1 00:00:02,070 --> 00:00:09,030 And the first talk is by Kate Pachal, who is talking about the dark matter searches 2 00:00:09,060 --> 00:00:10,440 at Atlas and CMS. 3 00:00:11,069 --> 00:00:13,649 Just checking Caterina you can hear me and see my slides? 4 00:00:15,150 --> 00:00:16,380 And 5 00:00:16,890 --> 00:00:19,410 also, if I fullscreen them just a second 6 00:00:21,570 --> 00:00:23,520 full screen, you can still see them. 7 00:00:24,660 --> 00:00:29,790 Yeah, everything works. And five minutes from the end of your talk, I will be signing a warning. 8 00:00:30,989 --> 00:00:31,889 Thanks a lot. 9 00:00:32,310 --> 00:00:35,700 Alright, um, first I just wanted to say thanks a lot to the organizers. You've 10 00:00:35,700 --> 00:00:39,090 done a lot of work to put together a really interesting online conference that 11 00:00:39,090 --> 00:00:42,780 I've gotten a lot out of. And thanks also to all of the participants who connected 12 00:00:42,780 --> 00:00:47,250 on a Saturday to listen to us talk about dark matter. That's much appreciated. So 13 00:00:47,250 --> 00:00:51,150 as Cate rina says, I'm here to give the experimental overview of some dark matter 14 00:00:51,150 --> 00:00:56,460 searches at Atlas and CMS. So I'm starting with a very, very baseline overview of 15 00:00:56,460 --> 00:01:00,000 what we're doing. Obviously, Simon's talk set the stage for this much, much better. 16 00:01:00,360 --> 00:01:03,600 But here's the doodle that you all will have seen a million times about dark 17 00:01:03,600 --> 00:01:06,900 matter searches and the different ways that we can do them. So if we assume we 18 00:01:06,900 --> 00:01:10,830 have some dark matter particle, which we will just say there's only one for now, 19 00:01:11,370 --> 00:01:14,850 nothing else, and that there is some form of interaction which connects it to the 20 00:01:14,850 --> 00:01:17,970 standard model, then there are typically three ways that we go about looking for 21 00:01:17,970 --> 00:01:21,810 this. So the first way would be to look for dark matter particles interacting and 22 00:01:21,810 --> 00:01:25,320 producing Standard Model particles. This is indirect detection, and is what you see 23 00:01:25,320 --> 00:01:30,510 at experiments like ice cube and super K. Next, you have dark matter and Standard 24 00:01:30,510 --> 00:01:33,840 Model particle scattering, which would be the direct detection experiments like Lux 25 00:01:33,840 --> 00:01:37,500 and Xenon. But there's of course, one more dimension, you can read this diagram, 26 00:01:37,530 --> 00:01:41,370 which would be standard model particles interacting in some way to produce dark 27 00:01:41,370 --> 00:01:47,640 matter particles. And this is what we have at colliders such as Atlas and CMS. So 28 00:01:47,670 --> 00:01:51,330 what you can see is that colliders really do make up a key part of this picture. All 29 00:01:51,330 --> 00:01:54,090 of these searches can access slightly different kinds of models, slightly 30 00:01:54,090 --> 00:01:59,070 different spaces. And so we do need all three components to build up a full search 31 00:01:59,070 --> 00:02:03,540 picture. Today, I'm just going to be talking about this one here, which is dark 32 00:02:03,540 --> 00:02:07,260 matter searches at Atlas and CMS. And I should put a further warning, a ton of 33 00:02:07,260 --> 00:02:10,320 different things can constitute Dark Matter searches, I'm only talking about a 34 00:02:10,320 --> 00:02:13,140 small fraction of those, there's going to be a ton more things that would be 35 00:02:13,140 --> 00:02:18,240 relevant here. But really, we can't talk about all of them today. So let's start by 36 00:02:18,240 --> 00:02:22,230 thinking about what Dark Matter could really actually look like at the LHC. So 37 00:02:22,230 --> 00:02:27,720 essentially, this blob from the previous slide, what is that? This depends a ton on 38 00:02:27,720 --> 00:02:32,910 our assumptions. And again, we've just had an entire talk about this. But to take it 39 00:02:32,910 --> 00:02:37,080 down to how we've been addressing this in the experiments, at very low energies, you 40 00:02:37,080 --> 00:02:39,840 really don't need a clearer picture than that. You can get away with just leaving 41 00:02:39,840 --> 00:02:42,930 the blob blurry and have an effective field theory that treats this as a four 42 00:02:42,930 --> 00:02:46,710 point interaction. And you're pretty much done. And this is valid as long as you 43 00:02:46,710 --> 00:02:50,580 have situations where the momentum transfer is small compared to the fundamental process. 44 00:02:50,580 --> 00:02:54,480 So for example, for direct and indirect detection experiments, that's valid, but 45 00:02:54,480 --> 00:02:58,260 at the LHC most of the time, there's one exception later in this talk, but most of the time, 46 00:02:58,260 --> 00:03:01,950 our high energies mean we really need a more complete picture than that. And so then 47 00:03:01,950 --> 00:03:05,100 we have to ask ourselves, what should that be? There's a ton of different models 48 00:03:05,100 --> 00:03:07,680 available, as you've seen with different levels of completeness, and 49 00:03:07,680 --> 00:03:10,830 simplification, and each of them has different dominant signatures 50 00:03:10,830 --> 00:03:14,250 experimentally, which is where this comes into becoming a really important question 51 00:03:14,250 --> 00:03:18,780 for us. So in this talk, I'm going to discuss Dark Matter by a few very 52 00:03:18,780 --> 00:03:24,870 simplified models, all fitting into the minimalist version of Simon's landscape, 53 00:03:25,020 --> 00:03:28,680 and highlighting the analyses that best constrain each and I'll be focusing on new 54 00:03:28,680 --> 00:03:33,300 results with full Run Two data as often as possible in the two experiments. So this 55 00:03:33,300 --> 00:03:37,170 is a picture that is a very useful diagram from the LHC dark matter working group 56 00:03:37,170 --> 00:03:41,760 that sort of clusters together these different models into regions, I guess 57 00:03:41,760 --> 00:03:45,210 where this over here is basically our E F T s. And it's not really what we do at 58 00:03:45,210 --> 00:03:49,890 the LHC anymore. In the middle here with simplified minimal Dark Matter models 59 00:03:49,890 --> 00:03:53,580 where we have just like maybe one or two mediators or one dark matter particle, 60 00:03:53,580 --> 00:03:56,460 everything relatively simple. This is pretty much where we're at right now 61 00:03:56,460 --> 00:03:59,940 today. And so I'll be using a couple of these models to frame the discussion. 