1 00:00:01,140 --> 00:00:06,450 Okay, so my name is Doug Schaffer from the University of Chicago. I'll be talking on 2 00:00:06,450 --> 00:00:11,670 behalf of Atlas and CMS for searches for the Higgs boson decaying to invisible 3 00:00:11,670 --> 00:00:17,370 particles. So probably like you, I've been spending a lot of time on the couch 4 00:00:17,370 --> 00:00:22,860 recently, staying at home. But when we were able to get out and investigate the 5 00:00:22,860 --> 00:00:27,930 world, we saw quite a few indications that the standard model is not complete, 6 00:00:28,680 --> 00:00:35,040 including things like flat Flat rotation curves, remnants from the cosmic microwave 7 00:00:35,040 --> 00:00:42,330 background, and images like the bullet cluster in which you had slower gases and 8 00:00:42,330 --> 00:00:47,730 faster gases from those that interacted electromagnetically. So we have this 9 00:00:47,730 --> 00:00:51,780 strong evidence that there's a lot of matter out there that we currently are not 10 00:00:51,780 --> 00:01:01,410 observing. And so one possible way at the LSE to look for this is Using the Higgs 11 00:01:01,410 --> 00:01:08,010 boson. So we know that the Higgs boson couples to various particles and gives 12 00:01:08,010 --> 00:01:13,530 them mass. For example, the leptons acquire their mass through Cal coupling. 13 00:01:13,890 --> 00:01:20,100 So potentially some similar interaction is occurring also for these dark matter 14 00:01:20,100 --> 00:01:25,980 particles. And so usually we would just measure the Higgs width. And we would know 15 00:01:25,980 --> 00:01:31,920 whether or not it was decaying to additional invisible particles. But 16 00:01:31,950 --> 00:01:35,880 unfortunately, the Higgs intrinsic width is much smaller than the detector 17 00:01:35,880 --> 00:01:43,620 resolutions that listen CMS. So then there are various models to predict how if the 18 00:01:43,650 --> 00:01:49,320 dark matter is less than half of the Higgs mass, if it's a scalar, vector thermionic 19 00:01:49,620 --> 00:01:57,750 Dark Matter ways in which you could couple. So this is how we might observe 20 00:01:57,750 --> 00:02:02,010 the Higgs decay, how it's produced. Then how we actually search for it, because 21 00:02:02,010 --> 00:02:06,240 we're not going to measure any of these dark matter particles in the ATLAS 22 00:02:06,240 --> 00:02:11,910 detector assuming you know it's not interacting. So the production mechanisms 23 00:02:11,910 --> 00:02:17,880 that we'll focus on here are mostly the vector both on fusion, z, H, and th these 24 00:02:17,880 --> 00:02:23,850 are lower cross sections, then were used, like GG F to discover the Higgs, but they 25 00:02:23,850 --> 00:02:29,700 come with lower backgrounds, especially at higher net. So, missing transverse 26 00:02:29,700 --> 00:02:35,340 momentum, as you know, one of the important things on both the Atlas and CMS 27 00:02:35,850 --> 00:02:42,690 in order to state that we have if we found something Oh, and so I thought it was 28 00:02:42,690 --> 00:02:49,590 worthwhile to look at a couple comparisons of how the ETS is constructed in the two 29 00:02:49,590 --> 00:02:54,690 experiments, and I think the the main driving difference that you see between 30 00:02:54,690 --> 00:03:00,960 the two is is mostly in the soft terms. So the soft terms are, you know, in Your jets 31 00:03:00,960 --> 00:03:09,060 or leptons, your photons, they're soft signatures that we're not entirely sure 32 00:03:09,450 --> 00:03:16,800 which vertex they come from. And also, they are much lower energy, so it's less 33 00:03:16,800 --> 00:03:23,970 clear how to calibrate them. So on Atlas, the approaches taken to use only primary 34 00:03:23,970 --> 00:03:31,350 vertex tracks and to remove all of the associated clusters, the soft clusters on 35 00:03:31,350 --> 00:03:39,120 CMS, there's a different approach used to recreate