WEBVTT 1 00:00:02.010 --> 00:00:02.310 Louie Dartmoor Corpe: OK. 2 00:00:03.179 --> 00:00:04.410 Federico Leo Redi: OK OK, see the recording. 3 00:00:04.620 --> 00:00:05.609 Louie Dartmoor Corpe: We can yes. 4 00:00:05.790 --> 00:00:07.230 Federico Leo Redi: Yes, I will assign the course. 5 00:00:07.830 --> 00:00:11.940 Louie Dartmoor Corpe: Okay perfect so Okay, just to repeat, for the benefit of the recording. 6 00:00:12.719 --> 00:00:18.840 Louie Dartmoor Corpe: Welcome to the second day of the LP workshop, we will be talking now about new LP search results and predictions that below you see. 7 00:00:19.110 --> 00:00:29.760 Louie Dartmoor Corpe: We have three talks at this session before a coffee break and we'll start with Victoria talking about the Atlas displaced courage agenda houses, so we can see your slides we are you ready and we'll let you know. 8 00:00:31.410 --> 00:00:33.150 Victoria Sánchez Sebastián: Okay, thanks okay. 9 00:00:33.210 --> 00:00:39.810 Victoria Sánchez Sebastián: So yeah this is Victoria from licorice with him and we present that will run to a resource so in. 10 00:00:40.560 --> 00:00:53.220 Victoria Sánchez Sebastián: A desert a look for purse of noodle lonely scholars, that the K hedonic harmonically either in the clinical emitter or we also have some sensitivity, a with him in the outer edge of the electronic perimeter. 11 00:00:53.970 --> 00:00:56.970 Victoria Sánchez Sebastián: We use as benchmark even sector models, you can see other. 12 00:00:57.930 --> 00:01:05.100 Victoria Sánchez Sebastián: In these models that are hidden sector in this and their mother are connected, they are having a child motion which then the case to appear so long, the scholars. 13 00:01:05.610 --> 00:01:14.250 Victoria Sánchez Sebastián: All of these hidden sector particles are new than the standard model so they are only detected when they take they came back to the standard model through up your premiums. 14 00:01:15.090 --> 00:01:23.160 Victoria Sánchez Sebastián: And in this case we're considering lady case of this color a both a family's from this correctly or reconstructed a single digits. 15 00:01:23.700 --> 00:01:31.650 Victoria Sánchez Sebastián: And therefore, the search a uses a search looks for two days that we call courage to the dates that narrow. 16 00:01:32.160 --> 00:01:38.040 Victoria Sánchez Sebastián: Because the showers translate a checklist because these are neutral particles within spec he's sending a detector. 17 00:01:38.550 --> 00:01:54.240 Victoria Sánchez Sebastián: And then we require that the majority of any of the energy of these jets is in the header any color emitter in for these redefined the collaboration, which is this ratio of energy deposited in the hundred meter over energy in the ECON and we required that this is high for this. 18 00:01:55.980 --> 00:02:01.410 Victoria Sánchez Sebastián: So a the main backgrounds, that we have our standard moment he gets a, that is the main background. 19 00:02:01.860 --> 00:02:08.370 Victoria Sánchez Sebastián: A, though, this process has an ability to mimic signal a, but it has a very high cross section so it's aluminum background. 20 00:02:09.000 --> 00:02:18.540 Victoria Sánchez Sebastián: Then secondly we have been used background, which are millions produce a from the interactions of the process of the bunch a interacting with a bean pie, or even some focusing elements. 21 00:02:19.020 --> 00:02:29.160 Victoria Sánchez Sebastián: And when these will this monster parallel to the room and when they go through the detector in, for example, the gallery meter they might undergo brimstone in our reconstructed, as this place jets. 22 00:02:29.700 --> 00:02:38.670 Victoria Sánchez Sebastián: And finally, we have cost makes a which actually are the lowest rate and they have an individual contribution to the search, so you will see that we don't have an explicit selection to remove these. 23 00:02:39.510 --> 00:02:48.060 Victoria Sánchez Sebastián: Also, we we use machine learning to stay in the fi signal events in the background system at uc now qualified ABC the method so. 24 00:02:49.110 --> 00:03:00.000 Victoria Sánchez Sebastián: Here you can see an overview of the analysis as you see, we start with to a dedicated signature to their signature driven triggers which are the conditions triggers you have to have them. 25 00:03:00.630 --> 00:03:05.940 Victoria Sánchez Sebastián: A after selection we apply these machine learning to study mentioned first we start with a project neural net. 26 00:03:06.510 --> 00:03:10.560 Victoria Sánchez Sebastián: Either classify states the standard model multitude beaming use background or signal. 27 00:03:11.130 --> 00:03:18.900 Victoria Sánchez Sebastián: In the second step is availability that uses the output of these neural network and some even level variables to classify events as signal background. 28 00:03:19.620 --> 00:03:26.580 Victoria Sánchez Sebastián: I want to emphasize that for both the training of these machine learning tools and from the final selection criteria, a, we have to selections. 29 00:03:27.000 --> 00:03:35.490 Victoria Sánchez Sebastián: A, which are optimized for the for loyalty signals which you can see here the different ranges of masses a discovers and the hi fi signals. 30 00:03:36.300 --> 00:03:46.110 Victoria Sánchez Sebastián: And we're now we go through each of the steps of these analysis so first the calibration triggers these are a they will this is yes, I. 31 00:03:46.650 --> 00:03:56.190 Victoria Sánchez Sebastián: done to identify narrow chocolate chips, which have a high called ratio, it will have two types in different that different are different in the Level one item. 32 00:03:56.700 --> 00:04:03.630 Victoria Sánchez Sebastián: A first we have the height the trigger a for which we have when we first have the though 60 which applies or 60 db first one. 33 00:04:04.140 --> 00:04:12.210 Victoria Sánchez Sebastián: Is was the lowest temperature scale three there for 2015 2016 or so so active to impact of the 17 2018. 34 00:04:12.690 --> 00:04:19.440 Victoria Sánchez Sebastián: A but a for those two years the lowest and freescale was the top 100 So what we do is apply and all of these two triggers. 35 00:04:20.220 --> 00:04:28.380 Victoria Sánchez Sebastián: And the second today is the loyalty, which applies are lower and energy threshold, but requires that the jets this isolated from excel deposits. 36 00:04:29.130 --> 00:04:35.010 Victoria Sánchez Sebastián: And these trigger was introduced in 2016 in other time was only efficient for loyalty samples but. 37 00:04:35.400 --> 00:04:42.600 Victoria Sánchez Sebastián: After some modifications in 2018 he became very efficient for all signal samples in I, you can see this in the blood on the left. 38 00:04:43.170 --> 00:04:52.320 Victoria Sánchez Sebastián: If you compare the open and feel squares, you can see how the efficiency of the loyalty trigger has changed between these two versions, a for high the samples. 39 00:04:52.920 --> 00:05:02.280 Victoria Sánchez Sebastián: So, with all of these modifications in these version of the analysis, what we do is to apply a logical or of these triggers regardless of this election that we were targeting. 40 00:05:03.180 --> 00:05:09.060 Victoria Sánchez Sebastián: A on the right, you can see an esteem, or how they can ratio did a tragedy sorry coloration trigger looks like. 41 00:05:09.540 --> 00:05:16.050 Victoria Sánchez Sebastián: And as you can see, we have two versions are so based on the tempe we have the nominal three that that runs all of this election. 42 00:05:16.500 --> 00:05:24.870 Victoria Sánchez Sebastián: And then we have annoyed with a more inclusive to that they know I saw today that is used to collect that data sample that doesn't run a baby Boomer algorithm. 43 00:05:25.380 --> 00:05:37.380 Victoria Sánchez Sebastián: In we use a combination of the nominal and then I switched over to the to target the end, finally, just to mention that we have also another version of the trigger that France in empty ones crossings for collection of academics that acid. 44 00:05:39.180 --> 00:05:46.380 Victoria Sánchez Sebastián: Okay, then the second element, and this is quite a normal for this version is that this project neural network. 45 00:05:46.830 --> 00:05:53.610 Victoria Sánchez Sebastián: That is trained on a project basis in order to identify these places in signal from standard model like or be like it's. 46 00:05:54.450 --> 00:06:04.800 Victoria Sánchez Sebastián: The input by others that we use a low level features, including inner detector tracks karima the energy deposits and me on track segments that are matched to the debt, we are we're considering. 47 00:06:05.280 --> 00:06:12.450 Victoria Sánchez Sebastián: In the output, we have three new year networks course that predict if the if the data is produced by nlp a multi data on the. 48 00:06:12.990 --> 00:06:22.830 Victoria Sánchez Sebastián: On the right, you can see a signal score, for example, for the two things in a for the training of these neural net where we use three samples a first one in. 49 00:06:23.520 --> 00:06:31.050 Victoria Sánchez Sebastián: A signal we use Monte Carlo simulation of signaling we we use the dates associated with lonely particle that the case in the gallery meters. 