1
00:00:01,560 --> 00:00:08,280
So I think we can start with the
presentation. So the first speaker is iden

2
00:00:08,280 --> 00:00:14,970
groomer from Atlassian. Who will present
electronically the case of light states

3
00:00:14,970 --> 00:00:22,200
and the rate for please item, you already
share the screen. Thanks for speaking. Hi,

4
00:00:22,260 --> 00:00:23,460
I'm here. Yeah.

5
00:00:24,810 --> 00:00:26,520
Right. Yes, thanks for the introduction.

6
00:00:26,580 --> 00:00:34,230
I'll be presenting results from both Atlas
and CMS today on the B SMU and the Tao

7
00:00:34,230 --> 00:00:42,930
two, three new results. So this will cover
both CMS and Atlas. Like I was saying

8
00:00:43,260 --> 00:00:47,760
they're very similar detectors. They do
have some differences though. The the

9
00:00:47,760 --> 00:00:52,200
inner diameter on CMS is four centimeters
with 80 million readout channels for the

10
00:00:52,200 --> 00:00:57,870
inner tracker, and the Atlas is 3.3 with
100 million, but maybe most significant

11
00:00:57,870 --> 00:01:05,940
for this talk is the The difference in the
magnetic field of 3.8 Tesla for CMS and to

12
00:01:05,940 --> 00:01:11,880
Tesla for for the ATLAS detector. So to
jump right into the BS to me of the

13
00:01:11,880 --> 00:01:19,980
result, the Atlas result can be found at
this link here, and it's performed on 26.3

14
00:01:20,310 --> 00:01:27,900
inverse femto barns effectively at 13 TV
in the 2015 and 2016 data as well as the

15
00:01:27,930 --> 00:01:35,700
25 inverse of two barns from the run one
data and CMS also here is shown with the

16
00:01:35,700 --> 00:01:42,540
link to their paper here, and 36 inverse
femto barns are used in 2016 data

17
00:01:43,020 --> 00:01:48,540
separated into two categories, just based
on operational instabilities experienced

18
00:01:48,540 --> 00:01:54,870
with the microstrip detector in CMS. So,
that's separated into 2016 A and 2016 B

19
00:01:55,440 --> 00:02:01,890
and then the 25 interoceptive burns from
run one are also Included in this CMS

20
00:02:01,890 --> 00:02:11,160
results, and I'll discuss that further. So
the motivation for looking at BST mew is

21
00:02:11,190 --> 00:02:17,160
primarily the smallness and precision of
the predictive branching fractions. And

22
00:02:17,340 --> 00:02:20,970
you can see the Standard Model predictions
here on the slide for the branching

23
00:02:20,970 --> 00:02:26,700
fractions vs mew and B to review at orders
of 10 to the minus nine and 10 to the

24
00:02:26,700 --> 00:02:33,060
minus 10 respectively. We can also probe
the standard model in measuring the

25
00:02:33,060 --> 00:02:38,850
effective lifetime. So CMS has a nice
result on the effective lifetime of the

26
00:02:38,940 --> 00:02:45,270
BST mew and the experimental world average
for the heavy state heavy mass Eigen

27
00:02:45,270 --> 00:02:52,920
state. lifetime of BS is what is shown on
the slide. any significant deviations from

28
00:02:52,920 --> 00:02:59,220
the standard model would would indicate
new physics. So, these could occur in in

29
00:02:59,220 --> 00:03:04,470
models such as MSM more minimal flavor
violation. And a couple of Fineman

30
00:03:04,470 --> 00:03:10,410
diagrams of these if these types of models
are shown on this slide as well. The

31
00:03:10,410 --> 00:03:15,240
branching fraction is measured using the
master equation here on the slide, you

32
00:03:15,240 --> 00:03:24,000
have the yield of B sub d a b or b sub s.
And, and its efficiencies, the event

33
00:03:24,000 --> 00:03:28,500
efficiencies times the acceptance and then
this is measured with respect to the

34
00:03:28,500 --> 00:03:34,110
normalization channel, B plus two j cy k
plus, which is abundant and has a very

35
00:03:34,110 --> 00:03:39,180
well measured branching fraction. And so
we also then have the hadron ization

36
00:03:39,180 --> 00:03:45,870
fraction in the master equation as well.
The analysis for both CMS and Atlas are

37
00:03:45,870 --> 00:03:51,840
performed as blind analyses where we
conceal the signal regions and then set

38
00:03:51,840 --> 00:03:57,420
all of the procedures for performing the
analysis and extracting the signal until

