1
00:00:00,810 --> 00:00:08,700
Yes. Hi. It's a pleasure to
talk to you about the latest results from

2
00:00:08,700 --> 00:00:16,410
last March actually, on electroweak penguin
decays from LHCb. So on slide 2,

3
00:00:16,440 --> 00:00:20,820
I thought it would be nice to just include
a small picture of myself with this remote

4
00:00:20,820 --> 00:00:24,690
setup that we have. So this is me trying
to teach quantum mechanics at the

5
00:00:24,720 --> 00:00:30,750
university. And so I was hoping to stand
in front of you like this, maybe with a

6
00:00:30,750 --> 00:00:38,100
nicer shirt, but it is what it is.
So my outline is that I'll start with a

7
00:00:38,220 --> 00:00:42,870
brief introduction to electroweak penguins,
and detector and angular decompositions

8
00:00:43,140 --> 00:00:47,730
because those are the ingredients that we
need to explain these two new new papers

9
00:00:47,730 --> 00:00:54,540
that came out in March from LHCb, on the B
to K*mumu angular analysis update and

10
00:00:54,540 --> 00:01:05,580
the limits on the B to ee branching ratio.
On slide four, what would be considered

11
00:01:05,610 --> 00:01:10,920
as electroweak penguins are the B to sll
transitions. So I'm not talking about

12
00:01:11,100 --> 00:01:15,270
where the gamma is on shell, I'm talking
about the diagram shown on the right here.

13
00:01:16,590 --> 00:01:21,900
These are flavour changing and neutral. So
these are forbidden at tree level in the

14
00:01:21,900 --> 00:01:27,900
Standard Model. So they are very
suppressed, and called rare decays. And

15
00:01:27,900 --> 00:01:32,070
because of that, sensitive to new
physics in the loops, or even at tree

16
00:01:32,070 --> 00:01:38,220
level, if that's the form that new physics
takes. We can categorise these penguins

17
00:01:38,250 --> 00:01:44,250
in three classes: the ratios of different
flavours of leptons, which we call R(X),

18
00:01:44,400 --> 00:01:49,830
but that's the subject of the talk by Miriam,
which is right after mine. And the

19
00:01:49,830 --> 00:01:54,450
measurements of branching fractions and
angular observables. We measure these

20
00:01:54,450 --> 00:01:58,410
as a function of the Q-squared which is
the invariant mass of, in this case, the

21
00:01:58,410 --> 00:02:03,240
dilepton pair. Because as you can see on
the bottom plot as a function of Q-

22
00:02:03,240 --> 00:02:06,600
squared, there's a lot of different
physics going on depending on where you

23
00:02:06,600 --> 00:02:13,440
are. Also the form-factor calculations
or theory calculations are done in a

24
00:02:13,440 --> 00:02:17,670
different way. At high Q-squared, it
comes from lattice QCD, at low Q-squared

25
00:02:17,670 --> 00:02:20,670
typically from light-cone sum rules, and in
the middle, there are some effective

26
00:02:20,670 --> 00:02:28,140
parameterisations which we'll come back to
at the end. On slide five, just briefly,

27
00:02:28,140 --> 00:02:33,180
the LHCb detector, as this is the first in
the session. Important for these two analyses

28
00:02:33,240 --> 00:02:40,260
is the high vertex resolution that we have
to separate combinatorial background from

29
00:02:40,320 --> 00:02:47,070
our signals. Good tracking or momentum
resolution for the narrow mass peaks also

30
00:02:47,070 --> 00:02:53,010
for signal to background ratio, particle
identification to reject hadronic decays

31
00:02:53,070 --> 00:02:57,960
in favour of our leptons and then the
calorimetry to trigger on the electrons,

32
00:02:57,990 --> 00:03:04,710
and the muon system which is highly
efficient in triggering on muons. Slide

33
00:03:04,710 --> 00:03:08,910
six. Over the past few years, a lot of
wonderful results came out of this

34
00:03:09,030 --> 00:03:14,040
beautiful detector. So what you see here
are for different b2sll transitions, the

35
00:03:14,040 --> 00:03:17,760
branching ratio as a function of the Q-
squared. And you can see that

36
00:03:17,760 --> 00:03:24,330
consistently, the data points in black are
below the theory predictions. Not so much,