62 00:04:00,630 --> 00:04:01,710 We're going to have in a minute. 63 00:04:01,950 --> 00:04:06,060 But then of course, there are all of these much more complete models as well. These 64 00:04:06,060 --> 00:04:10,230 were mentioned by Simon, they're a little beyond today's scope. One of the ones that 65 00:04:10,230 --> 00:04:14,460 we obviously have a lot of experimental limits in terms of is supersymmetric 66 00:04:14,460 --> 00:04:17,640 models. But we've had a lot of different talks on those in this conference. So we 67 00:04:17,640 --> 00:04:22,170 don't need another one right now. I want to start by talking about mono x 68 00:04:22,170 --> 00:04:27,390 signatures. So this is a really key final state in all of our dark matter searches. 69 00:04:27,480 --> 00:04:32,130 And what is it. Basically you have missing energy plus a physical object. So we have 70 00:04:32,130 --> 00:04:36,000 something usually a jet, but you could radiate anything off of an incoming Quark 71 00:04:36,000 --> 00:04:41,070 in our mystery diagram. And this gives us both something we can trigger on, and the 72 00:04:41,070 --> 00:04:45,000 momentum that we can measure which is important because then we can also have 73 00:04:45,180 --> 00:04:50,220 missing energy via an unobserved system of particles on the other side of our event. 74 00:04:50,640 --> 00:04:53,610 If we look for momentum balance, we see that it isn't there. This is our missing 75 00:04:53,610 --> 00:04:57,690 energy. This tells us there were invisible particles. The great thing about this, as 76 00:04:57,690 --> 00:05:01,440 you can see is that it still doesn't matter what the blob is. That's really 77 00:05:01,440 --> 00:05:05,640 model independent, no matter what your interaction is between the dark matter and 78 00:05:05,790 --> 00:05:08,730 the Standard Model, you can still get this signature because all you need is for one 79 00:05:08,730 --> 00:05:11,310 of your incoming quarks to radiate something and then to make some dark matter. 80 00:05:12,210 --> 00:05:15,210 What's interesting though, is that once you assume a model, a lot of other 81 00:05:15,210 --> 00:05:18,990 signatures become powerful as well. So today, we're actually going to go beyond 82 00:05:18,990 --> 00:05:21,930 the Mono jet signature, we're not going to talk about mono jet. And we're going to 83 00:05:21,930 --> 00:05:25,380 showcase instead, recent Atlas and CMS results in the areas of their own unique 84 00:05:25,380 --> 00:05:30,990 strengths. So let's start out by talking about an S channel Z prime model and the 85 00:05:30,990 --> 00:05:34,830 mediator searches that you get from that. So let's consider that we just stretch out 86 00:05:34,830 --> 00:05:38,310 our blob and we have a single interacting mediator x. This could be all kinds of 87 00:05:38,310 --> 00:05:44,370 things. It could be vector, scalar, pseudo scalar, whatever you like. But 88 00:05:44,370 --> 00:05:47,640 fundamentally, we're just going to have one massive mediator, one dark matter 89 00:05:47,640 --> 00:05:53,370 particle and two allowed vertices. For this set of comparisons we also then pick a 90 00:05:53,370 --> 00:05:56,430 vector, axial vector, Zed prime cause that has higher cross sections and it's a little 91 00:05:56,430 --> 00:05:59,850 easier for us to handle. And when you think about this model, you can see that 92 00:05:59,850 --> 00:06:03,990 there are two key signatures. The first one is the mono x that we just talked about, 93 00:06:04,110 --> 00:06:07,920 where you have a decay of your mediator to the dark matter particles. And then you 94 00:06:07,920 --> 00:06:11,280 additionally have a radiated something that you can trigger on and look at and 95 00:06:11,280 --> 00:06:15,450 this gives us a mono x signature. You additionally have this one down here where 96 00:06:15,450 --> 00:06:19,440 you could have some coupling from your Zed prime to fermions. And you just have a 97 00:06:19,440 --> 00:06:23,430 pair of fermions in the final state. This is a visible channel. So this is really, 98 00:06:23,430 --> 00:06:26,850 really useful for us because we don't have to worry about the missing energy. And it 99 00:06:26,850 --> 00:06:29,940 means that we have a resonance we can directly reconstruct by looking at those 100 00:06:29,940 --> 00:06:34,350 two fermions that come out. So I also want to highlight because we're going to be 101 00:06:34,380 --> 00:06:38,040 using this model a lot throughout this talk. There's only two new couplings in 102 00:06:38,040 --> 00:06:42,480 it. There's the coupling, in this case to quarks but could be leptons as well 103 00:06:42,930 --> 00:06:46,230 between your Zed prime and whatever fermions you have and there's the coupling 104 00:06:46,260 --> 00:06:51,210 of the Zed prime to dark matter. What this means is that the values we choose for 105 00:06:51,210 --> 00:06:56,520 these couplings strongly affect the limits we present. So we can have a large amount 106 00:06:56,520 --> 00:06:59,160 or a small amount of exclusion in this model pretty much by tuning these 107 00:06:59,160 --> 00:06:59,790 parameters. 