50 00:06:31.650 --> 00:06:40.500 Victoria Sánchez Sebastián: Secondly, we have a Monte Carlo simulation of what the movie yet and the final sample is a beep a data sample connected with the trigger that I mentioned before. 51 00:06:41.130 --> 00:06:51.180 Victoria Sánchez Sebastián: In for these we we use the digits identified as the data people and a okay here, you can see how the neural network looks like. 52 00:06:51.750 --> 00:06:56.040 Victoria Sánchez Sebastián: And as you can see, it has two parts, first we have the mania network, which is a combo. 53 00:06:56.790 --> 00:07:08.190 Victoria Sánchez Sebastián: combination of one, the combination of layers in long short term memory neural networks, which then are concatenate it and we have a the output of screen a nearly four scores that we use later in the analysis. 54 00:07:08.910 --> 00:07:14.670 Victoria Sánchez Sebastián: as well, we found a only using the mania network at the beginning, is that they were de de Monte Carlo differences in the output. 55 00:07:15.150 --> 00:07:19.800 Victoria Sánchez Sebastián: Add to this program that we have so Monte Carlo some person, a data samples in the training. 56 00:07:20.310 --> 00:07:31.380 Victoria Sánchez Sebastián: In the recent was mainly that we have some modeling issues in some bibles a, for example, the cluster timing and, if you look on the left on the right top you can see that a we're taking our control region in the there was. 57 00:07:32.640 --> 00:07:45.600 Victoria Sánchez Sebastián: A mismatch So what we need to improve this will to fix this was to other analysts on your network during training and to the output of the mania network and was trained a juice neither did control region in. 58 00:07:46.230 --> 00:07:49.950 Victoria Sánchez Sebastián: The ideas that he tried to separate data Monte Carlo jets in these controlled region. 59 00:07:50.460 --> 00:08:03.630 Victoria Sánchez Sebastián: And during training the the performance of these network was fed back to the mania network sets that in the optimisation a documentation a included this information so that's that a there were no differences in the catalog. 60 00:08:04.440 --> 00:08:14.220 Victoria Sánchez Sebastián: In in this is every interview technique in adolescent you can see that at the end adversary indeed in this country, region, the the performance is much better looking the below. 61 00:08:16.080 --> 00:08:26.250 Victoria Sánchez Sebastián: The following element, a separate ability which uses as input a information from the jets identified by these neural network as mostly men like or more be like. 62 00:08:26.700 --> 00:08:36.840 Victoria Sánchez Sebastián: and also some evil level information, and this was during a with see a combination of signal samples against the data in the main objective was to discriminate from beep. 63 00:08:37.230 --> 00:08:45.840 Victoria Sánchez Sebastián: However, these big data sample contains the standard moment yet events as well, and while he's found is that this video series communicates a signal from audience background, as well as from. 64 00:08:46.470 --> 00:08:55.680 Victoria Sánchez Sebastián: An indicated here in the deadlines how a different regions of the vtt identify be like jets multijet like gets in in signal dates. 65 00:08:57.630 --> 00:09:05.970 Victoria Sánchez Sebastián: So, then, we moved to heaven cleaning up any selection a What we require is that we want to apply the recently method, so we wanted to have to create environments. 66 00:09:06.570 --> 00:09:17.340 Victoria Sánchez Sebastián: And for these we wanted to well basically having multiple backgrounds versus making these correlations So what we wanted to have is only one payback on population in the final selection, which was more digits. 67 00:09:17.880 --> 00:09:31.380 Victoria Sánchez Sebastián: In for these two they're matching cuts and dependability score and some further cleaning carpet applied rejected and other other spirits events and here listed the final selections but in backup you can see the full list of a few selections. 68 00:09:32.610 --> 00:09:39.810 Victoria Sánchez Sebastián: So, after this we have like a background estimation, as I mentioned, we we apply it, the likelihood based in within your CBC the method. 69 00:09:40.200 --> 00:09:46.920 Victoria Sánchez Sebastián: So as much as most of you know ABC method prefers to integrate these variables to design the plane into four regions. 70 00:09:47.460 --> 00:09:56.220 Victoria Sánchez Sebastián: Sets that in the back of background only hypotheses a background events in the signal region a can be estimated with a number of background events in regions B, C and D. 71 00:09:56.820 --> 00:10:09.210 Victoria Sánchez Sebastián: But we use a modified version that if it is similar time mostly background and signal models in this account for singer contamination some regions B, C and D, which is look, if you look at the plug on the right, we we have some of it. 72 00:10:10.230 --> 00:10:16.680 Victoria Sánchez Sebastián: And, in our case, we defined the vc be playing by the ABC ABC this a B, C D has course. 73 00:10:17.040 --> 00:10:30.000 Victoria Sánchez Sebastián: In also these variables that we always find very useful, which is there some of the entire between the signal the candidate and it's near stuck with the larger entity, we find a very small correlation in data, so this is useful, we can. 74 00:10:31.110 --> 00:10:40.890 Victoria Sánchez Sebastián: This is take that we have to do a but we we did some further tests to verify that beep in comics were not present in the vc the plane and will not spoil the method. 75 00:10:41.520 --> 00:10:51.810 Victoria Sánchez Sebastián: So for being a what we did was to to take a the a big data sample collected by these triggers and I mentioned in okay thanks. 76 00:10:52.620 --> 00:10:56.280 Victoria Sánchez Sebastián: And okay what we found is that the the height is election, there was no being regenerated. 77 00:10:57.120 --> 00:11:08.040 Victoria Sánchez Sebastián: In so that was fine and for the loyalty selection one event was found in region a bath a what we're saying is that all events in the vc be playing a look a standard more than multijet like. 78 00:11:08.490 --> 00:11:19.650 Victoria Sánchez Sebastián: And since we know we have these contributions in the in the data sample that was fine a four course makes what was done was to use the data is collected by these empty ones crossings coverage to get. 79 00:11:20.460 --> 00:11:38.970 Victoria Sánchez Sebastián: An estimate the contribution of ABC or, of course, mixing the ABC playing for the main selection and for these a we took a We calculated the factors to to this conversion and first one was the fact that a delay factor that takes into account that we have less a. 80 00:11:40.170 --> 00:11:40.710 Federico Leo Redi: Dr g. 81 00:11:42.150 --> 00:11:42.690 Which is very. 82 00:11:44.400 --> 00:11:44.790 Federico Leo Redi: Human. 83 00:11:45.270 --> 00:11:47.760 Victoria Sánchez Sebastián: Right and listening so for someone as big as well. 84 00:11:49.110 --> 00:11:58.620 Victoria Sánchez Sebastián: Okay, so we have less than Defense crossings then filaments crossings, and the second factor, it took into account that a empty amongst Christians are much cleaner environment coalition's. 85 00:11:59.160 --> 00:12:03.450 Victoria Sánchez Sebastián: And with all of these into taking all of these into account, we found a few events in. 86 00:12:04.080 --> 00:12:14.760 Victoria Sánchez Sebastián: In in in the events in the city playing a where we can do statistical uncertainty, so we could confirm that even cosmetics are not likely to to interfere with the functioning of the ABC the method. 87 00:12:15.600 --> 00:12:26.970 Victoria Sánchez Sebastián: So after all of these cross checks, we analyze the data from a will also run through data and seeing if utterly no significant access was found in upper limits were calculated. 88 00:12:27.570 --> 00:12:36.240 Victoria Sánchez Sebastián: A what we find is that the limits improve around a factor of 10 a with respect to the previous version of the analysis with 2015 2016 data. 89 00:12:36.810 --> 00:12:40.800 Victoria Sánchez Sebastián: A for the loyalty selection and I run a factor three for the high this election. 90 00:12:41.220 --> 00:12:48.150 Victoria Sánchez Sebastián: The the reason of these improvement is mainly the higher statistics, but also the triggers improvements and the better dysplasia diversification. 91 00:12:48.750 --> 00:13:00.930 Victoria Sánchez Sebastián: And this is especially for the case for the loyalty samples I indicated here in the blood on the left a how in the daily limits have improved a for one of the samples, they were using the media as the higgs. 92 00:13:01.530 --> 00:13:09.330 Victoria Sánchez Sebastián: A the higgs as an industry, and you can see, under write a summary of all of the limits that we get a full day hicks mediator and Defense color masters. 93 00:13:10.230 --> 00:13:19.860 Victoria Sánchez Sebastián: In the next slide I show a summary of other limits that we get a with different a media a mediator masses aim for higher lower me the message. 94 00:13:21.120 --> 00:13:22.470 Victoria Sánchez Sebastián: And when you have been. 95 00:13:24.330 --> 00:13:32.100 Victoria Sánchez Sebastián: And then okay in also a there are other searches looking for long enough part because, as you may know, in that last day looking for the same model. 