37
00:03:24,930 --> 00:03:29,940
maybe around one sigma, maybe a bit more.
So by themselves, they're not that

38
00:03:29,940 --> 00:03:35,460
spectacular, but combined in an effective
model independent approach, which I'll

39
00:03:35,460 --> 00:03:41,400
come back to later, this goes up to four
sigma or higher away from the Standard

40
00:03:41,400 --> 00:03:48,480
Model. So should we get excited? Maybe we
should start to get excited by about

41
00:03:48,480 --> 00:03:55,440
now. Yes. On slide seven. Besides
these branching ratios, we can measure

42
00:03:55,860 --> 00:04:01,980
angular observables. The K* is a spin-
one vector, and decays to a kaon and a

43
00:04:01,980 --> 00:04:08,340
pion. This kaon-pion pair has
different polarisation amplitudes.

44
00:04:08,340 --> 00:04:14,220
These depend on the Wilson coefficients,
which describe the type of physics that

45
00:04:14,220 --> 00:04:19,050
can happen at the vertex. And they also
depend on the form factors. So, this is

46
00:04:19,050 --> 00:04:24,780
the theory input with some
uncertainty associated to it. Now, in our

47
00:04:24,780 --> 00:04:30,300
fits to the data, which is a function of
Q-squared, and with these three angles,

48
00:04:30,300 --> 00:04:36,150
that you can see in the figure on the
right, we have eight observables. We can

49
00:04:36,150 --> 00:04:40,260
describe them in different bases. The Si
observables are the CP averaged

50
00:04:40,290 --> 00:04:46,320
observables. We can combine those to
make the Pi observables. These are

51
00:04:46,320 --> 00:04:51,600
constructed in such a way to minimise the
theory uncertainties. And for this the P5

52
00:04:51,600 --> 00:04:55,500
prime - for which the calculation is
shown on the bottom right - is this famous

53
00:04:55,590 --> 00:04:59,640
observable that has this big
discrepancy with the Standard

54
00:04:59,640 --> 00:05:00,540
Model expectation.

55
00:05:04,290 --> 00:05:08,880
On slide eight: why am I talking so
enthusiastically about this K*mumu

56
00:05:08,880 --> 00:05:14,010
decay? Well, the previous analysis,
which was using Run-1 data only, so

57
00:05:14,010 --> 00:05:20,280
three inverse femtobarn for LHCb, showed 
some reasonably significant

58
00:05:20,280 --> 00:05:24,690
discrepancies with the Standard Model in
this P5-prime observable. So in

59
00:05:24,690 --> 00:05:30,780
certain bins of Q squared, that goes up to
about three sigma away. And this was a

60
00:05:30,840 --> 00:05:35,520
source for a lot of interest and
discussion, of course. Now, this analysis

61
00:05:35,520 --> 00:05:39,750
that I will talk about today, it
effectively doubles this dataset as you

62
00:05:39,750 --> 00:05:44,700
can see on the on the bottom. So it's the
same amount of decays we roughly

63
00:05:44,700 --> 00:05:51,600
add to this. On slide nine, I just
briefly want to mention the steps we do in

64
00:05:51,600 --> 00:05:56,430
this analysis. We trigger on the muons,
which is highly efficient. We demand

65
00:05:56,430 --> 00:06:01,200
a high vertex quality and impact
parameter criteria, shown on the figure

66
00:06:01,200 --> 00:06:05,760
on the right. We demand particle
identification to reject against what we

67
00:06:05,760 --> 00:06:10,230
call ‘peaking backgrounds’, which mimic the
signal. These mostly come from Lambda-b

68
00:06:10,230 --> 00:06:15,930
decays and Bs decays, where we miss-
identify a proton or kaon as a pion.