108 00:07:00,900 --> 00:07:03,600 This is something to bear in mind when you look at any limit plots in it. 109 00:07:04,470 --> 00:07:08,220 Okay, from there, let's get on to our first experimental result, we're going to 110 00:07:08,220 --> 00:07:11,820 be talking about dijet resonances. And I want to highlight this really nice, new 111 00:07:11,820 --> 00:07:16,620 paper from CMS here. So dijet resonance is, in some sense, the absolute simplest 112 00:07:16,620 --> 00:07:20,670 of the mediator searches. If you have qq going to Zed prime as an allowed vertex, 113 00:07:20,700 --> 00:07:24,630 then obviously you can have Zed prime back to qq as your decay channel, all you 114 00:07:24,630 --> 00:07:29,040 need to do is take your two jets, find the invariant mass and look in it for some 115 00:07:29,040 --> 00:07:33,300 sort of new resonance. If you have a background made entirely of Q C D jets, then 116 00:07:33,300 --> 00:07:37,680 you expect this mass spectrum will be smoothly falling. As you can see this line 117 00:07:37,680 --> 00:07:41,550 of black dots in the right hand side plot here, where signals will appear as peaks 118 00:07:41,550 --> 00:07:45,600 that you will be able to pick up on top of that background. Now traditionally, the 119 00:07:45,600 --> 00:07:48,360 way that we've done these searches is we use an approach to the background 120 00:07:48,360 --> 00:07:51,900 estimation where we essentially fit a functional form that is smoothly falling 121 00:07:51,900 --> 00:07:56,850 like we expect the Q C D distribution to parameterize the background. This new 122 00:07:56,850 --> 00:08:00,780 analysis from CMS has done something really fun they've actually used the data 123 00:08:00,810 --> 00:08:04,650 which has jets going really forward in the detector to predict the shape of the 124 00:08:04,650 --> 00:08:08,370 spectrum in the center, when you have jets in the center of the detector, you can see 125 00:08:08,370 --> 00:08:10,890 in the bottom here that there's actually really good agreement between the two 126 00:08:10,890 --> 00:08:17,760 methods, which are the red and the green leftovers. But the fact that they're using 127 00:08:17,760 --> 00:08:21,060 this method instead means that you can have much wider signals and still 128 00:08:21,060 --> 00:08:25,500 constrain them compared to what was possible with the fit only search. And 129 00:08:25,500 --> 00:08:29,490 this allows us to set limits up to much, much higher couplings than were done in 130 00:08:29,490 --> 00:08:33,390 the past because larger couplings tend to give you a wider width of this resonance. 131 00:08:33,720 --> 00:08:39,570 So this means that the, the CMS result was actually able to extend limits on this 132 00:08:39,570 --> 00:08:43,530 model all the way up to cases where you have a width to mass ratio up to 50%, which 133 00:08:43,530 --> 00:08:48,990 for us is really quite novel and you can see this really nice limit plot here. The 134 00:08:48,990 --> 00:08:52,200 next thing I want to talk about is if you have a lepto-philic Zed prime. If you 135 00:08:52,200 --> 00:08:56,370 allow a coupling from the mediator to leptons, then you can have a dilepton 136 00:08:56,370 --> 00:08:59,700 final state as well and this is a very clean powerful decay channel which has 137 00:08:59,700 --> 00:09:03,390 lower backgrounds than dijets has and can also be a very important sort of 138 00:09:03,390 --> 00:09:10,530 mediator search. So you can see this new Atlas result here, we have electron pair 139 00:09:10,800 --> 00:09:14,580 channel on the left hand side and muons on the right hand side. One thing you'll 140 00:09:14,580 --> 00:09:17,700 notice here is that because lepton triggers have much lower thresholds, we're 141 00:09:17,700 --> 00:09:21,480 able to access much lower masses here than the dijet search could get to. So 142 00:09:21,480 --> 00:09:25,290 these axes go down a lot lower. The other thing is that the resolution is better in 143 00:09:25,290 --> 00:09:31,020 the e e channel. And this means that we have these really fine peaks, which are, again, 144 00:09:31,050 --> 00:09:34,260 a big improvement on what we can do in dijets. So in the situation where you 145 00:09:34,260 --> 00:09:38,400 might have a coupling to leptons this is a really great signature to look at as 146 00:09:38,400 --> 00:09:43,800 mediators. You can also of course, assume that you have heavy quarks instead of 147 00:09:43,800 --> 00:09:48,180 regular quarks and we can look for a TT bar resonance. This is a new Atlas full 148 00:09:48,180 --> 00:09:51,720 run 2 result that came out quite recently. In this case, you would want to 149 00:09:51,720 --> 00:09:56,910 identify the hadronically decaying tops, which turn into large radius jets and 150 00:09:56,910 --> 00:10:00,480 have various substructure information in high mass that can make them identifiable 151 00:10:00,480 --> 00:10:05,700 as coming from top Quarks. So we would tag two top like jets and look at their 152 00:10:05,700 --> 00:10:10,230 invariant mass as well. Now, in order to sort of assist with this, the analysis was 153 00:10:10,230 --> 00:10:14,400 looking at the tracks that are contained within these high radius, or large radius 154 00:10:14,400 --> 00:10:17,850 jets to identify when there's a B jet inside each of them. Ideally, you would 155 00:10:17,850 --> 00:10:23,310 expect there to be one B jet within each of your top candidates. There's also 156 00:10:23,310 --> 00:10:26,640 often another situation that we fail to tag one adequately. And so there are two 157 00:10:26,640 --> 00:10:30,540 signal regions for this analysis, one with just one B tag required, where you can see a 158 00:10:30,540 --> 00:10:34,050 smaller signal contribution, and one with two b's required where you can actually 159 00:10:34,050 --> 00:10:39,900 see a fairly dominant signal shape, but then taking the two in combination gives a 160 00:10:39,930 --> 00:10:43,560 stronger limit. And so you can see that nice limit obtained from the statistical 161 00:10:43,560 --> 00:10:48,030 combination of the one and two b signal regions right here. I want to highlight 162 00:10:48,030 --> 00:10:51,480 that these three curves here are different theoretical cross sections, of course, 163 00:10:51,480 --> 00:10:54,420 where the theory cross section crosses the black line, that's the mass limit being 164 00:10:54,420 --> 00:10:58,320 set on this model by the analysis. But you can see here the difference between 165 00:10:58,320 --> 00:11:02,460 the blue dotted line and the red dotted line is the difference in the width of the 166 00:11:02,490 --> 00:11:06,360 mediator that you assume. And so that can really, really change the the actual 167 00:11:06,360 --> 00:11:07,380 exclusion that's obtained. 