96 00:13:32.640 --> 00:13:46.650 Victoria Sánchez Sebastián: And recently some nice summer a lot we're done is my son of his work and, indeed, we, we can see that our results extended the limits in the intermediate range between a displacement this isn't the ID and music the meter is based practices. 97 00:13:47.520 --> 00:13:52.410 Victoria Sánchez Sebastián: You can see you're totally a further Massachusetts so many ballads for masters in this link. 98 00:13:53.700 --> 00:14:12.690 Victoria Sánchez Sebastián: And finally, a we in for these version we included efficiency map that could be used for interpretation so just a very complex to have to understand, we have some more information in backup but the idea is that a here in the region efficiency is calculated a separated by a by. 99 00:14:14.190 --> 00:14:22.140 Victoria Sánchez Sebastián: By lps in okay in the being is calculated a take into account the LP kinematics take a position and decay type. 100 00:14:22.710 --> 00:14:40.890 Victoria Sánchez Sebastián: So more information on the some backup ABS, this can be used for interpretation in for during the accuracy of these maps a for the high the selection, we see that for all efficiencies around 0.5% percent accurate on 25% and for the loyalty selection a formula efficiencies around. 101 00:14:43.020 --> 00:14:45.210 Victoria Sánchez Sebastián: resource library to 33%. 102 00:14:46.230 --> 00:14:57.030 Victoria Sánchez Sebastián: So yeah that's all so basically we perform these foreign to a analyse so a looking for a purse of noodling lips particles they came in the color into. 103 00:14:57.570 --> 00:15:01.350 Victoria Sánchez Sebastián: nursing if you can't access is found with respect to the data even back on prediction. 104 00:15:01.800 --> 00:15:14.730 Victoria Sánchez Sebastián: And so we have read the limits in the improve with respect to the previous round of analyses thanks to all of these improvements, as I mentioned before, and the use of luminosity so the I think that's all from my side. 105 00:15:16.500 --> 00:15:25.020 Louie Dartmoor Corpe: Perfect Thank you very much Victoria really nice talk, and so I think we have time for a few questions and i'm going to close this question. 106 00:15:28.620 --> 00:15:31.230 Louie Dartmoor Corpe: There is one hand up from Larry Please go ahead. 107 00:15:32.280 --> 00:15:34.650 Larry Lee: hi thanks so much for that I stopped um. 108 00:15:35.070 --> 00:15:38.520 Larry Lee: I was wondering for the bid. 109 00:15:40.470 --> 00:15:48.420 Larry Lee: If you could say that more about which bit effects actually effect, but the signal region and, in particular. 110 00:15:49.740 --> 00:16:06.210 Larry Lee: The the trigger that you're using for for collecting a big sample like there are a lot of different kinds of bid is that sample from the trigger representative of the kind of effects that show up in the signal region, and if you just say a bit more about that. 111 00:16:08.190 --> 00:16:09.600 Victoria Sánchez Sebastián: Okay okay i'm not sure. 112 00:16:10.230 --> 00:16:11.670 Victoria Sánchez Sebastián: I only went to be able to. 113 00:16:11.760 --> 00:16:18.840 Victoria Sánchez Sebastián: To answer properly in what can you repeat it was no the trigger because a I didn't fully understand. 114 00:16:19.350 --> 00:16:23.010 Larry Lee: I guess I see you have some control sample taken with this dedicated. 115 00:16:24.660 --> 00:16:30.180 Larry Lee: As not as a trigger and and the like admixture of that sample from. 116 00:16:31.320 --> 00:16:39.030 Larry Lee: You know HALO versus being gas interaction versus upstream beam gas interacted like like you know these kinds of different effects that kick in. 117 00:16:39.540 --> 00:16:46.440 Larry Lee: How do you ensure that that's representative of what shows up in the signal region, especially by the time you throw this all into under on that. 118 00:16:47.730 --> 00:16:48.450 Victoria Sánchez Sebastián: Okay. 119 00:16:49.980 --> 00:16:51.870 Victoria Sánchez Sebastián: Sure, so I. 120 00:16:54.630 --> 00:16:56.670 Victoria Sánchez Sebastián: yeah i'm not sure if I know how to. 121 00:16:57.870 --> 00:16:58.740 Larry Lee: me as your hand up. 122 00:16:58.770 --> 00:16:59.820 Victoria Sánchez Sebastián: yeah, I guess, I guess, I. 123 00:17:01.140 --> 00:17:01.440 Victoria Sánchez Sebastián: needed. 124 00:17:01.530 --> 00:17:02.490 Victoria Sánchez Sebastián: A ticket will be. 125 00:17:02.850 --> 00:17:13.470 Emma Torró Pastor: yeah we need many texts to test the to make sure that is the case and and indeed they are really, really similar the on the main difference between the sample will collect with the trigger. 126 00:17:13.650 --> 00:17:16.500 Emma Torró Pastor: And what we actually have in the nominal trigger is the timing. 127 00:17:16.950 --> 00:17:28.590 Emma Torró Pastor: So in the trigger we only select deep with negative timing and that most of it, and so we get a big sample then so large statistics that we can use in the training. 128 00:17:29.100 --> 00:17:39.870 Emma Torró Pastor: Then the ones we collect and the nominal trigger i've mostly been with positive timing and, of course, we cannot reject those us at the carrier level just be a timing cut. 129 00:17:40.680 --> 00:17:53.220 Emma Torró Pastor: And yeah but, but this so aside from timing, the rest of characteristics are really, really similar so basically the same is just they arrive at different times some other detector. 130 00:17:54.540 --> 00:17:57.120 Larry Lee: That is that because of the different. 131 00:17:58.350 --> 00:17:58.710 Emma Torró Pastor: They come. 132 00:17:59.520 --> 00:18:02.370 Emma Torró Pastor: They just becoming different buttons crossings. 133 00:18:02.790 --> 00:18:05.460 Larry Lee: or two Okay, thank you, thanks. 134 00:18:07.320 --> 00:18:10.230 Louie Dartmoor Corpe: Thank you, we have time for question from products as well. 135 00:18:11.940 --> 00:18:14.820 Prabhat Solanki: So I had a question regarding this party when we did it. 136 00:18:15.180 --> 00:18:16.380 Prabhat Solanki: was just wondering wondering. 137 00:18:17.580 --> 00:18:21.690 Prabhat Solanki: What kind of party event information we use in this, but even. 138 00:18:24.240 --> 00:18:24.930 Victoria Sánchez Sebastián: Okay, I. 139 00:18:25.350 --> 00:18:27.990 Victoria Sánchez Sebastián: got the listing backup am so sorry. 140 00:18:28.440 --> 00:18:33.870 Victoria Sánchez Sebastián: In Okay, so one part is the so information for the debt, so I. 141 00:18:35.130 --> 00:18:35.850 Victoria Sánchez Sebastián: forgot the list. 142 00:18:37.980 --> 00:18:43.080 Victoria Sánchez Sebastián: A I guess the EDA sorry I guess you might say, that is because I completely forgot. 143 00:18:44.340 --> 00:18:44.940 Victoria Sánchez Sebastián: A but. 144 00:18:46.230 --> 00:19:06.900 Emma Torró Pastor: yeah so there's two types of information, we are the first type of be the outputs of the personality, so we include both the output of the possibility so discourse as the better than neural network for the two projects that are most likely to be our research. 145 00:19:08.310 --> 00:19:19.770 Emma Torró Pastor: And then we have like topology information, like the separation nita or the angular supply separation between these two tickets or things like. 146 00:19:20.820 --> 00:19:23.010 Emma Torró Pastor: the totality of the event. 147 00:19:23.070 --> 00:19:28.170 Emma Torró Pastor: hd or using a steel rate is one of the other variables that we. 148 00:19:31.530 --> 00:19:31.890 Prabhat Solanki: Thank you. 149 00:19:34.470 --> 00:19:34.590 Emma Torró Pastor: Thank. 150 00:19:35.580 --> 00:19:44.130 Louie Dartmoor Corpe: You, so I guess, I will move on, but I just wanted to make one one final remark about these efficiency maps, and so I think it was going to penultimate slide. 151 00:19:44.880 --> 00:19:54.540 Louie Dartmoor Corpe: And the, so this is OK, as a full disclosure I was part of this this particular team, but I just wanted to highlight that this is. 152 00:19:55.020 --> 00:20:01.260 Louie Dartmoor Corpe: A nice thing to add, in particular since the last workshops we've had a lot of focus on reinterpretation of LP results. 153 00:20:02.220 --> 00:20:12.570 Louie Dartmoor Corpe: This is parametric as a function of both lps PT the key position eater and K types so depending what kind of coke or laptop the case to. 154 00:20:13.560 --> 00:20:22.920 Louie Dartmoor Corpe: The information will be on have data and the the formula basically to calculate we've been index from those truthful quantities will be available, but the really nice thing is, you can see that these maps are. 155 00:20:23.400 --> 00:20:34.740 Louie Dartmoor Corpe: Pretty accurate within kind of 25% in the efficiency, been using only truthful information which is hopefully going to be a nice shortcut to be able to use this so theorist feedback is very welcome. 156 00:20:35.790 --> 00:20:49.950 Louie Dartmoor Corpe: Okay, I was just reading opposition to that Thank you again very much Victoria, for this very nice talk and I think we can move on to the second talk, which will be about and that you see be future dark sector sensitivity predictions from. 157 00:20:52.050 --> 00:20:55.440 Louie Dartmoor Corpe: From Michael Williams, so what can see you connected. 