69
00:06:17,070 --> 00:06:20,880
Then we use multivariate techniques to
suppress further combinatorial

70
00:06:20,880 --> 00:06:27,630
background. Then, we define eight bins
in Q-squared, in the green regions on the

71
00:06:27,630 --> 00:06:32,520
bottom right plot. We avoid the cc bar
resonances because they come from a

72
00:06:32,520 --> 00:06:37,860
different process. But we do use these,
the J/psi and the psi(2S), as control

73
00:06:37,860 --> 00:06:44,280
channels to completely check all of our
analysis steps before we unblind the data,

74
00:06:44,280 --> 00:06:49,440
so before we look at the final results.
Then we perform this maximum

75
00:06:49,440 --> 00:06:53,850
likelihood fit in these three angles and Q
squared, where the signal parameters, the

76
00:06:53,850 --> 00:06:58,830
Si or the Pi’s are shared between all our
datasets, because the physics should be

77
00:06:58,830 --> 00:07:08,880
the same. On slide 10, just a word
on the angular response. The fact that

78
00:07:08,910 --> 00:07:16,050
LHCb has a very peculiar shape, optimised
for B-hadron decay detection, the

79
00:07:16,050 --> 00:07:20,730
acceptance and also the the selection and
reconstruction, affect how our data

80
00:07:20,730 --> 00:07:25,590
looks in Q-squared and the three angles.
So for instance, on the right, you see the

81
00:07:25,590 --> 00:07:29,370
cosine of one of those angles, for
different Q-squared bins. It's definitely

82
00:07:29,370 --> 00:07:33,210
not flat, and it's also different between
the bins. So we need to understand that,

83
00:07:33,510 --> 00:07:40,020
and this we study with high statistics,
simulated samples. We fit the shapes

84
00:07:40,020 --> 00:07:46,740
using Legendre polynomials in these four
dimensions. And finally, as you can see on

85
00:07:46,740 --> 00:07:51,390
the right, there's also spin zero
components in the K-pi system, which we

86
00:07:51,390 --> 00:07:56,730
also need to include in our angular
description, which we do. This

87
00:07:56,730 --> 00:08:01,320
modelling of the acceptance shape is
actually the largest source of systematic

88
00:08:01,320 --> 00:08:05,970
error, as you can see on the table here.
It depends a bit on which which observable

89
00:08:05,970 --> 00:08:10,920
you look at. But still, this is a lot
smaller than the statistical error that we

90
00:08:10,920 --> 00:08:15,180
have. So we're still limited by the amount
of data that we have, by a factor of four

91
00:08:15,180 --> 00:08:21,840
to 20, even. On slide 11 there are the
new results. I show

92
00:08:21,840 --> 00:08:26,490
you the P5-prime versus Q-squared
again. There are actually eight

93
00:08:26,490 --> 00:08:30,060
angular observables, not just this one.
But this is the most popular one to look

94
00:08:30,060 --> 00:08:34,740
at. What I can say is we have a very good
agreement between our two data sets.

95
00:08:34,740 --> 00:08:40,140
That means we are internally consistent.
This is shown in the plot of S5 versus

96
00:08:40,140 --> 00:08:44,940
Q-squared, on the bottom here. I just
picked one, so that's black versus red.

97
00:08:45,660 --> 00:08:51,180
The local tension in these two Q-squared
bins in P5-prime remained about the

98
00:08:51,180 --> 00:08:56,730
same, but it's somewhat less. Even though
our statistical error, of course, from the

99
00:08:56,730 --> 00:09:03,600
experimental side, decreased a bit. 
This has a reason, because the theory

100
00:09:03,780 --> 00:09:07,350
predictions also changed a bit. The
form factor uncertainties actually

101
00:09:07,350 --> 00:09:11,460
decreased. But the sub-leading
corrections to the interpolation, it

102
00:09:11,460 --> 00:09:16,110
actually became more conservative. 
Together with a small shift in central

103
00:09:16,110 --> 00:09:22,140
value of the observation, this results in just
a slight decrease in discrepancy with

104
00:09:22,140 --> 00:09:27,810
the Standard Model. But the behaviour of
all of our angular observables is now more

105
00:09:27,810 --> 00:09:35,370
consistent with a single change in C9
which is the vector boson coefficient.