168 00:11:09,179 --> 00:11:13,979 Now, I mentioned that low mass is one of the advantages of the dilepton analysis. 169 00:11:14,249 --> 00:11:19,199 What if we want to go even lower in any object that we have, then the simple 170 00:11:19,469 --> 00:11:22,949 resonance searches we've seen a moment ago? Well, it's really challenging to 171 00:11:22,949 --> 00:11:27,959 search for low mass mediators, because of our triggers. This is something that Simon 172 00:11:27,959 --> 00:11:31,529 mentioned a moment ago, but essentially, the decay products from our resonance have 173 00:11:31,529 --> 00:11:35,129 to have high enough momentum to pass a trigger so that we save the event. And 174 00:11:35,129 --> 00:11:39,869 in ATLAS and CMS, we have two main methods to work around this. One of them is to 175 00:11:39,869 --> 00:11:44,069 trigger on associated objects, which are produced with the resonance that you're 176 00:11:44,069 --> 00:11:47,789 looking for, and the other is to perform the analysis at the trigger level. So I'm 177 00:11:47,789 --> 00:11:50,729 just going to start by talking about triggering on initial state radiation. 178 00:11:50,729 --> 00:11:55,019 This is on some associated object. Imagine x is the resonance that we want to find 179 00:11:55,019 --> 00:11:59,069 here. It's going to decay to two jets but the jets are too soft to pass a trigger. 180 00:11:59,309 --> 00:12:03,539 If we have a hard radiation of an additional jet or a photon or something else, we can 181 00:12:03,539 --> 00:12:06,389 use that to trigger and then look in the rest of the event for the thing that we 182 00:12:06,389 --> 00:12:10,919 want to reconstruct. Now, if this is really, really quite light, then these 183 00:12:10,919 --> 00:12:14,669 decay products will end up being boosted. And this is the CMS result that I want to 184 00:12:14,669 --> 00:12:18,149 talk about at this moment. In this case, the decay products of the boosted Zed 185 00:12:18,149 --> 00:12:22,559 prime are so close together that they become a single large radius jet. You can 186 00:12:22,559 --> 00:12:25,949 then look at the distribution of energy within the large radius jet in the same 187 00:12:25,949 --> 00:12:29,789 way as we top tag to select for topology that looks like there are sort of 188 00:12:29,789 --> 00:12:34,259 two separate energy clusters. And this is something we can use to reject Q C D and 189 00:12:34,259 --> 00:12:38,249 identify the signal. This then gives you something where you can estimate the 190 00:12:38,249 --> 00:12:43,139 background using events which failed the large-R jet tagging selections. And you can see 191 00:12:43,139 --> 00:12:47,849 the really nice estimate here and this is then used to push a resonant search all 192 00:12:47,849 --> 00:12:49,739 the way down to around 50 GeV, 193 00:12:50,010 --> 00:12:51,330 which is pretty fantastic. 194 00:12:52,889 --> 00:12:56,159 Atlas recently put out an additional full Run 2 result which is looking for the 195 00:12:56,159 --> 00:13:00,539 same kind of Zed prime mediator but where you have initial state radiation of a W or 196 00:13:00,539 --> 00:13:04,319 Zed instead. And so this is essentially looking for dijet production in 197 00:13:04,319 --> 00:13:07,949 association with something that gives you a high p T lepton and the lepton is used for 198 00:13:07,949 --> 00:13:11,789 triggering. Here again, you can see that it's possible to push the dijet spectrum 199 00:13:11,789 --> 00:13:18,089 down to around 250 GeV, which for such a search is really quite low. And once 200 00:13:18,089 --> 00:13:22,529 again, fit this smooth distribution that goes all the way up to around 7 TeV, and then 201 00:13:22,529 --> 00:13:27,269 search in it for excesses. As with the other things that we've been showing here, 202 00:13:27,269 --> 00:13:30,899 you can see there aren't any, but it's pretty neat to be able to access this 203 00:13:30,899 --> 00:13:34,679 range when we have a look. Once again, here, the fit is sent to the combined e 204 00:13:34,679 --> 00:13:38,009 mu channels, and you can see a limit plot on the right hand side, you'll note 205 00:13:38,009 --> 00:13:41,039 that the intersection of the theory curve with the observed limit is at quite a bit 206 00:13:41,039 --> 00:13:44,879 of a lower mass here than the previous limits we've seen in this model. And 207 00:13:44,879 --> 00:13:48,359 that's essentially because of the lower probability when you expect your radiated 208 00:13:48,359 --> 00:13:52,949 W or Zed in the initial state as well. So the next thing is bringing us back to the 209 00:13:52,949 --> 00:13:56,279 trigger level which is really quite an exciting topic. So we've got two cool new 210 00:13:56,279 --> 00:13:56,879 results 211 00:13:57,570 --> 00:13:58,620 to highlight right here. 212 00:14:00,299 --> 00:14:04,979 I can see that I switched up my links by mistake. This is obviously a CMS result. Ah 213 00:14:04,979 --> 00:14:09,479 yes, CMS result is down here, Atlas result is up there. Both have done 214 00:14:09,479 --> 00:14:13,199 trigger level jet analyses. But the one I want to highlight right here is an 215 00:14:13,199 --> 00:14:16,589 interesting combination of the initial state radiation and the trigger level 216 00:14:16,589 --> 00:14:22,019 approaches. So this is a new result from CMS, which is using three jets to have 217 00:14:22,049 --> 00:14:29,309 enough combined