39
00:03:57,450 --> 00:04:01,290
until those are set and then and then
unblind the data You can see the blinding

40
00:04:01,290 --> 00:04:07,620
regions here are in dotted lines with the
signal Monte Carlo for B sub s as well as

41
00:04:07,620 --> 00:04:14,760
be to me view, Monte Carlos, and those are
fit with double Gaussians does the signal

42
00:04:14,760 --> 00:04:19,260
regions are defined slightly differently
between Atlas and CMS but but it's a it's

43
00:04:19,260 --> 00:04:24,000
a similar idea. And Monte Carlo
simulations are generated for the signals

44
00:04:24,000 --> 00:04:28,950
as well as the background regions. And
then there's also a Monte Carlo generated

45
00:04:28,950 --> 00:04:32,310
for the P plus two j sy k candidates
reference channel.

46
00:04:33,780 --> 00:04:38,640
The background is composed of common
tutorial background, which is a dominant

47
00:04:39,810 --> 00:04:44,790
common story of background. The it
consists of muons from uncorrelated hadron

48
00:04:44,790 --> 00:04:48,960
decays, and it's characterized by not
having much dependence on the debut on

49
00:04:48,960 --> 00:04:54,810
invariant mass. A BDT is used to separate
the continuum background for both CMS and

50
00:04:54,810 --> 00:05:00,840
Atlas. The BTT discriminator values are
shown on this slide. indicated the

51
00:05:00,840 --> 00:05:05,460
boundaries are indicated with the arrows
here. So, you have the sideband data

52
00:05:05,550 --> 00:05:12,330
compared to the signal MonteCarlo a value
greater than closer to one indicates more

53
00:05:12,330 --> 00:05:19,710
signal like the partially reconstructed
the Ks contribute as as a background here

54
00:05:19,710 --> 00:05:24,750
in the analysis, you can see them in this
a lot of information here on this slide

55
00:05:24,750 --> 00:05:30,360
for the partially reconstructed decays
these have mostly accumulate in the low

56
00:05:30,360 --> 00:05:38,220
sideband region of the of the analysis.
And the peaking background is also shown

57
00:05:38,220 --> 00:05:44,490
here in the middle plot. You have the
hadrons misidentified as nuanced in the

58
00:05:44,730 --> 00:05:51,150
background. And so, these are these are
simulated with Monte Carlo as well. The

59
00:05:51,150 --> 00:05:56,430
reference channel extraction is shown here
on the slide. This is B plus two j cy k

60
00:05:56,430 --> 00:06:05,760
plus for both CMS and Atlas and The the
extractions are shown here. So, the

61
00:06:06,420 --> 00:06:12,780
backgrounds that contribute to the fit of
the of the signal here are the P plus two

62
00:06:12,780 --> 00:06:16,950
j cy k plus decays as the signal that can
be that can be both suppressed to be plus

63
00:06:16,950 --> 00:06:22,020
two j Sai pi plus the partially
reconstructed decays and the continuing

64
00:06:22,020 --> 00:06:31,050
background as well. And in CMS, the data
is split between the 2016 A and 2016 b but

65
00:06:31,050 --> 00:06:36,510
it's also further split into pseudo
rapidity regions of the most forward new

66
00:06:36,510 --> 00:06:41,460
on the yields for Atlas and CMS are also
shown on this slide, but this is just

67
00:06:41,610 --> 00:06:47,790
showing the reference channel yields. And
then finally, the signal extraction and

68
00:06:47,790 --> 00:06:53,280
results are presented on this slide. The
Standard Model expectation for atlases 91

69
00:06:54,120 --> 00:07:01,830
events for N sub s, the B sub s yield and
N sub D is taking And then compared to to

70
00:07:01,830 --> 00:07:09,180
the extraction of the Atlas results 80 and
minus 12 for density. CMS is also

71
00:07:09,180 --> 00:07:14,490
determined for each each BDT bin and data
subset category that I explained in the

72
00:07:14,490 --> 00:07:20,820
previous slide. And and their results for
the yield are presented here as well. And

73
00:07:20,880 --> 00:07:25,770
so, this has shown for Atlas for the
highest vdt been and for CMS for all bins

74
00:07:26,640 --> 00:07:36,450
in in all data subset categories including
one the branching fractions are compared

75
00:07:36,450 --> 00:07:40,410
here. So you have the BW on the vertical
axis and the vs to medium on the

76
00:07:40,410 --> 00:07:48,000
horizontal axis and, and the confidence
level lines are shown in the in the curves

77
00:07:48,240 --> 00:07:54,540
on the plot as well. The central point is
shown in black for the combination of run

78
00:07:54,540 --> 00:07:59,970
one and run two and compare it to the
standard model and read in both in both

79
00:08:00,000 --> 00:08:07,170
Results there. So, the the boxes show the
final results of the Bs dimeo measurements

80
00:08:07,350 --> 00:08:14,940
are shown in the upper lines and the the
upper limits on the branching fractions of

81
00:08:14,940 --> 00:08:21,720
the two new are also shown here. For
Atlas, the name on construction is used

82
00:08:21,720 --> 00:08:24,630
for, for performing the comparison of the

83
00:08:26,580 --> 00:08:29,610
for performing the results of the printing
fraction limit.

