniques related to these new on corrections and this be tagging and a little bit of practice at navigating a poet, and these boot files. 107 00:41:26.190 --> 00:41:31.200 Julie Hogan: So we will come, we will come back tomorrow. 108 00:41:32.880 --> 00:41:44.610 Julie Hogan: we'll go over the solutions to the challenge, just like we did at the beginning of this morning and then we'll have the analysis will us and the beginning of the cloud computing lesson we will see you guys tomorrow. 109 00:41:53.640 --> 00:41:54.330 Julie Hogan: Thank you bye. 110 00:41:55.890 --> 00:41:56.580 Dillon Fitzgerald: bye everyone.