sum momentum to pass a trigger. And then saving only the calorimeter 218 00:14:29,309 --> 00:14:33,389 jet information and nothing else so that by saving a minimal amount of data for a 219 00:14:33,389 --> 00:14:36,899 large number of events, you can overcome the trigger rate reduction, which would 220 00:14:36,899 --> 00:14:41,009 otherwise happen. They then only take the two hardest jets to make the invariant 221 00:14:41,009 --> 00:14:44,459 mass and assume the third one was just initial state radiation. But when doing 222 00:14:44,459 --> 00:14:44,879 this, 223 00:14:45,000 --> 00:14:46,620 what you can see is that there's a really 224 00:14:47,700 --> 00:14:52,830 fairly high statistics distribution that again, can get down to around 300 GeV which 225 00:14:52,830 --> 00:14:56,520 can be used to search for excesses with the signal shapes that you can see here. 226 00:14:57,540 --> 00:15:00,450 There's also a new result I think is really interesting, which is using the 227 00:15:00,450 --> 00:15:05,160 same trigger level strategy, so again, saving a lot of data, or saving a small 228 00:15:05,160 --> 00:15:08,430 amount of data for a large number of events, but doing this for the first time 229 00:15:08,430 --> 00:15:13,110 for muons, and then doing a muon dilepton invariant mass search, 230 00:15:13,230 --> 00:15:17,940 similar to the one that we did before. So you can see right here the distribution of 231 00:15:18,300 --> 00:15:23,910 the invariant mass of muons in this CMS analysis, you can see a nice Zed peak 232 00:15:23,910 --> 00:15:29,820 here. And then as you go downwards, around 50 GeV, the red line here, which is from 233 00:15:29,820 --> 00:15:34,110 the standard triggers, drops off as trigger pre sales kick in, whereas the green line 234 00:15:34,110 --> 00:15:38,820 here represents what's gained by doing the scouting or trigger level analysis and 235 00:15:38,820 --> 00:15:41,430 taking the muons just at the trigger level. And you can see that at the lowest 236 00:15:41,430 --> 00:15:44,280 end here, it's about a two order of magnitude gain, which is really, really 237 00:15:44,280 --> 00:15:50,130 cool. And so this region is now able to be searched for new excesses, in a much more 238 00:15:50,340 --> 00:15:54,900 sensitive way than we had before. The last thing I want to talk about in the resonance 239 00:15:54,930 --> 00:15:58,740 search section is diboson resonances. This is a little bit different from the 240 00:15:58,740 --> 00:16:02,460 simplified model we've been talking about so far, but in some models, you can 241 00:16:02,460 --> 00:16:06,540 essentially allow a Zed prime to couple to W and Z bosons. And then depending on the 242 00:16:06,540 --> 00:16:09,840 other couplings, you have various available production modes. For that you 243 00:16:09,840 --> 00:16:14,370 could have gluon-gluon fusion, Drell-Yan, and V B F. It really depends. So this is a 244 00:16:14,370 --> 00:16:18,330 recent Atlas search, which has a huge number of different signal regions and 245 00:16:18,330 --> 00:16:21,990 plots. I've only picked out one here just to sort of highlight, but they're looking 246 00:16:21,990 --> 00:16:27,480 at 0, 1 and two lepton final states, assuming one of the vector bosons goes to 247 00:16:27,480 --> 00:16:32,040 q q and you can get another one of these large radius jets, and then taking a look 248 00:16:32,040 --> 00:16:35,340 at either the invariant mass distribution or the transverse mass depending on how 249 00:16:35,340 --> 00:16:39,330 much energy is lost to neutrinos, and searching in that distribution for 250 00:16:39,330 --> 00:16:43,860 resonance peaks. So you can see one such distribution up here. This is the W W 251 00:16:44,430 --> 00:16:49,320 with one lepton invariant mass distribution, assuming gluon-gluon 252 00:16:49,320 --> 00:16:54,570 fusion and selecting for that. And as a result, the limit on Z prime to W W, which 253 00:16:54,570 --> 00:16:59,730 is being set down here in the bottom plot. Okay, this brings me to the next section 254 00:16:59,730 --> 00:17:04,230 of this talk. We're going to move away now from that s channel, Zed prime 255 00:17:04,230 --> 00:17:07,320 simplified model, and we're going to have a brief interlude to talk about invisible 256 00:17:07,320 --> 00:17:12,390 decays of the Higgs. This is another form of simplified model connection 257 00:17:12,390 --> 00:17:16,530 between the standard model and dark matter, which is a really powerful 258 00:17:16,530 --> 00:17:20,850 benchmark for us to be checking out. So in this case, we're going to assume that we 259 00:17:20,850 --> 00:17:24,330 have one new dark matter particle and then it couples the Standard Model only via the 260 00:17:24,330 --> 00:17:29,460 Higgs. Now if we assume that we understand the total Higgs production cross section, 261 00:17:29,850 --> 00:17:34,650 it's natural to ask how much room is there for BSM decays within that. All of the 262 00:17:34,650 --> 00:17:38,310 numbers that you see here I've taken from some ATLAS documentation, which is in the 263 00:17:38,310 --> 00:17:42,570 link in the next slide. But roughly speaking, the combination of the observed 264 00:17:42,570 --> 00:17:47,370 Higgs decays that we have now sets an upper limit on the remaining possible 265 00:17:47,370 --> 00:17:51,540 branching fraction of the Higgs to invisible or BSM decays at around 30%. 266 00:17:52,260 --> 00:17:56,340 This is a larger number than I expected. My understanding is that this is because 267 00:17:56,370 --> 00:17:59,550 there's quite a bit of wiggle room in the theoretical production cross section, but 268 00:17:59,550 --> 00:17:59,910 maybe 269 00:18:00,659 --> 00:18:04,439 If anyone has more details on that, feel free to put it in after the talk. 270 00:18:04,620 --> 00:18:08,250 Regardless, this actually leaves us quite a bit of wiggle room for having some BSM 271 00:18:08,250 --> 00:18:12,930 Higgs decays, even though we've had as many measurements as we have on the Higgs. 