these clusters by various factors 36 00:03:39,120 --> 00:03:44,850 to correct for the amounts of pileup. Now, there have been advances in many of these, 37 00:03:44,850 --> 00:03:48,300 but these are the methods that are definitions that are used in the analyses 38 00:03:48,300 --> 00:03:56,760 that I'll describe later. Okay, moving on to some of the searches. So, Higgs to 39 00:03:56,760 --> 00:04:02,850 invisible searches, as I showed come from quite a few different Higgs production 40 00:04:02,850 --> 00:04:09,630 mechanisms, and all of these are combined to give one ultimate limit and also the 41 00:04:09,660 --> 00:04:18,480 run one and run two analyses were combined to have an even better limit. And in 42 00:04:18,480 --> 00:04:23,880 addition to these, you can also beyond just these direct searches, you can use 43 00:04:23,880 --> 00:04:28,680 Higgs coupling measurements, so use all the visible channels as well as this input 44 00:04:29,400 --> 00:04:38,850 to put additional tighter limits get down to around 16% expected 30% observed in the 45 00:04:38,850 --> 00:04:45,090 Atlas adn versus empty barn, couplings measurement, similar results from CMS and 46 00:04:45,090 --> 00:04:47,670 Atlas using 36% of arms. 47 00:04:48,960 --> 00:04:55,380 Okay, so both Atlas and CMS with up to 36 and verse two bar and set limits of 26 and 48 00:04:55,380 --> 00:05:01,560 19%. What are the channels that feed into these? Well, the most sensitive is the BBF 49 00:05:02,100 --> 00:05:07,770 followed by the Z decay electronically plus Higgs boson through Higgs strong, and 50 00:05:07,770 --> 00:05:15,630 then the channel VH, where the B goes hydraulically. So I'll summarize through 51 00:05:15,630 --> 00:05:23,070 these first few in order of sensitivity. So the first one is the DBF analysis. So 52 00:05:23,070 --> 00:05:31,620 this channel looks for two jets that are well separated in the detector and have 53 00:05:31,620 --> 00:05:37,080 large and great mass and then a large missing transverse momentum that's 54 00:05:37,440 --> 00:05:49,380 balancing these two jets. So the Atlas analysis with 139% of Barnes requires this 55 00:05:49,380 --> 00:05:54,810 large met this large invariant mass and then the name of the game is you come with 56 00:05:54,810 --> 00:05:59,970 large backgrounds and you need some way to control those. So the way that it's done 57 00:06:00,000 --> 00:06:07,800 In the Atlas analysis is that the Z decay to visible leptons is a control region 58 00:06:07,800 --> 00:06:13,680 used to constrain z decay to neutrinos and then the W, where you have a visible 59 00:06:13,680 --> 00:06:19,470 leptons, and you've identified it is used to constrain the normalization of the W or 60 00:06:19,470 --> 00:06:23,550 you've lost one. So you can see the control regions in the upper right. And 61 00:06:23,550 --> 00:06:29,610 you can see the prefix and postfix agreement in the lower right as a function 62 00:06:29,640 --> 00:06:36,480 of digit and variant mass. And you can see there very good agreement, because of this 63 00:06:37,290 --> 00:06:43,440 ATLAS experiment proceeded to set limits on the Higgs branching ratio to invisible 64 00:06:43,590 --> 00:06:51,000 and set a limit of 13% to be compared to the 36 and first from the barn analysis 65 00:06:51,000 --> 00:07:02,430 with 28%. Okay, so what changed between these analyses the 3600 hours And the 139 66 00:07:02,430 --> 00:07:07,650 inverse empty barn. So there's re optimization to recover around 50% more 67 00:07:07,650 --> 00:07:14,520 signal on the same dataset. They improved lepton detail, reduce the wl new 68 00:07:14,520 --> 00:07:22,500 background from 48% to 33%. Increased acceptance for the Z to Ll control region. 