158 00:20:56.310 --> 00:20:57.000 Michael Williams: Can you hear me. 159 00:20:57.390 --> 00:20:59.970 Louie Dartmoor Corpe: We can hear you if you can show your side yeah. 160 00:21:00.000 --> 00:21:02.760 Michael Williams: I mean this line this up here. 161 00:21:03.360 --> 00:21:14.250 Louie Dartmoor Corpe: Okay perfect, yes, we can see your slides yeah full screen, so you have 12 minutes for this talk and we'll let you know when there's five minutes left Please go ahead. 162 00:21:14.910 --> 00:21:15.120 Michael Williams: alright. 163 00:21:15.240 --> 00:21:17.880 Michael Williams: So i'm just bringing up the chat and things, just in case. 164 00:21:19.290 --> 00:21:28.500 Michael Williams: This should be quick, I was actually asked to give this talk dealing with some updated dark sector projections that we did as part of the snow mass process. 165 00:21:29.730 --> 00:21:37.050 Michael Williams: So, starting with the dark photon so when we're talking about dark photons http we're. 166 00:21:37.620 --> 00:21:49.410 Michael Williams: Talking about the minimal model so not where there's a large coupling to the higgs but where the connection between the dark sector and the ordinary matter is by kinetic mixing between the dark photon and the ordinary photon. 167 00:21:51.030 --> 00:21:57.600 Michael Williams: This model is a is always represented in this two dimensional plane of the mass of the proton and the coupling strength. 168 00:21:59.280 --> 00:22:06.540 Michael Williams: The dark Gray, or the existing limits, other than HDTV and the yellow here is the CV results from run to. 169 00:22:09.390 --> 00:22:14.760 Michael Williams: And so, so far, like CP is only produced results with data from the debut on final state. 170 00:22:15.780 --> 00:22:22.650 Michael Williams: These are completely consistent with predictions that myself just eat Taylor Filton at all did. 171 00:22:23.520 --> 00:22:30.120 Michael Williams: A number of years ago, before that motivated this search and and as part of that work we actually predicted the sensitivity and run three and then. 172 00:22:30.690 --> 00:22:40.170 Michael Williams: pretty much for all future runs and so, by the time you on threshold we didn't have to do any updates here, since the predictions seem to match with the data, so we just expect that this will. 173 00:22:40.860 --> 00:22:49.950 Michael Williams: come to fruition, however, in the lower mass to below the dime you on threshold in run to lh CB did not actually have. 174 00:22:50.550 --> 00:22:56.760 Michael Williams: A electron ID in in the first software trigger states, there was no collimator reconstruction, the time for that. 175 00:22:57.330 --> 00:23:02.130 Michael Williams: And the original plans for the run three trigger also did not have that and so. 176 00:23:02.790 --> 00:23:10.740 Michael Williams: Originally, we had come up with a way of looking for dark photons that involve using radiative charm decays and the idea, there was essentially to use the charm part of the decay to get through. 177 00:23:11.130 --> 00:23:17.310 Michael Williams: The first couple trigger stages and then eventually when you get to having the full reconstruction, you can look for the dark photon. 178 00:23:18.150 --> 00:23:24.720 Michael Williams: However, in 2018 we actually managed to upgrade the trigger to be able to just directly look for dark photon D plus C minus. 179 00:23:25.200 --> 00:23:36.570 Michael Williams: And then run three, we also have now the full Calvin reconstruction from from the first stage onward so there's no real reason to do the rate of charm approach. 180 00:23:37.320 --> 00:23:45.330 Michael Williams: Of course, any dark photons produced and ready to charm would also be picked up by an inclusive search, but many, many more, that would be produced students a Python decays. 181 00:23:45.960 --> 00:24:04.050 Michael Williams: will also be available and so we've updated the sensitivity to account for that major change in the trigger and so now, when you take a union of the dialect con and the dime you and you get this large region here that's covered that should be covered in the next run. 182 00:24:05.910 --> 00:24:13.740 Michael Williams: Essentially, this this this lower edge here is more or less just set by the size of the vertex detector these these searches assume. 183 00:24:14.280 --> 00:24:25.440 Michael Williams: That we do what we've done so far, which is only use the case that happened within the vertex detector if you went beyond that maybe you could do better, and just talk about that we haven't studied that in detail, yet. 184 00:24:27.030 --> 00:24:29.970 Michael Williams: Okay, so that's the update on the dark photon. 185 00:24:31.620 --> 00:24:39.960 Michael Williams: higgs portal, so this is another place where at least to be his world leading limits to these are in grey a here, so we updated these it for a few things so one. 186 00:24:40.710 --> 00:24:48.150 Michael Williams: The ice results are actually from run one data and at this point, the first paper was now seven years old, and there was a second paper using. 187 00:24:48.600 --> 00:24:59.790 Michael Williams: a slightly different decay, which is six years old, they were using a slightly older model for the scale or hadron coupling, this is a non conservative gct regime so it's tricky. 188 00:25:01.380 --> 00:25:09.000 Michael Williams: More so i'll pretty much all newer predictions Codex be no longer part of what it tells us a newer model. 189 00:25:10.050 --> 00:25:19.980 Michael Williams: So what we did first is we actually just updated the old the LSD results use that new model so everything's available in the public in the public data that you can actually do that. 190 00:25:21.120 --> 00:25:30.060 Michael Williams: So we did that and that's what this Gray, is here, so these Gray regions are just updated to use the same scale or hadron model that everyone's using these days. 191 00:25:32.370 --> 00:25:38.460 Michael Williams: And then we also updated predictions for the demure on K, which is what the results are based on. 192 00:25:39.360 --> 00:25:50.910 Michael Williams: The naive job that was done in some other places that are mostly shown for projections just took the results from run one and just sort of scaled them by some luminosity based factor. 193 00:25:51.570 --> 00:25:56.460 Michael Williams: But, in fact, the TV limits are strongly dependent on the lifetime. 194 00:25:57.060 --> 00:26:04.380 Michael Williams: And so you can see that we did a terrible job of this and there's not a constant scale factor you actually do a lot better higher mass than you do it lower mass. 195 00:26:04.980 --> 00:26:10.980 Michael Williams: And this is because it higher masters in the run one day you're actually probing the prompt regime so at some point you start probing. 196 00:26:11.430 --> 00:26:19.470 Michael Williams: Long live the case and there's less background, so you do better there, we also added into this an estimate of what you could get by considering hydroponic. 197 00:26:20.130 --> 00:26:30.180 Michael Williams: The case so plus or minus 10 K plus K minus and so at lower masses, it seems that you can you can actually do better, is because the branching fraction. 198 00:26:30.690 --> 00:26:40.710 Michael Williams: Of a dark eggs into those funnels states is close to one compared to the rather small branch infraction into the new on a final state, so this is something that at least you can easily explore. 199 00:26:42.330 --> 00:26:47.220 Michael Williams: And you know so that that's something we'll see how that how that plays out it looks like he can do substantially better. 200 00:26:47.610 --> 00:26:55.920 Michael Williams: Low mass but, just like in the dark proton case ultimately you're limited by the lifetime acceptance of the vertex detector since again we assume that you decay there. 201 00:26:57.930 --> 00:27:04.710 Michael Williams: Finally, we also produce the first projections for accion like particles that coupled predominantly to glue ons. 202 00:27:06.420 --> 00:27:13.110 Michael Williams: So we've never led is not done a search here, this was a there was a paper that myself and you Tom Daniel wrote. 203 00:27:13.890 --> 00:27:28.710 Michael Williams: Several years ago where we actually worked out how these things decay and they decay into very terrible things like you know pi pi and pipeline gamma, and so there was no searches done for this, but we were able to estimate roughly what always speaking do these are incredibly conservative. 204 00:27:29.760 --> 00:27:40.470 Michael Williams: Estimates so the backgrounds are coming from some run one one inverse femto barn study of EDA prime decays. 205 00:27:40.950 --> 00:27:46.950 Michael Williams: So we're scaling up the backgrounds by a factor of three orders of magnitude, so they end up at rather absurd. 206 00:27:47.640 --> 00:27:59.940 Michael Williams: level something like a million background events per mass been and run six which i'm sure won't happen, it must be possible to do better, but anyway, you get conservative estimates that way, we also are using a rather. 