272 00:18:13,110 --> 00:18:16,800 So the natural question asked is does this include invisible decays to dark matter? 273 00:18:17,280 --> 00:18:21,870 These are really challenging searches. The goal of them is to set an upper limit on 274 00:18:21,870 --> 00:18:25,470 the possible Higgs to invisible branching ratio. For comparison, the Standard Model, 275 00:18:25,470 --> 00:18:29,010 of course, predicts some Higgs to invisible decays, but on the order of point 1%, so 276 00:18:29,010 --> 00:18:32,400 we really don't have to worry about that. Now, here's the nice new result that's 277 00:18:32,400 --> 00:18:36,270 just come out of Atlas. This is looking for VBF production of a Higgs with the 278 00:18:36,270 --> 00:18:40,590 Higgs decaying to dark matter or other invisible particles. The topology that's 279 00:18:40,590 --> 00:18:44,250 useful for picking this out is that you have very forward jets. This helps you to 280 00:18:44,250 --> 00:18:48,390 reject just V plus jets production. There's an obvious remaining challenging 281 00:18:48,390 --> 00:18:53,430 background from the production of VBF Zed going to invisible. So the approach to 282 00:18:53,430 --> 00:18:57,900 this analysis is to trigger on the missing energy. Look for QCD dijets that are back 283 00:18:57,900 --> 00:19:03,270 to back in eta, but not in phi. There are multiple control regions defined which 284 00:19:03,300 --> 00:19:07,020 allow the constraint of the W and Z based backgrounds. You can see these along the 285 00:19:07,020 --> 00:19:12,300 side here. And then a data driven fit is done to predict the multijet background in 286 00:19:12,300 --> 00:19:15,990 these signal regions right here. And what you can see is this red line here 287 00:19:15,990 --> 00:19:19,410 represents the maximum amount of Higgs to invisible decay that we could be having 288 00:19:19,410 --> 00:19:24,360 without having observed it in the data. And this corresponds to a upper limit 289 00:19:24,390 --> 00:19:28,050 on the branching ratio of 13%, both observed and expected, which is a really 290 00:19:28,050 --> 00:19:31,020 nice improvement over what we've had before and well below what we can do by 291 00:19:31,020 --> 00:19:35,100 just looking at the visible channels. There's also this plot included which is a 292 00:19:35,100 --> 00:19:38,880 really nice comparison to the direct detection limits. In this case, an EFT 293 00:19:38,880 --> 00:19:42,630 framework documented at the link there is used to translate the results into limits 294 00:19:42,630 --> 00:19:45,990 on the wimp-nucleon cross section, which is what direct detection experiments tend 295 00:19:45,990 --> 00:19:50,040 to use. And this is valid essentially, because the relevant mass scale here is 296 00:19:50,070 --> 00:19:54,690 the Higgs which is quite a bit lighter than any sort of additional processes that 297 00:19:54,690 --> 00:19:58,200 would be happening in this model. You can also see there's two different lines which 298 00:19:58,200 --> 00:20:01,950 are being set as limits by this analysis which correspond to the 299 00:20:01,950 --> 00:20:06,390 differences in the relationship between Higgs to invisible branching fraction and 300 00:20:06,390 --> 00:20:09,570 this cross section, depending on whether your dark matter is scalar or fermion. 301 00:20:10,200 --> 00:20:13,170 And you can see that the limits end at half the mass of the Higgs as we would 302 00:20:13,170 --> 00:20:17,910 expect. Now, the last thing I want to talk about is to 2 HDMa models, and it's the 303 00:20:17,910 --> 00:20:21,810 dominant signatures that come with that. So if we move to a simplified model, where 304 00:20:21,810 --> 00:20:25,860 we think about a two Higgs doublet model, plus a pseudo scalar - again, this is just one of 305 00:20:25,860 --> 00:20:29,550 many options, but it's one that we've done a number of interpretations in - then like 306 00:20:29,550 --> 00:20:34,440 the Zed prime model, you can search in a met plus x signature, we have our typical 307 00:20:34,440 --> 00:20:38,250 monojet signature over here, or you can look for visible mediator decays, 308 00:20:39,090 --> 00:20:43,830 something like this. But these, you'll note, all of these situations here prioritize 309 00:20:43,830 --> 00:20:47,670 third generation couplings and signatures which have vector bosons and Higgses. So 310 00:20:47,670 --> 00:20:52,470 this is a bit different than the Z prime situation. Here we have TT bar, BB bar plus 311 00:20:52,470 --> 00:20:57,120 met or all heavy flavor. And over here we have mono Z H signatures which have a 312 00:20:57,120 --> 00:21:01,710 special role. So the first thing I want to talk about here is the heavy flavor plus 313 00:21:01,710 --> 00:21:05,100 missing energy analysis. So this would be this Feynamn diagram here where you have a 314 00:21:05,130 --> 00:21:09,900 TT bar and then invisible decay of this mediator. So the thing to do here is to 315 00:21:09,900 --> 00:21:13,680 search for Dark Matter basically looking for two tops and high missing momentum 316 00:21:13,680 --> 00:21:19,740 final state. In this case, this nice new analysis from Atlas selects a one lepton 317 00:21:19,740 --> 00:21:23,160 decay to sort of balance high stats with suppression of QCD and then requires 318 00:21:23,160 --> 00:21:29,250 at least two b tags to keep the top-like events. The signal region then is defined 319 00:21:29,250 --> 00:21:33,240 here optimized for dark matter and you can see a multi-bin fit is performed in this 320 00:21:33,240 --> 00:21:37,080 distribution, which is the angular, angular difference between the missing ET and the 321 00:21:37,080 --> 00:21:43,170 lepton. You can see unsurprisingly, this is dominated by ttZ where the Z decays 322 00:21:43,170 --> 00:21:47,760 invisibly and that is in this case constrained with a dedicated lepton-based 323 00:21:47,760 --> 00:21:51,840 control region. You can see the nice exclusion plot which comes up here. So 324 00:21:51,840 --> 00:21:57,720 this is a very new and very cool analysis. We have now mono V or Higgs which as we 325 00:21:57,720 --> 00:22:01,710 pointed out earlier also has a special sensitivity to these 2 HDM plus X models. 