106
00:09:35,370 --> 00:09:42,120
Speaking of Wilson coefficients, just
one slide on effective theory. This is a

107
00:09:42,120 --> 00:09:48,000
tool that helps us write the process as a
sum over effective operators, that

108
00:09:48,000 --> 00:09:52,980
describe the physics happening at this
effective four point interaction on the

109
00:09:52,980 --> 00:09:53,760
bottom right block,

110
00:09:55,980 --> 00:09:59,520
and their Wilson coefficients. These
parameterise the amount of which kind of

111
00:09:59,520 --> 00:10:04,170
physics happens at this vertex. The
idea is that this allows for a model

112
00:10:04,170 --> 00:10:08,520
independent interpretation of which kind
of new physics is happening here. But also

113
00:10:08,520 --> 00:10:11,850
it allows to combine different
measurements of different processes which

114
00:10:11,850 --> 00:10:18,990
are sensitive to a subset of the same
Wilson coefficients. Now, in this Q-

115
00:10:18,990 --> 00:10:22,260
squared region where these bins deviate,
the C9 and the C10 - the vector and

116
00:10:22,260 --> 00:10:27,930
axial vector - are actually the most important
ones. When we allow a freedom in C9

117
00:10:27,930 --> 00:10:34,350
and C10 in a fit, you can see
on the bottom left that we have about 2.8

118
00:10:34,350 --> 00:10:38,850
sigma deviation from the Standard Model.
But mostly this appears to be in the C9

119
00:10:38,850 --> 00:10:44,010
direction, so the vector direction.
If we only allow to vary C9,

120
00:10:44,010 --> 00:10:48,240
you see the middle plot, we are actually
3.3 standard deviations away with just

121
00:10:48,270 --> 00:10:52,950
this K*mumu measurement. And
then on the right, for comparison, you see

122
00:10:52,950 --> 00:10:58,830
when we change C9 to minus one, which
seems to be this preferred value, the

123
00:10:58,830 --> 00:11:04,350
Standard Model and the data -
the theory and the data -

124
00:11:04,380 --> 00:11:09,600
match better by eye. So this is going from the
solid line or the blue blocks to these

125
00:11:09,600 --> 00:11:18,960
blue dashed lines. Slide 14. I
mentioned combinations of different

126
00:11:18,960 --> 00:11:22,380
measurements contributing to the same
Wilson coefficients, so it's a natural

127
00:11:22,860 --> 00:11:27,840
moment to move on to the second analysis I
wanted to highlight, which is the

128
00:11:27,900 --> 00:11:33,330
measurement of B to ee. This is also a
b2sll transition, but it has a different

129
00:11:33,360 --> 00:11:38,130
decay topology, to leptons only. And
because of helicity suppression, this is

130
00:11:38,130 --> 00:11:42,990
even more suppressed. If you see this
equation in the branching ratio here, the

131
00:11:42,990 --> 00:11:46,560
helicity suppression occurs in front of
this C10 term, which is the same C10

132
00:11:46,560 --> 00:11:52,140
as we saw for this angular analysis.
But there are also scalar and pseudo-scalar

133
00:11:52,140 --> 00:12:01,170
contributions which were not there. 
They contribute to the

134
00:12:01,170 --> 00:12:06,720
overall picture of the coefficients, of
course, but the B to ee analysis is even

135
00:12:06,720 --> 00:12:13,080
more suppressed in C10. So this is a very
good probe of scalar new physics. And even

136
00:12:13,080 --> 00:12:17,640
one observation, one event, would actually
be a very significant hint for new physics

137
00:12:17,640 --> 00:12:26,130
already. In slide 15, the B to mumu results.
Aiden already talked about it. So, I will

138
00:12:26,130 --> 00:12:30,930
not mention that now. But on the bottom
left, you can see our results for the

139
00:12:30,930 --> 00:12:36,690
search of B to ee, which resulted in no
observation, but a limit that is 30

140
00:12:36,690 --> 00:12:41,220
times better than the previous CDF one. 
The global picture of those

141
00:12:41,310 --> 00:12:44,430
coefficients will help constrain which
kinds of new physics

142
00:12:45,780 --> 00:12:47,310
might explain our other observations.