207 00:28:01.770 --> 00:28:07.350 Michael Williams: I guess conservative way of dealing with the unknown UV physics, so in the in the accion case you have this. 208 00:28:08.220 --> 00:28:17.820 Michael Williams: parametric Lee long term that goes like some scale that you don't know we just set that to one, which is what we've done in our other papers that corresponds to a quite conservative. 209 00:28:18.780 --> 00:28:28.230 Michael Williams: thing, because if you even just put in one TV or something like that you would get a bigger bigger number, but we generally just set that to one thing that we don't really know what it is. 210 00:28:29.070 --> 00:28:38.490 Michael Williams: So, in any event we predict here that led to be can do very, very well, probably, in reality, maybe a factor of two or three better if the backgrounds are. 211 00:28:39.330 --> 00:28:53.670 Michael Williams: handled in some sort of proper way, but clearly this is worth doing alright so that's that's my summary So this was just a short update of the sensitivities so dark photons we can cover more area now because we actually have the ability to do, inclusive, the blessing minus. 212 00:28:54.840 --> 00:28:58.980 Michael Williams: This was just something not something that at least we really needed before, but now we've worked to get it in there. 213 00:28:59.940 --> 00:29:14.400 Michael Williams: The higgs Porto again we updated for more modern scale or hedonic coupling added some projections for hydroponic the case and produced first projections for the accion my particles couple to glue on and that's it i'll take any questions thanks. 214 00:29:18.120 --> 00:29:29.280 Louie Dartmoor Corpe: brilliant, thank you for this very nice talk about these these predictions, which is very, very exciting I think so let's open the floor for questions we have one from Fidel Please go ahead. 215 00:29:30.780 --> 00:29:35.190 Federico Leo Redi: Yes, hi sorry I didn't have the chance to ask you this question before but. 216 00:29:36.330 --> 00:29:49.920 Federico Leo Redi: Can you elaborate a bit more on the the electoral final states and specifically what they expect would be the challenges of foreign three and then later on for. 217 00:29:51.000 --> 00:30:03.450 Federico Leo Redi: For for me, but essentially seems Okay, there has been this trigger breakthrough which is impressive, how do you, what do you think is left as a preparatory work or what has to be done. 218 00:30:04.920 --> 00:30:06.480 Michael Williams: uh yeah, so I think. 219 00:30:07.500 --> 00:30:15.000 Michael Williams: there's a few things so one thing that we're working on now, so the fact that we can we can reconstruct and identify plus or minus inclusively. 220 00:30:15.900 --> 00:30:24.870 Michael Williams: At 40 megahertz is great, except that the standard model gamma star is for the prompt case the standard model gamma starting plus C minus phrase basically infinity. 221 00:30:25.920 --> 00:30:30.000 Michael Williams: So what we're working on, we can't just keep all of those, even though we know what we have. 222 00:30:31.020 --> 00:30:37.560 Michael Williams: So what we're actually working on is just filling monitoring histograms with e plus or minus mass for the prompt case. 223 00:30:38.790 --> 00:30:42.510 Michael Williams: This way we don't actually care what the rate is so this, this would be really like. 224 00:30:43.740 --> 00:30:51.180 Michael Williams: As far as I know, the first analysis ever done out of monitoring histograms but it looks like it'll work so that's one and then. 225 00:30:52.710 --> 00:30:59.430 Michael Williams: We so one of the things we had to work really hard on in the in run to is to actually do a data driven material map of the vertex detector. 226 00:31:00.090 --> 00:31:07.260 Michael Williams: And so the vertex detector has been completely upgraded and replace and run three so that'll that'll something we have to actually redo. 227 00:31:07.890 --> 00:31:20.250 Michael Williams: So I have to go redo the data driven maps is just uses material interactions and protecting to figure out, you know you look at where the vertices are and use that to figure out where the material is in reality will have to redo that. 228 00:31:21.570 --> 00:31:32.460 Michael Williams: And then, in the run to data and run through data, we still have to really completely validate the material vetoing for the E, plus a minus electrons are messy. 229 00:31:33.840 --> 00:31:38.820 Michael Williams: But we're working on that now, I mean I don't think it's going to cause any problems but it's not like. 230 00:31:38.880 --> 00:31:39.240 Michael Williams: You know. 231 00:31:39.330 --> 00:31:43.110 Michael Williams: You just do exactly what you did with new ons and expect it's going to work perfectly. 232 00:31:44.310 --> 00:31:46.020 Federico Leo Redi: And you weren't related ad. 233 00:31:48.330 --> 00:31:48.810 Michael Williams: i'm. 234 00:31:49.110 --> 00:31:49.650 Michael Williams: Sure, I mean. 235 00:31:49.680 --> 00:31:55.920 Michael Williams: Obviously, the better, you can do your electron ID in in the first trigger stage, then you'll you'll do better to some extent. 236 00:31:57.150 --> 00:32:01.500 Michael Williams: Although you know the gamma starting E plus C minus rate is so absurd that. 237 00:32:02.700 --> 00:32:05.670 Michael Williams: there's a floor there right so even with sort of crude P ID. 238 00:32:06.960 --> 00:32:12.870 Michael Williams: You you're you're tending to have a pretty high fraction of actual each plus a minus. 239 00:32:13.890 --> 00:32:21.150 Michael Williams: There so yeah certainly we're looking at improved electron ID and anything we get will definitely use. 240 00:32:22.170 --> 00:32:34.020 Michael Williams: But you know there's a there's a floor there, which is just there that everything is a plus or minus at some point, and I think that that's it's not orders of magnitude below just using really basic crude electron it. 241 00:32:36.690 --> 00:32:37.290 Federico Leo Redi: Okay, thanks. 242 00:32:39.300 --> 00:32:43.320 Louie Dartmoor Corpe: Okay, I think we have time so first question, please go ahead. 243 00:32:44.400 --> 00:32:45.030 Louie Dartmoor Corpe: yeah. 244 00:32:45.060 --> 00:32:45.690 Louie Dartmoor Corpe: hi Mike. 245 00:32:45.780 --> 00:32:54.930 Susanne Westhoff: Thank you for your talk um I have a question about production for the darks kayla and for the items what production channels to include. 246 00:32:56.430 --> 00:33:13.410 Michael Williams: um yeah So these are just assuming that use the same penguin decays that were used in the published hig skip the portal, the scale or portal search so in the in the published result from run one it's be to K X and vt K star X so uses those two. 247 00:33:15.120 --> 00:33:19.650 Michael Williams: And so, for this for these projections for simplicity, we just assume that used exactly the same. 248 00:33:20.760 --> 00:33:24.720 Susanne Westhoff: rate might even be higher right, including towns, for example. 249 00:33:26.130 --> 00:33:36.150 Michael Williams: um so yeah so well, including other char like cancer tough for us because they don't decay, very often inside the British editor. 250 00:33:38.370 --> 00:33:48.150 Michael Williams: Typically, but I mean certainly the the inclusive penguin decay rate for bees is like a factor of three or five higher than just these two particular decays. 251 00:33:48.630 --> 00:33:52.170 Susanne Westhoff: Okay, so you assume basically to to dominated by PD case. 252 00:33:54.030 --> 00:33:55.080 Michael Williams: yeah yeah I think. 253 00:33:55.650 --> 00:33:59.790 Michael Williams: that's that's a good assumption, at least for this for this mass range. 254 00:34:00.270 --> 00:34:13.170 Susanne Westhoff: yeah I mean on the lock screen, you probably won't see that much but one more reason, so this is more comment than a question so since you talked about that locks thing in loop and just like production from messenger case. 255 00:34:14.010 --> 00:34:22.350 Susanne Westhoff: This the locks have been resolved so better results, basically better theory predictions are available now and it's mostly been done. 256 00:34:22.650 --> 00:34:37.620 Susanne Westhoff: By a group around parties, Norbert and Martin bond and the item so so it might be worth checking with them and using this updated better predictions than putting an arbitrary scale into the lock. 257 00:34:38.160 --> 00:34:46.890 Susanne Westhoff: Again it's not it won't make a huge to America difference, but just for keeping up with the theoretical developed, and I would suggest, just to do that. 258 00:34:47.370 --> 00:34:56.130 Michael Williams: yeah I mean so in the actual in this actual snowmass White Paper we we say in the beginning, like we're doing this, you know there's other stuff out there. 259 00:34:57.690 --> 00:35:01.350 Michael Williams: I was kind of hoping that there'd be some attempt of like in the snowmass of. 260 00:35:02.760 --> 00:35:12.030 Michael Williams: Trying to have some kind of consistency, because then you know, like it moves it moves the results around by by a factor of two or three I think when you when you go between these different choices. 261 00:35:12.510 --> 00:35:13.080 Maybe. 262 00:35:14.190 --> 00:35:18.570 Michael Williams: You know it'd be nice if everyone was using the same choice, because then you're getting a fair comparison of the experiments. 