69 00:07:23,160 --> 00:07:28,980 Also optimizing electron selection recovered 34% more. So this is really 70 00:07:28,980 --> 00:07:33,720 increasing your precision on that dominant z to new new background. In addition, 71 00:07:33,720 --> 00:07:39,810 simulation statistics were an issue. So there was quite a bit of development to 72 00:07:40,140 --> 00:07:45,000 generate three times the amount of simulated cross section with the same CPU 73 00:07:45,540 --> 00:07:50,550 using matrix element filters. And so then you can see the change in the 74 00:07:50,550 --> 00:07:57,990 categorization as opposed to just a three been MJ fit. There's now 11 bins and you 75 00:07:57,990 --> 00:08:03,420 can see the various increase incense signal the background as you go toward a 76 00:08:03,420 --> 00:08:12,840 higher m JJ and lo d phi JJ. Okay, the CMS analysis started CMS experiment also did a 77 00:08:13,200 --> 00:08:19,920 similar study, currently with 36 members from the barn. As opposed to the Atlas 78 00:08:19,920 --> 00:08:25,380 results. The main differences are that there is no third jet Ito. So this allows 79 00:08:25,380 --> 00:08:31,530 for more sensitivity to ggf Higgs production, there's no explicit Tao veto, 80 00:08:31,710 --> 00:08:37,980 and then slightly lower mgh a threshold also increases the GTF contribution 81 00:08:37,980 --> 00:08:44,310 contributions. And then the CMS result uses the shape fit. So the previous result 82 00:08:44,580 --> 00:08:50,370 has a control region for each signal region been so it's not a shape fit. They 83 00:08:50,370 --> 00:08:56,910 actually use the mtj shape to better constrain the backgrounds and so this 84 00:08:56,910 --> 00:09:05,460 leads to CMS having a 25% expected limit with 36% barn it'd be exciting to see with 85 00:09:05,460 --> 00:09:06,360 the full data set. 86 00:09:08,519 --> 00:09:14,759 The next most sensitive channel is the CH where the Z indicates left tonically. So 87 00:09:14,819 --> 00:09:21,089 the dominant background here is debose on z to two leptons and the other one going 88 00:09:21,089 --> 00:09:28,919 to neutrinos. And this as well as the W z. backgrounds are both modeled using 89 00:09:28,919 --> 00:09:36,209 simulation. And you can see on the right the end result of the Met distribution. 90 00:09:37,349 --> 00:09:43,439 And there's some slight excess, although still well within the uncertainties. And 91 00:09:43,439 --> 00:09:55,589 so there there's an expected 39% at 95% confidence level and observed of 67%. One 92 00:09:55,589 --> 00:10:01,499 exciting channel from CMS that was not included in the previous combination Is TT 93 00:10:01,499 --> 00:10:11,429 h. So this is 36 inverse dumpster barn results that was reinterpreted from a Suzy 94 00:10:11,429 --> 00:10:20,099 top partner search and it combines 01 and to leptons channels you can see in the 95 00:10:20,099 --> 00:10:28,289 lower right and then looks in various categories of met. And so the zero and one 96 00:10:28,289 --> 00:10:35,129 are about are the most sensitive expected channels, but the combined result is 48% 97 00:10:35,129 --> 00:10:43,439 expected and 46% observed so it's a really nice addition to results. So just to 98 00:10:43,439 --> 00:10:49,259 summarize all of the current results starting on the bottom with the run one 99 00:10:49,259 --> 00:10:58,259 plus run to 36% of barn results you can see that there are expected of around 15 100 00:10:58,259 --> 00:11:07,499 and 17% for Atlas CMS and then this th channel, then is a strong addition to 101 00:11:07,889 --> 00:11:13,229 their currently combined channels. And then you can see comparing to just the 36% 102 00:11:13,229 --> 00:11:19,559 of our and this is this is going to add quite a bit. And then the updated full run 103 00:11:19,559 --> 00:11:30,719 to Atlas results at 13% now sets the best limit on Higgs branching to invisible. So 104 00:11:30,719 --> 00:11:37,079 then one thing that we can do with the searches is to try to compare to direct 105 00:11:37,079 --> 00:11:45,209 detection experiments. It always comes with some copy some caveats, but it seems 106 00:11:45,209 --> 00:11:53,669 that we can in some decay channel or in some wimp models provide complementarity 107 00:11:53,669 --> 00:11:58,139 to some of the direct detection experiments especially at low went best 108 00:11:59,189 --> 00:12:08,339 and so you can see it Below, around 10 G, you can add sensitivity to lowers wimps. 