84
00:08:31,200 --> 00:08:35,760
The lifetime measurement from CMS was
presented in their paper as well. So a two

85
00:08:35,760 --> 00:08:41,100
dimensional unbid maximum likelihood fit
is performed on both the decay time the

86
00:08:41,100 --> 00:08:47,940
proper decay time and the diamond on mass.
And this fit includes the signal and each

87
00:08:47,940 --> 00:08:52,260
background component, as you can find on
the slides here that combinatorially

88
00:08:52,260 --> 00:08:58,860
peaking and semi metallic decay again. So
the tau two Mumia results is presented

89
00:08:58,860 --> 00:09:05,130
here 1.7 and And you can compare it to the
experimental world average of the heavy

90
00:09:05,400 --> 00:09:09,300
mass I can state lifetime 1.6 time

91
00:09:10,950 --> 00:09:11,910
picoseconds

92
00:09:13,620 --> 00:09:21,990
I'm going to shift over to tau two three
mew now. And so for Atlas the paper you

93
00:09:21,990 --> 00:09:28,620
can find here in the journal and this is
performed with 20.3 inverse femto barns of

94
00:09:29,220 --> 00:09:35,340
at the center of mass energy ATV. And the
CMS results is also presented today and

95
00:09:35,370 --> 00:09:44,130
this is what 33 inverse femto burns 13 TV
data. motivation for performing this

96
00:09:44,130 --> 00:09:50,160
analysis is the is that if we were able to
see charged leptons flavor violation, it

97
00:09:50,160 --> 00:09:55,380
would be a major breakthrough in our
understanding, and the branching fraction

98
00:09:55,410 --> 00:10:00,450
is expected to be very small, by the
standard model. You can see it's Let's

99
00:10:00,450 --> 00:10:05,670
suppose it's predicted to be less than 10
to the minus 14, but some extensions to

100
00:10:05,670 --> 00:10:09,840
the Standard Model lead to branching
fractions many orders of magnitude greater

101
00:10:09,840 --> 00:10:16,920
than this and and could be in within reach
of experimental confirmation. And so,

102
00:10:17,130 --> 00:10:21,330
these these branching fractions could be
held up to 10 to the minus 10 or 10 to the

103
00:10:21,330 --> 00:10:27,750
minus eight. And a couple of Fineman
diagrams and models beyond the standard

104
00:10:27,750 --> 00:10:36,480
model are shown or shown here. Atlas
analysis uses a source of left tailed

105
00:10:36,480 --> 00:10:43,380
leptons from decaying from a W boson to a
towel and a neutrino and the town neutrino

106
00:10:43,380 --> 00:10:48,300
from the W bosonic. appears in the
analysis as a missing transverse energy.

107
00:10:48,630 --> 00:10:54,390
You can see the diagram here on the right
where the W goes on is produced in the

108
00:10:54,690 --> 00:11:00,960
primary vertex and subsequently decays to
a townie hotel and neutrino which is

109
00:11:00,960 --> 00:11:07,920
picked up as the missing inverse energy,
that missing transverse energy. And the

110
00:11:08,040 --> 00:11:13,050
nuance subsequently that when a tau decays
to three neurons, this is characterized by

111
00:11:13,470 --> 00:11:20,670
also having a small opening angle cone
around those three neurons. The branching

112
00:11:20,670 --> 00:11:28,470
fraction is calculated with respect to the
WW yield and here we have the acceptance

113
00:11:28,470 --> 00:11:34,860
and the efficiencies. So, this is the
master equation for this analysis. Atlas

114
00:11:35,340 --> 00:11:42,090
picks up on pre selects events three with
three new on vertex vertices and our a

115
00:11:42,090 --> 00:11:47,220
three neon vertex and it and these events
are required to meet blue selection

116
00:11:47,220 --> 00:11:53,550
criteria. These selections include the
three neon mask vertex fit, and or the