326 00:22:02,250 --> 00:22:07,200 This is a CMS result which took a bunch of existing limits and added in a W W and Zed 327 00:22:07,200 --> 00:22:12,300 Zed pair of channels for this. So you can see here this plot is for the Higss to W W, 328 00:22:12,300 --> 00:22:16,050 where they actually trained a BDT on the momenta, transverse masses, angular 329 00:22:16,050 --> 00:22:21,750 variables and so on. And you can see this nice distinguishing between the signal and 330 00:22:21,750 --> 00:22:25,260 the background. For the Zed Zed, it was just a typical missing energy 331 00:22:25,290 --> 00:22:29,130 distribution. But the combination of the two has gone into this really nice limit plot 332 00:22:29,130 --> 00:22:32,640 here where you can see the the dominant channel across all of this is still really 333 00:22:32,640 --> 00:22:36,540 Higgs to bb bar, but that the combination has actually pushed down and has set 334 00:22:36,810 --> 00:22:39,210 exclusions across a very wide range. This is really nice to see. 335 00:22:40,560 --> 00:22:44,070 You can also of course, look for physical mediator decays. So if you assume that the 336 00:22:44,070 --> 00:22:50,520 A can decay visibly, if it's very light, you would actually reconstruct even a, an A 337 00:22:50,760 --> 00:22:55,710 decay to two gluons as a single small radius jet. So this cool Atlas analysis 338 00:22:55,740 --> 00:23:02,100 leans quite heavily on a machine learning based estimation of this. So the first 339 00:23:02,100 --> 00:23:08,010 step then was to train on a single small radius jet in signal to look for that two 340 00:23:08,010 --> 00:23:12,600 pronged substructure and learn the resonance mass from that. A second step then took 341 00:23:12,600 --> 00:23:16,560 that learned mass and various substructure variables and trained a selection neural 342 00:23:16,560 --> 00:23:21,600 net. And then for events which pass the selection neural net, you take your Zed to LL 343 00:23:21,600 --> 00:23:26,550 candidate, and this single, small radius jet, you make the invariant mass of those 344 00:23:26,550 --> 00:23:29,850 three things. And if you have a new resonance, it should add up to the mass of 345 00:23:29,850 --> 00:23:33,270 the Higgs. And so this little window around the Higgs mass is then used as a 346 00:23:33,630 --> 00:23:39,090 signal region and searched for any excess. Okay, this brings me to the end of the 347 00:23:39,090 --> 00:23:42,720 specific analyses that I wanted to show today. But I just want to remind you how 348 00:23:42,720 --> 00:23:47,070 many more things there are that we do that are also Dark Matter searches, even if we 349 00:23:47,070 --> 00:23:50,790 don't cover them in a talk like this. So this is just a reminder that 350 00:23:50,790 --> 00:23:55,710 supersymmetry can also be a dark matter model. When we look for R-parity 351 00:23:55,710 --> 00:23:59,970 conserving SUSY, what this means is that we require there's an even number of SUSY 352 00:24:00,000 --> 00:24:04,560 particles in each interaction. And that in turn means that the lightest SUSY particle 353 00:24:04,560 --> 00:24:05,550 has to be stable. 354 00:24:05,970 --> 00:24:07,740 And this makes it a good Dark Matter candidate. 355 00:24:08,460 --> 00:24:12,300 Depending on various other parameters that you tune. The quality of the Dark Matter 356 00:24:12,300 --> 00:24:15,180 candidate will depend on your other assumptions, because various parameter 357 00:24:15,180 --> 00:24:18,900 choices can give you something with too high relic density and so on. But this 358 00:24:18,900 --> 00:24:23,100 is one of the powerful, more complicated models that we should be thinking of as 359 00:24:23,100 --> 00:24:26,370 part of our dark matter search program. And so if you want more details on this, 360 00:24:26,400 --> 00:24:31,110 I've put a link to one of the other talks from LHCP right here, go and check it out. I want 361 00:24:31,110 --> 00:24:34,260 just to use my last couple of minutes to talk about putting all of these together 362 00:24:34,260 --> 00:24:38,040 into a single big picture. It's really hard to keep track of all of these 363 00:24:38,040 --> 00:24:41,490 scattered analyses, if we don't have these simplified models that let us put them all 364 00:24:41,490 --> 00:24:46,230 in a single plot and take a look at how they all interrelate. So this is the first 365 00:24:46,230 --> 00:24:51,180 plot that I want to show for this. This is taking all of our dijet-like resonance 366 00:24:51,180 --> 00:24:55,470 searches and putting them onto a plot where we have a coupling to quarks on the 367 00:24:55,470 --> 00:24:56,160 y axis, we 368 00:24:56,160 --> 00:24:56,940 have the mass 369 00:24:57,030 --> 00:25:01,260 of the mediator that we're searching for on the x axis, and everything sort of 370 00:25:01,260 --> 00:25:05,760 above the solid lines here is something that we've excluded. This has been nicely 371 00:25:05,760 --> 00:25:09,120 updated. I think this is the most fresh off the press result I'm showing today 372 00:25:09,120 --> 00:25:14,160 since we just made this public yesterday. But you can see here the new limits from 373 00:25:14,160 --> 00:25:18,480 the full Run 2 dijet and di-b and how those are really strongly constraining us 374 00:25:18,480 --> 00:25:23,370 down towards this low coupling and high mass corner. The other thing that we can do, of 375 00:25:23,370 --> 00:25:29,790 course, is put this more clearly into the rest of the Dark Matter picture for these Zed prime 376 00:25:29,970 --> 00:25:33,510 models. And these are some summary plots you'll've seen a lot probably from Atlas 377 00:25:33,510 --> 00:25:40,440 and CMS where we have in this case, dark matter mass on the y axis, mediator mass 378 00:25:40,440 --> 00:25:44,700 again on the x axis, and we have to specify all of the couplings rather than setting 379 00:25:44,700 --> 00:25:49,650 limits on those directly. What you can see here is Mono X limits from things like 380 00:25:49,650 --> 00:25:53,550 mono jet fill in this corner, and then you have pretty much independence of the 381 00:25:53,550 --> 00:25:57,660 Dark Matter mass when you look at the resonance searches. This is also new this 382 00:25:57,660 --> 00:26:01,800 week from CMS and includes now the full Run 2 dijet analysis that we just looked 383 00:26:01,800 --> 00:26:05,910 at earlier, and it was so nice. I highlight the couplings here because I 384 00:26:05,910 --> 00:26:09,090 want to point out that if you change them, you get a completely different picture. 