263 00:35:18.600 --> 00:35:25.830 Susanne Westhoff: But the most precise prediction, I would suggest and it's not hard to obtain you just need to run the couplings ones from. 264 00:35:26.070 --> 00:35:37.770 Susanne Westhoff: Whatever high scale you choose and if there's an agreement to set that say 10 TV then it's an ambiguous what you get out and you just use numerical recited as a prediction. 265 00:35:38.070 --> 00:35:50.640 Michael Williams: yeah i'm aware of these it doesn't seem to be any consistency, so it was kind of hoping there'd be a consistency and then we go back and update because they figured they are that we would choose the same thing as everyone else, given that you do have to make a choice. 266 00:35:51.180 --> 00:36:04.050 Susanne Westhoff: I mean you can always we scale as long as you're doing single couplings cans and don't use multiple couplings and in certain model, then you could scale, the rate, if you have to correct resigned. 267 00:36:05.910 --> 00:36:09.090 Susanne Westhoff: yeah yeah I also don't want to get this to tank for you. 268 00:36:09.300 --> 00:36:13.350 Michael Williams: Today nice if there would be some just agreement of like this is. 269 00:36:13.350 --> 00:36:24.750 Michael Williams: What we do, and then we won't do it like I would definitely like that and, and I mean i'm familiar with these papers or you're talking about so I mean i'm happy with if there's some agreement on picking a scale and going with it. 270 00:36:25.830 --> 00:36:34.890 Michael Williams: So far that doesn't have to be a case so mismatched things but it looks like about a factor of two or three is what it does to the like any. 271 00:36:35.280 --> 00:36:36.540 Any outcase especially. 272 00:36:37.620 --> 00:36:41.550 Michael Williams: Around roughly by that, depending on which of the choices, you make. 273 00:36:42.420 --> 00:36:46.560 Susanne Westhoff: that's good to know and very nice experimental progress so that's really nice to see. 274 00:36:47.730 --> 00:36:48.150 Michael Williams: Thanks. 275 00:36:49.710 --> 00:36:57.450 Louie Dartmoor Corpe: Thanks Thank you RON for the discussion, I think, can can now probably move to the last talk of the session, which is from and asunder. 276 00:36:57.510 --> 00:37:00.240 Louie Dartmoor Corpe: On the recent atlas display stock photon jets. 277 00:37:00.270 --> 00:37:00.810 result. 278 00:37:02.160 --> 00:37:04.560 Louie Dartmoor Corpe: And yeah, of course, the discussion can continue on matter most. 279 00:37:06.750 --> 00:37:10.440 Alessandro Biondini: yeah can you do me also Can you see you know starting full screen. 280 00:37:10.800 --> 00:37:13.620 Louie Dartmoor Corpe: Perfect we see the slides, and we are you betting on so. 281 00:37:14.220 --> 00:37:15.300 Alessandro Biondini: Thank you very much so. 282 00:37:16.350 --> 00:37:16.950 Alessandro Biondini: and 283 00:37:17.280 --> 00:37:19.620 Alessandro Biondini: Also, thank you for giving me the opportunity to present. 284 00:37:20.550 --> 00:37:28.290 Alessandro Biondini: er the results from the search for lifelong the particles produced the origin at 13 TV became two pairs of laptops. 285 00:37:28.680 --> 00:37:36.990 Alessandro Biondini: setup fee with other subjects are also more or simply known as displaced after lunch at the analogy so start with a bit of an introduction, so, in general, we have. 286 00:37:37.890 --> 00:37:46.740 Alessandro Biondini: Many the DSM scenarios in which we have access to cover we decouple to do some our model, and in particular interest in the case of attack for them from exported again. 287 00:37:47.310 --> 00:37:57.540 Alessandro Biondini: Then through vector quarter to laptop or or right either answer and in this scenario, we are discovering parameter be nice armada in the sector, the specter portakabin which is. 288 00:37:58.140 --> 00:38:06.450 Alessandro Biondini: The die for the that far less lifetime and, in particular, is more complex means that we have this is not eligible left hand up for them. 289 00:38:06.840 --> 00:38:13.380 Alessandro Biondini: And if they are pharmacy significantly lower than the six months, then we will talk with medicaid so this means we have like seven feet. 290 00:38:13.710 --> 00:38:21.630 Alessandro Biondini: Even grassroots calling me that that's of leptons or not we should affect us that full on jets and also show on the upper part of this slide. 291 00:38:22.590 --> 00:38:31.050 Alessandro Biondini: To the to our our benchmark senior models, so the on the one on that side this stuff are busy even ballet model which is a bit more. 292 00:38:32.040 --> 00:38:44.100 Alessandro Biondini: Reference for most of the the one is being interpreted in the past version of the searcher while VHF channel that we have now started as an interpretation for different to what it is and in the lower a couple of Blocks so showing. 293 00:38:45.210 --> 00:38:49.740 Alessandro Biondini: Where the address ritual is it stands in the bathtub for our limits. 294 00:38:51.390 --> 00:38:59.940 Alessandro Biondini: As from that for lunch at Seattle So how do we tackle them in your doctor so after that for them is they came to me on so research for our climate in your monitoring them your spectrum either. 295 00:39:00.330 --> 00:39:06.540 Alessandro Biondini: Not matching jetta proximity ID, and these are close by in our search for an ethics that are calm. 296 00:39:06.930 --> 00:39:12.930 Alessandro Biondini: And the in this case, the main background with cosmic renewals wife, the performance, the game to electron soap opera. 297 00:39:13.260 --> 00:39:23.490 Alessandro Biondini: or let's block, for example, we then searched for our local admin and instruction jets in each car without a SIP trunk see me in the in the ID and also showing on the upper part. 298 00:39:24.420 --> 00:39:37.770 Alessandro Biondini: Why, it is also important not to tackle both dedicates two electrons from you and says, this gives you a basically a way to tackle different muscle and just have feedback for the forage, for example, me on the case i'm not available at very low masses. 299 00:39:39.030 --> 00:39:40.740 Alessandro Biondini: And as for the reconstruction just. 300 00:39:42.270 --> 00:39:47.820 Alessandro Biondini: A are we, representing the outside of sexual deviancy for us some of our benchmark. 301 00:39:48.420 --> 00:39:58.170 Alessandro Biondini: signal scenarios for for for me to pj we are able to reconstruct set up to the hospital middle middle layer White for the CDP J, which is the. 302 00:39:58.800 --> 00:40:08.850 Alessandro Biondini: We have deficiency report on the right side, we are able to resolve within the HR volume so we'll turn different different Alex why each foot to the pj types. 303 00:40:10.140 --> 00:40:18.150 Alessandro Biondini: As organizers three year for the first time in the fall down to one answer, we are not only investigating the room fusion production, but we also are the. 304 00:40:18.480 --> 00:40:23.640 Alessandro Biondini: Investigation of he associated production or vegetables on with a w both them. 305 00:40:24.120 --> 00:40:36.150 Alessandro Biondini: And therefore we then Target two different we are busy or two different to get started this project, for we have to use something together that will be three years, in particular for really bj use this narrows country years. 306 00:40:36.600 --> 00:40:43.920 Alessandro Biondini: which are based on Ms information, and we should have an incandescent Level one a plus, then I scan on the level three or four seven. 307 00:40:44.730 --> 00:40:57.780 Alessandro Biondini: That are Cone with what the nuances of why not to be matched to unlocking the ID and why, for the city pj we use the scholarship that he gets which will also be familiar since there's been told by a bitter from the commercial team. 308 00:40:59.700 --> 00:41:10.260 Alessandro Biondini: Like me, or I don't know me and for this we're searching for the slot manufacturing and narrow jets in the car with pete on the tracks as well and. 309 00:41:11.130 --> 00:41:23.250 Alessandro Biondini: On the other hand, for who gave us the fact that we can trigger from the electronic The case of w balsam so in this case, we think that mountain region or more standard and with higher efficiency trigger a single the electronic signature. 310 00:41:25.200 --> 00:41:31.500 Alessandro Biondini: And that's for a bit of an overview of the analysis challenges, as already mentioned, we have these two different categories of. 311 00:41:32.700 --> 00:41:44.610 Alessandro Biondini: Do the fact that we tackle both Jeff and the WHO with Jeffrey additional couple more complex on veto veto in the present our project on people intuition, we have provided with WHO. 312 00:41:45.720 --> 00:41:58.050 Alessandro Biondini: And then we have the mvp J type a number selection involving specific Apps on the constituents, which leads us to have six or seven other channels three four Jeff directly from who so who we have. 313 00:41:58.740 --> 00:42:03.060 Alessandro Biondini: The Channel where it will love to follow them on became to me on, so we have opportunity pj channel. 314 00:42:03.570 --> 00:42:12.630 Alessandro Biondini: are fully CDP json to CDP json and also channel, where we have on the UK somewhere wonderful this again to me and undoubtedly into an actual satellite other answer. 