117
00:11:53,550 --> 00:11:58,560
three neon vertex fit as well as the
missing energy selections. The full

118
00:11:58,560 --> 00:12:03,720
selection lists as can be found In the
background and the backup slides, a BDT

119
00:12:03,720 --> 00:12:08,820
discriminator is trained on a signal Monte
Carlo and the background events in the BDT

120
00:12:08,820 --> 00:12:13,230
training region, you can see the feed at
training regions in this in this bullet

121
00:12:13,230 --> 00:12:18,450
here, and a loose cut on the BDT is
applied where x zero is set equal to minus

122
00:12:18,450 --> 00:12:24,180
0.9, where BDT runs between minus one and
one and this just removes a very

123
00:12:24,180 --> 00:12:28,710
background like events. Finally, tight
selection criteria are applied on top of

124
00:12:28,710 --> 00:12:34,770
this. So, tight selection and the loose
PDT cut are applied to look at the

125
00:12:34,770 --> 00:12:39,000
analysis variables. So, you see the
analysis variables here for the transverse

126
00:12:39,060 --> 00:12:44,940
missing energy as picked up by the tracker
and the significance of the three neon

127
00:12:44,940 --> 00:12:45,630
vertex that

128
00:12:47,010 --> 00:12:48,900
usually means Thank you.

129
00:12:51,210 --> 00:12:55,860
So the results are shown here on this
slide you have the BDT distribution, the

130
00:12:55,860 --> 00:13:00,960
optimal BDT cut is found based on
optimizing the x spected upper limit of

131
00:13:00,960 --> 00:13:07,590
the branching fraction that's set to 9.0
point 933. You can see the second Monte

132
00:13:07,590 --> 00:13:15,030
Carlo difference between the background as
well here and the three Mian mass

133
00:13:15,030 --> 00:13:19,710
distribution is shown on this slide. You
have the loose selection criteria the

134
00:13:19,710 --> 00:13:25,260
loose PDT cut and the tight selection
criteria data shown in the black points

135
00:13:25,500 --> 00:13:33,600
and one event passes the tight selection
criteria plus the tight BDT cut a fit to

136
00:13:33,600 --> 00:13:39,330
the BDT. The sideband data is also shown
here just to show you what the background

137
00:13:39,330 --> 00:13:45,630
looks like in the signal region. And an
observed upper limit on this branching

138
00:13:45,630 --> 00:13:53,550
fraction is 3.76 times 10 to the minus
seven and a 90% confidence level. The CMS

139
00:13:53,550 --> 00:14:00,600
results is shown on this slide the
production of the of the center This is

140
00:14:00,600 --> 00:14:07,590
from D marathons two to town neutrino as
well as B. So, the production of these is

141
00:14:07,590 --> 00:14:13,290
shown on these bullets here. The events
are selected based on two neurons in a

142
00:14:13,290 --> 00:14:18,780
track and the branching fractions are
measured with respect to the normalization

143
00:14:18,780 --> 00:14:25,020
channel D to phi pi, the data are
separated based on mass resolution

144
00:14:25,020 --> 00:14:29,340
categories. So, you have three nice
resolution categories and a BDT is trained

145
00:14:29,340 --> 00:14:35,100
in each each one of the categories and,
and you can see the BDT output in this

146
00:14:35,100 --> 00:14:40,350
slide here. The BDT regions are optimized
based on the search sensitivity and signal

147
00:14:40,350 --> 00:14:48,600
extraction. And And so, the top two bins
are used for for signal extraction. The

148
00:14:48,600 --> 00:14:52,560
mass resolution distributions are shown
here you have the three different

149
00:14:52,560 --> 00:14:58,290
categories for each of the mass
resolutions. The signal Monte Carlo, the

150
00:14:58,290 --> 00:15:04,110
signal here shown in red is normalized as
if the branching fraction were set to be

151
00:15:04,320 --> 00:15:11,280
minus 10 to the minus seven. And the
yields are shown in the signal and data

152
00:15:11,310 --> 00:15:17,490
categories for the full mass range, as
well as the signal range in parentheses.

153
00:15:18,150 --> 00:15:24,180
The upper limit for the cmas results is
8.8 times 10 to the minus eight at a 90%

154
00:15:24,180 --> 00:15:32,370
confidence level. And so this is a very
nice result as well. So the summary of the

155
00:15:32,370 --> 00:15:36,840
results are shown here you have the
branching fraction measurements of the BS

156
00:15:36,840 --> 00:15:43,350
to mew as well as the upper limits of the
bmu from both Atlas and CMS Atlas in blue

157
00:15:43,350 --> 00:15:50,130
and CMS and read. The effective lifetime
measurement measured by CMS is shown here.