109 00:12:10,079 --> 00:12:16,979 Okay, so then the last analysis I wanted to summarize is a bit different than the 110 00:12:16,979 --> 00:12:24,269 standard ones. So this is semi visible case to case. So this is just a simplified 111 00:12:24,269 --> 00:12:32,699 model trying to make metabolics assumptions and, and so this one looks for 112 00:12:33,299 --> 00:12:40,199 a dark photon, so a very light massless photon coupling to the Higgs and the way 113 00:12:40,199 --> 00:12:45,719 this is produced is with the photon and in addition to the dark photon So, you'll, 114 00:12:45,899 --> 00:12:55,319 you'll search for met plus photon and the search from CMS is done in the Z h z 115 00:12:55,319 --> 00:13:01,289 decaying WEP tonically channel and is put out with 137 interest rates To barns, and 116 00:13:01,289 --> 00:13:11,309 excludes hates branching ratio to gamma gamma dark of less than 5%. Okay, and then 117 00:13:11,309 --> 00:13:17,849 just some conclusions. So the Standard Model, Higgs production of invisible 118 00:13:17,849 --> 00:13:24,059 decays is only around point 1%. And so we can look to see for extensions of this. 119 00:13:25,349 --> 00:13:29,999 Beyond the Standard Model physics, as you can see, there's a rich regime of possible 120 00:13:29,999 --> 00:13:35,939 ways to search for Higgs to invisible with DBF leading the way and exciting new 121 00:13:35,939 --> 00:13:42,359 channels like tgh to be included, and then we'll continue to see more full run to 122 00:13:42,359 --> 00:13:47,219 results coming out. That should further push sensitivity. And I just wanted to end 123 00:13:47,969 --> 00:13:55,109 on one projection from CMS about the potential sensitivity with 124 00:13:56,490 --> 00:14:02,040 well first to be 300% the barn that we hope to get Soon, in addition to longer 125 00:14:02,040 --> 00:14:10,260 scale, HL hc projections that get down to around 4% sensitivity from the CMS 126 00:14:10,260 --> 00:14:16,020 experiment. So it'll be exciting to see with more data to look more closely it 127 00:14:16,110 --> 00:14:19,020 Higgs to invisible to case. That's it. Thanks. 128 00:14:19,710 --> 00:14:24,300 Thank you very much, Doug. This was a very nice and complete overview with a very 129 00:14:24,330 --> 00:14:25,200 nice result. 130 00:14:27,180 --> 00:14:29,340 Any questions, comments? 131 00:14:31,950 --> 00:14:33,480 I have one, maybe if I can. 132 00:14:35,040 --> 00:14:39,930 Hey, jack. Thanks. This was a really interesting talk. I was wondering if there 133 00:14:39,930 --> 00:14:47,640 are plans to reinterpret the Atlas? TT plus met Susie search that john presented 134 00:14:48,210 --> 00:14:53,190 in this model, in the same way as a CMS result that you mentioned? 135 00:14:54,330 --> 00:15:03,060 Yes, sir. Yeah, there is there are plans to include similar analyses For Atlas, but 136 00:15:03,090 --> 00:15:07,380 you know, obviously this takes time. So I think this will come out relatively soon. 137 00:15:08,880 --> 00:15:09,120 Lots 138 00:15:11,880 --> 00:15:12,720 more questions. 139 00:15:16,289 --> 00:15:17,369 Maybe I have one. 140 00:15:18,960 --> 00:15:21,720 So the new result the vdf. 141 00:15:23,640 --> 00:15:29,730 So you mentioned a few improvements. Did you check where you would be with the 142 00:15:29,730 --> 00:15:32,130 expected limit without those improvements? 