315 00:42:13.290 --> 00:42:22.020 Alessandro Biondini: While for the bridge the focus is on the casing Bobby sleepy Jason, this is the occasion which we are really bombarded with Jeff and we can benefit from the combination of these two. 316 00:42:22.290 --> 00:42:32.760 Alessandro Biondini: And so we like to see PGA and you sleepy Jason the last one, we have for Jeff about, we also have an exclusive once the reputation that to strengthen you mo D meet Sunday CPA. 317 00:42:33.930 --> 00:42:49.980 Alessandro Biondini: And that's for the burger and rejection, I will talk about it more on the next few slides but we developed for different analogies dedicated to target to improve the background rejection are made madrasa which are the other objects from multijet and it was just for be a CDP Jay. 318 00:42:51.690 --> 00:43:01.050 Alessandro Biondini: And also being this bag around the world for me to be Jane men vegans from cosmic claim your answer and then, as for the found leg on this to make that we also have a few words about the data but. 319 00:43:01.620 --> 00:43:05.550 Alessandro Biondini: As we speak on for this llp unconventional. 320 00:43:06.300 --> 00:43:22.260 Alessandro Biondini: So just we use the stuff without even a B, C D megatron estimate for whichever and vote on the master oxidizing deeply enough to accommodate the bible's and then we speak for ABC the region from when we can estimate number of events industry knowledge on with the from Russia in the. 321 00:43:23.340 --> 00:43:31.680 Alessandro Biondini: Lower corner of life so we're now talk a bit more about these neurotoxins study from the gst target so me, maybe. 322 00:43:33.480 --> 00:43:33.690 Alessandro Biondini: The. 323 00:43:34.950 --> 00:43:40.500 Alessandro Biondini: most difficult challenge for different two analogies to divulge a huge advantage just beg around for multijet ambiguous jetson. 324 00:43:40.830 --> 00:43:50.340 Alessandro Biondini: which also prevented being added on to for for us to have an exclusion in the gay so, for example, of redefining the case of that for them and for this we developed for their food on to. 325 00:43:50.670 --> 00:44:02.610 Alessandro Biondini: Our project publish on your network exploding 3D images and this is one of the first users in our search service dnn based on loan inputs, which are unread direction and karima that cluster information and also showing your. 326 00:44:04.410 --> 00:44:12.180 Alessandro Biondini: All these images actually look like and we investigates then train them submitted some other learning jetta and just from the that for them the case. 327 00:44:12.720 --> 00:44:23.760 Alessandro Biondini: And we can see the out score of our tiger in the lower right Plata where the out which one is for serious the tiger is trying to 149 04 better on jet set. 328 00:44:24.300 --> 00:44:33.360 Alessandro Biondini: and see our nicely, we can discriminate between the excluding the Australia, that represents the multijet multilateral ensure our signal benchmark more that's giving really. 329 00:44:35.250 --> 00:44:42.780 Alessandro Biondini: Big big improvement from the evidence to register them from being this backgrounder Similarly, we cannot use what is usually standard. 330 00:44:43.650 --> 00:44:50.130 Alessandro Biondini: For a non long in the 90s, so to make that type of JET cleaning to adjust beeps because this will mean also affecting our signal. 331 00:44:50.460 --> 00:45:05.220 Alessandro Biondini: So in this case also we use a gentleman just trained on signal sample and allowed to even beep sample for the rejection and we use this quarter of that target based on this, together with a cut on the JET timing and also younger separation between the jets. 332 00:45:06.660 --> 00:45:11.490 Alessandro Biondini: And as for the cosmic Ray nuance that we developed up, then you want deep your network factor. 333 00:45:11.880 --> 00:45:23.220 Alessandro Biondini: Which isn't says inputs, the timing of the actual impact permanent information of the in your answer and for also for this one, the target score responsive be shown on the lower part. 334 00:45:23.940 --> 00:45:34.320 Alessandro Biondini: With the background cost so cosmic from anti Bush crossings in in the dash the waiting area and the signal which nicely Victor pick that one. 335 00:45:35.070 --> 00:45:44.100 Alessandro Biondini: And then it's for the final destination, as already mentioned before we use this a B, C D meetup and, in particular your so take care of possible seemingly good job type this in our region. 336 00:45:44.610 --> 00:45:49.860 Alessandro Biondini: is doing a similar kind of story of signal and mega events, and we have no significant contribution from. 337 00:45:50.310 --> 00:46:00.750 Alessandro Biondini: With nontraditional from the beginning, this background at the macd level, and we have just a minor obviously about contribution from our cosmic creative background, which is subjective and cross check settle on. 338 00:46:01.590 --> 00:46:11.610 Alessandro Biondini: To chapter 17 bucks on the CD and multiple location region show really and I started off the A, B, C D, a move of the nola in all of our ABC planes. 339 00:46:12.420 --> 00:46:18.150 Alessandro Biondini: and also your listing or whether the claim that number to use for different obesity playing for our signal regions. 340 00:46:18.600 --> 00:46:33.090 Alessandro Biondini: And also put it on example your debut CD for the two co op je je je je FF on the lower part of the light from the signal of a B, C D nicely weakness in the region and the forum to obesity on the left hand side. 341 00:46:34.800 --> 00:46:47.100 Alessandro Biondini: And this is also a bit of our a then a summary of this material selection of these two table on the right hand side for the WHO who thought I would see Mary Johnson on the left hand side for the Jeff once. 342 00:46:47.730 --> 00:46:59.490 Alessandro Biondini: We pass on the number of any type of the pjs the aforementioned the a big to get selections and dedicated cats or dogs or send the entity quantities. 343 00:47:00.840 --> 00:47:13.740 Alessandro Biondini: As for that certainty a moving on we have that, of course, the other even bedroom estimation brings a similar concept that you're in the middle, that is, accordingly, a calculated propagating this is the uncertainty of the country Johnson. 344 00:47:14.370 --> 00:47:23.430 Alessandro Biondini: And as for the experience of systematics we have them nicely summarized here in this table for our six Sigma region, Sir, I like in the different contribution for the toddler. 345 00:47:23.850 --> 00:47:38.550 Alessandro Biondini: A uncertainty with for Jeff anything you want to take that sentence is being deleted systematic central region you're in just a certain distance by the engine that are still being the overlords have to be in this generation uses steam dominated by Nice, that is, the uncertainty. 346 00:47:40.380 --> 00:47:41.040 Alessandro Biondini: Okay, thank you. 347 00:47:43.980 --> 00:47:51.750 Alessandro Biondini: OK, and then, as for the boss feet, that is not background estimate, so we have that perfect and buzzfeed estimates were compatible with the one one Sigma. 348 00:47:52.050 --> 00:47:58.860 Alessandro Biondini: and show you understand what that unfortunately, the unbranded results show no access with respect to be yes, but the background, so we can see in the last two columns the. 349 00:47:59.160 --> 00:48:08.550 Alessandro Biondini: Number of expected events and seek natalie Johnson, and you observed, one for dollar 60 in the regions and all of these are were comparable to the Multi muscle, where we are left to. 350 00:48:08.910 --> 00:48:15.570 Alessandro Biondini: With setting limits on iPhone busy in a church and models and that's probably miss we put limits on restaurant. 351 00:48:16.380 --> 00:48:25.920 Alessandro Biondini: With a CLS meet on the using as a political leader and then the limits that we obtain for a given that follow on to our next w deployment style with love with a lifetime we're winning meetup, for example, that's shown in the. 352 00:48:26.310 --> 00:48:32.220 Alessandro Biondini: brought on the lower right to where we have the upper limits on the virtual should fall on their way actually send you that follow on property here. 353 00:48:32.910 --> 00:48:37.890 Alessandro Biondini: In the X axis and then this limiters function of the deck for the muscle than interpolated between the available. 354 00:48:38.190 --> 00:48:46.680 Alessandro Biondini: signal must points, and we need a according to the branching ratio, and this will be shown next slide there and I would just say he had a few more words about the different. 355 00:48:47.130 --> 00:48:50.550 Alessandro Biondini: combinations he says, I mentioned, this is the first time that the research, we are. 356 00:48:51.240 --> 00:48:59.580 Alessandro Biondini: Exploiting the acidic production model and it's Nice that there is a nice interplay between Jeff and who is showing this problem that shows the solution for the. 357 00:48:59.760 --> 00:49:08.580 Alessandro Biondini: One of them would be that funnel my sample so which we are bonnie inspiration for me fully CDP gh on that, since we have an underpass advocates who will act on something that further into. 358 00:49:09.120 --> 00:49:23.070 Alessandro Biondini: And we can actually see from the three groups that the Jeff for which is the blue car, we can see that there is dominating in the same region, and then the US Government complimentary who at smaller and higher to improving us and giving us a message. 