158
00:15:50,640 --> 00:15:55,860
And the upper limits on the Tao the three
new branching fractions from from both

159
00:15:55,860 --> 00:15:57,330
Atlas and CMS results are

160
00:15:57,780 --> 00:16:01,080
are nicely done and performed here.

161
00:16:02,520 --> 00:16:06,090
So thanks very much for your attention.
Please, please let me know if you have

162
00:16:06,090 --> 00:16:06,690
questions.

163
00:16:06,960 --> 00:16:11,220
Thank you very much for this very nice
presentation and also for respecting the

164
00:16:11,220 --> 00:16:17,850
time contracts. So question, you should
raise your hand using the bottom in the

165
00:16:17,850 --> 00:16:28,860
bottom part of the screen. I have a
question to start this time the discussion

166
00:16:29,280 --> 00:16:39,810
on about the B zero result on page number
B nine. You can go back to your 1999 huh?

167
00:16:40,050 --> 00:16:48,630
Yes, yeah, from the plus side. So I have a
difficulty with this all you have both.

168
00:16:49,710 --> 00:16:55,140
Either on one or two, you have a central
radio, let's say that is negative. So

169
00:16:55,200 --> 00:16:59,970
he's, for sure. Of course he's not
physical. It's saying not so I can

170
00:17:00,000 --> 00:17:06,150
Question that are related to this fact.
So, the first thing is if you study some

171
00:17:06,150 --> 00:17:12,210
possible sources of these effects that can
be maybe related to background subtraction

172
00:17:12,210 --> 00:17:17,850
something like that. And the second
question is, how did you manage these,

173
00:17:18,930 --> 00:17:26,160
this effect a when you go to the limit
says you start from a number that cannot

174
00:17:26,160 --> 00:17:26,640
be

175
00:17:27,990 --> 00:17:30,540
say should be a fluctuation cannot be

176
00:17:32,160 --> 00:17:33,450
the same from numbers.

177
00:17:34,530 --> 00:17:39,450
So, I can I can say that Atlas is fully
aware that it's an unphysical to have a

178
00:17:39,510 --> 00:17:45,630
negative branching fraction, but it's but
it is certainly possible to have this

179
00:17:45,630 --> 00:17:51,750
statistical this particular statistical
value can take on negative values. And so

180
00:17:51,750 --> 00:17:57,750
the reason that the name on construction
is implemented to perform to show the

181
00:17:57,780 --> 00:18:02,790
upper limit on the branching fraction
Exactly for this. So it's handled by

182
00:18:02,820 --> 00:18:06,810
instead of just doing likelihood contours.
We apply the name on

183
00:18:07,800 --> 00:18:10,170
bands in this analysis.

184
00:18:11,429 --> 00:18:16,139
Thank you. I don't know. And the other
thing, do you think that can be some

185
00:18:16,499 --> 00:18:21,209
background that is not completely
described since you have the same

186
00:18:21,269 --> 00:18:26,129
negative, not the same, but you have the
negative values both for the wrong one

187
00:18:26,129 --> 00:18:32,219
around to try to understand why this
isn't? Why do you think?

188
00:18:34,260 --> 00:18:37,620
So? I mean, I,

189
00:18:38,010 --> 00:18:44,730
I think that the negative value is just a
function of the statistic itself. So it's

190
00:18:44,730 --> 00:18:48,180
a low, low branching fraction compared to
BS. And

191
00:18:49,740 --> 00:18:54,510
I think, with more statistics, we'll be
able to probe this further.

192
00:18:56,220 --> 00:19:01,830
Thank you very much. So are there any
other question? Try to either end you have

193
00:19:01,860 --> 00:19:07,140
time just for one very short. Otherwise
you can move. Oh, yes.

194
00:19:07,800 --> 00:19:08,820
There is a question. Yes.

195
00:19:08,820 --> 00:19:13,560
Okay. There is one. Ah, Gregor, try to try
to speak.

196
00:19:18,449 --> 00:19:23,519
Quick question. Is there any reason you're
not measuring the effective lifetime in

197
00:19:23,519 --> 00:19:28,589
Atlas, you should be able to do this
similar precision down CMS.

198
00:19:31,230 --> 00:19:34,230
I think that there's no reason Atlas can't
do this.

199
00:19:35,910 --> 00:19:39,540
I think we got triggered. I think it's a
nice day like 10 cuts.

200
00:19:42,120 --> 00:19:42,780
I

201
00:19:47,640 --> 00:19:48,720
sorry, I mean,