385 00:26:09,210 --> 00:26:12,660 And so it's very important to keep in mind what you're including when making this 386 00:26:12,660 --> 00:26:16,980 kind of comparison. This is my last slide of content. This is just reminding you all 387 00:26:16,980 --> 00:26:20,250 that we should also be paying close attention to putting all of this in 388 00:26:20,250 --> 00:26:25,320 context with the non-collider world. It's important that we also keep track of how 389 00:26:25,320 --> 00:26:30,090 our results interplay with the direct and indirect detection results from other 390 00:26:30,090 --> 00:26:34,230 experiments. I'm showing here just two versions of this. One of these is limits 391 00:26:34,230 --> 00:26:38,190 using an axial vector Zed prime with couplings specified here, you can see this 392 00:26:38,190 --> 00:26:43,560 is a spin dependent limit. On this side, we have a spin independent Dark Matter 393 00:26:43,560 --> 00:26:48,210 nucleon interaction cross section limit for a vector mediator, you can see how 394 00:26:48,210 --> 00:26:51,930 very different these two pictures are once again, where we have power, where we don't. 395 00:26:52,380 --> 00:26:56,190 But making these plots and continuing to try to introduce these is something that we 396 00:26:56,190 --> 00:27:00,540 want to do to make sure we keep up as active a conversation as possible. This 397 00:27:00,540 --> 00:27:04,320 brings me to my conclusions. Sorry, I realize I'm a minute or two over. We 398 00:27:04,320 --> 00:27:08,460 explored the wide range of analyses here today, which can constrain dark matter at 399 00:27:08,460 --> 00:27:12,480 the LHC and tried to highlight the unique contributions of each of them. This 400 00:27:12,630 --> 00:27:15,990 included in this talk Higgs to invisible searches, resonance searches and missing 401 00:27:15,990 --> 00:27:19,530 energy based searches, which are all key contributors to our dark matter search 402 00:27:19,530 --> 00:27:23,760 program. There's more full Run 2 results in progress now, so stay tuned for 403 00:27:23,760 --> 00:27:26,790 some exciting updates. I know we have a few things aiming for ICHEP, so should be 404 00:27:26,790 --> 00:27:31,290 fun. And as a final reminder, our dark matter possibilities are really super 405 00:27:31,290 --> 00:27:34,620 diverse. There's a lot of broader Dark Sector searches, there's Susy searches, there's 406 00:27:34,620 --> 00:27:37,830 long lived particle searches, all of these can be handles on dark matter. So I've put 407 00:27:37,830 --> 00:27:41,880 down here a collection of talks around this week that you can go back and look at. 408 00:27:41,910 --> 00:27:43,980 There's still more, but it's a place to start. 409 00:27:44,549 --> 00:27:45,299 Thanks a lot. 410 00:27:46,980 --> 00:27:49,740 Thank you very much for the nice overview talk. It wasn't, ah, it was almost on time. 411 00:27:53,820 --> 00:27:59,850 It was closer to on time than not. So do we have any questions for the talk? 412 00:28:11,040 --> 00:28:16,080 At the moment it doesn't look like there's any audience questions. So 413 00:28:17,789 --> 00:28:18,839 maybe I will ask one? 414 00:28:21,690 --> 00:28:28,680 So where would you see the most exciting Run Three developments that could be 415 00:28:28,680 --> 00:28:30,540 coming up in this kind of analysis? 416 00:28:32,220 --> 00:28:33,360 Could be coming up? 417 00:28:36,120 --> 00:28:40,020 I guess it's just a personal question. It's not like there's a right or wrong 418 00:28:40,020 --> 00:28:44,370 answer anywhere else. But what woud you aim for 419 00:28:45,090 --> 00:28:50,160 in Run Three? I'd be interested to see us... I'd be interested to see us extending what 420 00:28:50,160 --> 00:28:53,520 we're doing with trigger level analyses. I know this is piggybacking a little bit on 421 00:28:53,520 --> 00:28:56,340 what Simon was just talking about, but I really do think we've only begun to 422 00:28:56,340 --> 00:28:59,790 scratch the surface there of what we could possibly do and that at the moment, we're 423 00:28:59,790 --> 00:29:03,150 being a little bit limited by the number of objects that we're actually using there. 424 00:29:03,390 --> 00:29:07,830 Jets and now newly muons from CMS, but we could be thinking more about, if we are 425 00:29:07,860 --> 00:29:13,440 able to, what other analyses like this could benefit from that approach to have a 426 00:29:13,440 --> 00:29:17,370 better look at the low mass limits. The other thing I think we need to do is think 427 00:29:17,370 --> 00:29:21,300 more about the simplified models that we're interested in using for our 428 00:29:21,300 --> 00:29:24,750 summaries. It's too easy to make a plot like this, and look at it and see no holes 429 00:29:24,750 --> 00:29:29,850 and say that we were done. And so I think we need to be pushing ourselves to examine 430 00:29:29,850 --> 00:29:35,610 more thoroughly, more realistic scenarios where we're making more clear to ourselves 431 00:29:35,610 --> 00:29:39,660 our limitations, and that should help us also to figure out what we need to focus 432 00:29:39,660 --> 00:29:39,870 on. 433 00:29:42,420 --> 00:29:42,990 Thank you.