143 00:15:33,720 --> 00:15:44,820 Ah, yeah. So as ever call, I think that it is roughly around scaling with a factor of 144 00:15:44,940 --> 00:15:53,820 two decrease from 28%. You would expect something like around 17 18% you have to 145 00:15:53,820 --> 00:15:59,070 keep in mind that there's increased pileup, and so then this results in some 146 00:15:59,070 --> 00:16:05,250 more pernicious backgrounds like Multijet. So it doesn't perfectly scale like the 147 00:16:05,250 --> 00:16:11,970 factor of two that you'd hoped from the 28%. But add something more around 17 18% 148 00:16:13,020 --> 00:16:16,140 Yeah, sorry, I don't have an exact number for you, but roughly that's 149 00:16:16,560 --> 00:16:24,090 so when the So, mainly This is a improve the acceptance and the categorization. So, 150 00:16:24,090 --> 00:16:26,400 which one is more 151 00:16:28,290 --> 00:16:29,460 more to the improvement? 152 00:16:30,030 --> 00:16:35,010 So, I think actually the things that contribute the most are actually this a 153 00:16:35,100 --> 00:16:42,420 decrease in the W background. And then I would say similarly, increasing the Z to 154 00:16:42,420 --> 00:16:49,740 Ll control region. So that is the precision on your largest uncertainty. So 155 00:16:49,740 --> 00:16:54,390 the more Zika Ll you have to constrain z to new new, the better. The analysis is 156 00:16:54,390 --> 00:17:00,960 going to do. Those, those two are two of the biggest improvements. And then if you 157 00:17:00,960 --> 00:17:06,570 want the additional, what was the biggest limiting factor in the previous analysis 158 00:17:06,570 --> 00:17:10,800 that was MC stats. So generating the specter of three was also a large 159 00:17:10,800 --> 00:17:16,980 contribution. In terms of the breakdown. I think the simulation if you would have 160 00:17:16,980 --> 00:17:22,560 stayed with the same amount of simulation would have been the biggest problem. And 161 00:17:22,560 --> 00:17:27,480 then if you just want pure analysis changes, then these leptons acceptances of 162 00:17:27,480 --> 00:17:30,090 the next biggest. Thank you. 163 00:17:32,430 --> 00:17:33,990 Any more questions coming? 164 00:17:34,710 --> 00:17:37,050 I have a question. That from Fadi. 165 00:17:37,260 --> 00:17:38,400 Yeah. Oh. 166 00:17:39,390 --> 00:17:50,670 So I'm curious about the gamma, the semi visible Higgs. So the question is, is this 167 00:17:50,670 --> 00:17:57,810 motivated by particular model or because I would, so it's the gamma gamma two per mil 168 00:17:58,290 --> 00:18:04,620 in the standard model, and Gamma dark is? Well, I mean, a couple of ships with, you 169 00:18:04,620 --> 00:18:11,580 know, epsilon charge to the top. And, you know, so any model of something like this 170 00:18:11,580 --> 00:18:15,240 would be super suppressed, but maybe I don't have something in mind that would 171 00:18:15,240 --> 00:18:20,910 give a big did you have something in mind that that is motivated by a particular 172 00:18:20,910 --> 00:18:21,720 model? Maybe? Sorry? 173 00:18:23,430 --> 00:18:28,950 I yeah, I must admit, I don't, I am pretty sure that there is a model and I think I 174 00:18:28,950 --> 00:18:32,760 can look it up and send it to you. I think that there's one referenced in the paper. 175 00:18:32,940 --> 00:18:38,430 Okay. I believe that there's the there are some papers that show that this could be 176 00:18:38,430 --> 00:18:44,790 large, which was the motivation to look for it. But yeah, I'm sorry, I don't 177 00:18:44,820 --> 00:18:46,500 remember the exact details. No, no, it's okay. 178 00:18:46,499 --> 00:18:48,869 You have the reference for the paper. Thank you. 179 00:18:55,019 --> 00:19:00,389 Any more questions? So you mean you're basically out of time, but maybe One more 180 00:19:00,389 --> 00:19:03,659 question to any of the speakers of the session. 181 00:19:08,670 --> 00:19:14,760 If not, thank you very much, Doug, and thank you to our speakers. It was a very 182 00:19:14,760 --> 00:19:18,960 nice session, I think and see you at the next part of our sessions.