143
00:12:50,490 --> 00:12:55,770
In slide 16, I'm done with the LHCb
measurements. But I just wanted

144
00:12:55,770 --> 00:13:00,510
to mention that there are many independent
groups who do these models-independent

145
00:13:00,510 --> 00:13:05,490
fits to a lot of these measurements
combined: in the order of 170. And they

146
00:13:05,490 --> 00:13:12,870
find somewhat consistently that C9
only, deviates from zero by four sigma to

147
00:13:12,870 --> 00:13:19,200
five sigma, depending on your
assumptions and which group you ask. But

148
00:13:19,200 --> 00:13:22,650
it all seems to be consistent with a
single change in a Wilson coefficient, this

149
00:13:22,650 --> 00:13:29,490
factor seems to be minus one, maybe. So
this is a very interesting result, how all

150
00:13:29,490 --> 00:13:33,510
of these measurements somehow seem to
point to a single explanation. I think

151
00:13:33,510 --> 00:13:40,290
that's fascinating. So any kind of vector
new physics is very hot at the

152
00:13:40,290 --> 00:13:48,510
moment, of course, because of this. On
slide 17, my conclusion. I showed you two

153
00:13:48,510 --> 00:13:54,120
new exciting results on b2sll transitions
from LHCb: the angular analysis K*mumu,

154
00:13:54,210 --> 00:14:00,840
and this new rare decay, B to ee, where we have
a limit. The discrepancy in this angular

155
00:14:00,840 --> 00:14:04,710
analysis with the Standard Model still
stands, although it did not increase

156
00:14:04,740 --> 00:14:10,710
because of various reasons. And all of
these results are not done with the full

157
00:14:10,710 --> 00:14:17,490
dataset yet: we’re still analysing the
remaining data that we took in Run-2, so we

158
00:14:17,490 --> 00:14:21,930
can expect the what we call ‘Run-1+2
legacy results’ soon. This also

159
00:14:21,930 --> 00:14:26,850
goes for the new B to mumu
analysis, which is coming out, hopefully,

160
00:14:26,850 --> 00:14:33,630
very soon. The B to ee analysis, and on top of
that, we actually have many other analysis

161
00:14:33,690 --> 00:14:39,750
ongoing. These are R(X) analyses that
Miriam will talk about after me, but also we

162
00:14:39,780 --> 00:14:44,520
finally have enough statistics to measure
the angular analysis of K*ee

163
00:14:44,520 --> 00:14:49,920
actually. So all kinds of very interesting
new results will hopefully come out soon

164
00:14:49,920 --> 00:14:51,690
with that more data.

165
00:14:53,130 --> 00:14:56,760
And that's it. Thank you. Thank you 
for your attention.

166
00:14:57,150 --> 00:14:59,970
Question to Jacco, please raise your hand

167
00:15:04,740 --> 00:15:08,040
Yeah, my chemical yosity just started.
Sure.

168
00:15:09,360 --> 00:15:15,690
So, you mentioned second edition and China
is that you can start of course not but

169
00:15:16,020 --> 00:15:21,870
there is one missing I remember the visa
are quite old the result from the bar

170
00:15:21,870 --> 00:15:31,530
about B plus C two three plus down that
was about two or something time 10 to the

171
00:15:31,530 --> 00:15:32,100
minus three

172
00:15:33,840 --> 00:15:35,910
is quite interesting. I think c said

173
00:15:36,360 --> 00:15:41,760
should be sensitive to the effects massing
and the the tweaks doubled more than

174
00:15:42,180 --> 00:15:47,400
So, do you think it is something that can
be feasible in Italy should be a thing. So

175
00:15:47,400 --> 00:15:50,400
since you use already final stage without?

176
00:15:51,269 --> 00:15:52,109
Yeah, so

177
00:15:53,460 --> 00:15:57,840
I will I will not say it's not possible?
Absolutely not. We've made an attempt at B

178
00:15:57,840 --> 00:16:02,940
to dow which is even more challenging. So
it's definitely something we we studied,

179
00:16:03,210 --> 00:16:09,870
we can study I should say. But it is
tricky at the LSE. Right? So these these

180
00:16:10,290 --> 00:16:15,870
measurements with without, you have to
make it takes time. I should say that,

181
00:16:16,020 --> 00:16:20,310
yeah, I would expect the result like that
coming from LGB soon, but it doesn't mean

182
00:16:20,340 --> 00:16:26,070
it's not something maybe we can all do.
And it adds to the to the full picture. So

183
00:16:26,100 --> 00:16:27,540
it's an interesting channel, I think.

184
00:16:29,040 --> 00:16:29,850
Thank you very much.

185
00:16:31,110 --> 00:16:32,190
Other question

186
00:16:34,500 --> 00:16:38,760
seems not so we can move to the third