359 00:49:25.110 --> 00:49:37.770 Alessandro Biondini: And I will show them inside the into the living in something with analysis in the Left left hand side for the misinterpretation on the inside 40 hmm interpretation in particularly. 360 00:49:38.220 --> 00:49:43.890 Alessandro Biondini: For somebody from the rbc, since these are new one for whichever direct comparison with the area searcher. 361 00:49:44.700 --> 00:49:54.480 Alessandro Biondini: And, in particular, we can see in the Left blocking read the the 10% contour for the one prompt the analogy sending the orange Cone is 10 percenter. 362 00:49:54.960 --> 00:50:03.210 Alessandro Biondini: led them to displace the exclusion, and we can see that, for the foreign to our we have a really significant. 363 00:50:04.050 --> 00:50:08.640 Alessandro Biondini: improvement to this is both thanks to the Internet of target central new production modes that exploded. 364 00:50:09.600 --> 00:50:13.980 Alessandro Biondini: And we also have these that we had our first exclusion affiliate trying to case which is allowing her to. 365 00:50:14.610 --> 00:50:18.870 Alessandro Biondini: produce and cover the region of the space so really all there for the masses. 366 00:50:19.440 --> 00:50:24.570 Alessandro Biondini: down to to muster to have the new on down to fuel me, and we also extended our each up to. 367 00:50:24.900 --> 00:50:34.920 Alessandro Biondini: me and it's also very nice to notice that we have for the learning channel and exclusion down to branch English of point 1% of its to that problem, the gate which is basically the standard model value. 368 00:50:35.160 --> 00:50:45.750 Alessandro Biondini: A prediction for X is equal to fix the cell sorry nicer solution and four, so we have the first thing interpretation into a church and models which is similar mission and drop them they're inside. 369 00:50:46.680 --> 00:50:55.890 Alessandro Biondini: And just put this results were presented earlier this year, also the as a comparable uptown problem, and you have now been backing into a payback which has to be submitted to the APP. 370 00:50:56.730 --> 00:51:04.110 Alessandro Biondini: And as for the team effort, we are also working now on a forum to be the best and I this we treat your fully a. 371 00:51:04.890 --> 00:51:11.190 Alessandro Biondini: start getting into the results for the end of this year and we've combined with Jeff and he and we also recently started. 372 00:51:11.940 --> 00:51:23.490 Alessandro Biondini: A prompt searcher fit for which we have been around one so we're really trying to do all the missing each other face Facebook spot we research and yeah that's what's happening promise. 373 00:51:26.820 --> 00:51:34.920 Louie Dartmoor Corpe: Thank you very much, I Center really nice presentation, for this very nice results see We already have a hand up from procedure, please go ahead. 374 00:51:36.120 --> 00:51:37.170 Rhitaja Sengupta: hello, can you hear me. 375 00:51:37.680 --> 00:51:43.800 Rhitaja Sengupta: Yes, yeah so in the ahl it narrows can vary search for another. 376 00:51:44.160 --> 00:51:55.380 Rhitaja Sengupta: me on with delta point five, so are you so this delta calculation, are you using the PT of these neurons or is it the physical separation in this neon chambers. 377 00:51:56.100 --> 00:52:08.610 Alessandro Biondini: yeah that is both the physical relations or the busier you get the harder to the distance in data plane, but then because of the trigger rates depending on the a PDF linear we'll have to put the. 378 00:52:09.960 --> 00:52:17.310 Alessandro Biondini: Law archive too busy, and the longevity of the have the me on the on Unfortunately, because of the T REX okay. 379 00:52:17.340 --> 00:52:19.920 Rhitaja Sengupta: So, you also have this physical separation. 380 00:52:20.640 --> 00:52:21.030 Alessandro Biondini: thing yes. 381 00:52:21.630 --> 00:52:33.480 Rhitaja Sengupta: Okay, and another thing so so you're talking about narrow jetson the hdl so here in this analysis you're not taking into consideration the case where the Doc. 382 00:52:34.020 --> 00:52:41.970 Rhitaja Sengupta: The Doc photon decays inside the non spectrometer into electrons so for electrons case it always has to be good before he can right. 383 00:52:43.200 --> 00:52:54.210 Alessandro Biondini: yeah in principle, if you will, about the gay so at this point the Channel act on the casing to the hospital very out for you to disentangle each will be able secularism you're basically so yeah it seemed interesting thing but. 384 00:52:55.050 --> 00:53:06.000 Alessandro Biondini: In principle, you're only sensible to instigate into two electrons or, like other ones in the in the edge calm, you can also see this also in exclusion visit this discontinuity you have. 385 00:53:06.480 --> 00:53:23.790 Alessandro Biondini: In the industry from below, to most of the nuances because the arranger of the exclusion engage how busy, the last time you access and then of course the the display the than the cabling ranger you access to is different from the young ones. 386 00:53:25.080 --> 00:53:30.330 Rhitaja Sengupta: And these narrow jets I mean how narrow are these I mean in terms of our I mean. 387 00:53:31.410 --> 00:53:35.430 Rhitaja Sengupta: When you are clustering these jets or what is the arlen for these jets. 388 00:53:36.030 --> 00:53:36.330 Sorry. 389 00:53:37.470 --> 00:53:43.590 Rhitaja Sengupta: So in the H can these narrow jets what is the our value for these jets when you are clustering them on. 390 00:53:45.570 --> 00:53:53.160 Alessandro Biondini: So the the we still have Okay, so you can actually do something, so the syllabus after them, between a busy week. 391 00:53:54.570 --> 00:53:55.350 Rhitaja Sengupta: Okay, this is it. 392 00:53:55.620 --> 00:54:04.050 Alessandro Biondini: You know, with the point four, but then inside this book calm for basically you ever maybe this will be helpful, for example, if you look. 393 00:54:04.440 --> 00:54:21.060 Alessandro Biondini: At the second major of these 3D images, you have enough, I am than the misery, the energy, you need, Sir, this is really what the image will will look like and so that's also what we use for discrimination with the with the with the CNN because this is our signal is a, this is a signal. 394 00:54:22.350 --> 00:54:33.390 Alessandro Biondini: can be data, and this will look at, of course, our Big Brother for for Meg rhonda and this one also allows us to do this, naturally, is Vanessa operation will be certificate. 395 00:54:34.830 --> 00:54:35.730 Rhitaja Sengupta: Okay, thank you. 396 00:54:39.300 --> 00:54:41.100 Louie Dartmoor Corpe: Okay, thank you very much, this question. 397 00:54:41.190 --> 00:54:45.480 Louie Dartmoor Corpe: I think we probably have time to squeeze in one more question if there is one. 398 00:54:47.460 --> 00:54:53.700 Louie Dartmoor Corpe: And if not, then I, I do have a curiosity so and, obviously, you have a callow dp J. 399 00:54:55.230 --> 00:55:07.020 Louie Dartmoor Corpe: channel which, on the face of it looks like quite similar to what we heard them, the first talk of the kind of tagger for the current ratio, which is also in the. 400 00:55:08.280 --> 00:55:18.630 Louie Dartmoor Corpe: In the perimeter but presumably they have like they have sensitivity to different that's a kinetic ranges can you maybe elaborate on this difference. 401 00:55:20.400 --> 00:55:22.140 Alessandro Biondini: yeah so we actually we done. 402 00:55:22.830 --> 00:55:31.740 Alessandro Biondini: A attacks or sort of what we leveraged in that also been a nice effort also from the gallery shooting, for example, reinterpreting the in their search. 403 00:55:33.390 --> 00:55:48.930 Alessandro Biondini: With our model and also we looked at actually in a result they were obtained our complimentary in the ranger also immediately goes also the, for example, the target of the master of the llp which we go a bit lower he must while with the courage to go and go up. 404 00:55:51.150 --> 00:56:00.990 Alessandro Biondini: up and become more on mass than we do because of the the visit apology that we interact with our data and we're also running to do something similar are also trying to. 405 00:56:02.010 --> 00:56:09.150 Alessandro Biondini: Do with the cost to try to get the server costs in printer printing or researching me in the model used by our restaurant. 406 00:56:10.920 --> 00:56:19.380 Louie Dartmoor Corpe: Perfect indeed and your you'd be sensitive to the masses and the career shooting where he would optimize more towards higher me data matters I guess know P masses. 407 00:56:21.540 --> 00:56:26.880 Louie Dartmoor Corpe: Great, thank you for the first time vacation and I think, in that case we can close here. 408 00:56:27.060 --> 00:56:29.460 Louie Dartmoor Corpe: And thank all of the speakers from this. 409 00:56:29.490 --> 00:56:45.060 Louie Dartmoor Corpe: session again so as Sandro Victoria and Michael so thanks everyone really nice session, and we continue about 10 minutes time after a short coffee break with more new LP results from around the legacy so thanks everyone.