WEBVTT

1
00:01:53.400 --> 00:01:57.569
Is it any better now? Yes that I can tell.

2
00:01:57.569 --> 00:02:01.019
Okay, beautiful.

3
00:02:05.250 --> 00:02:08.430
Silence.

4
00:02:08.430 --> 00:02:12.449
Silence.

5
00:02:14.699 --> 00:02:25.800
Okay, so final presentations, I'm, I move the group thing to the end because it's a video that I'll show.

6
00:02:25.800 --> 00:02:29.969
Before we go to your presentations, just like to tell you.

7
00:02:29.969 --> 00:02:44.669
Good luck in the future I'm available in the future. I'm even after you leave can discuss any legal and ethical topic doesn't just have to be quantum computing. So it drop me a note. Even after you leave discuss other courses, discuss life in general.

8
00:02:44.669 --> 00:02:49.620
You know, I'm available, but to listen to you, maybe offer advice, but probably not. And also.

9
00:02:49.620 --> 00:02:55.169
Please fill out the course survey that helps me a lot if we get a lot of responses and so on.

10
00:02:55.169 --> 00:03:02.550
Okay, so 1st and John and I'll.

11
00:03:02.550 --> 00:03:09.389
If you can take over, actually share your screen.

12
00:03:09.389 --> 00:03:15.270
If you're here, if not, if you need a few minutes to get started to get set up.

13
00:03:15.270 --> 00:03:21.120
Then, let me know, I.

14
00:03:22.259 --> 00:03:27.060
Yeah, I can see you.

15
00:03:27.060 --> 00:03:30.719
Good, you're sharing your screen, so I'll mute myself now.

16
00:03:40.949 --> 00:03:44.370
All right. Good.

17
00:03:44.370 --> 00:03:47.969
And start off on our research, or.

18
00:03:47.969 --> 00:03:52.110
And I'm John.

19
00:03:52.110 --> 00:03:56.849
Yeah, and, uh, our research is on the fundamentals implementation.

20
00:03:56.849 --> 00:04:00.479
Not only the entire state of events.

21
00:04:00.479 --> 00:04:03.599
Of a trap to on a quantum computer.

22
00:04:05.789 --> 00:04:18.660
So the way we'll be going through this, these are going to start off with how we just like an, in depth overview of our chapter. 9 infrastructure works. It's like how.

23
00:04:18.660 --> 00:04:26.488
What the cubic actually is how would trapping an eye on is the limitation. We're going to go into the stability.

24
00:04:26.488 --> 00:04:30.178
I've turned implementations of time.

25
00:04:30.178 --> 00:04:37.019
Quantum computers, and we're going to go into benchmarking what we came up with to test.

26
00:04:37.019 --> 00:04:40.468
The general state of computers right now.

27
00:04:40.468 --> 00:04:47.819
And then we'll end with that future road map. See on a start.

28
00:04:47.819 --> 00:04:52.288
It's a little bit different of how the cube it's implemented, but.

29
00:04:52.288 --> 00:05:01.949
Say we have these 2 categories, Nashville and artificial cube. It's so we've kind of got hammered home. National Cubics are these phenomenon with.

30
00:05:01.949 --> 00:05:07.798
Articles in elementary article, say, like measuring electrons in and it would have a certain.

31
00:05:07.798 --> 00:05:12.749
State down to the same thing with this.

32
00:05:12.749 --> 00:05:18.119
Elementary particle it's hilarity be horizontal vertical.

33
00:05:18.119 --> 00:05:21.449
For the keep it safe.

34
00:05:21.449 --> 00:05:27.928
However, these these are not easily manipulated or measured, so.

35
00:05:27.928 --> 00:05:31.048
Using, we're still using quantum mechanics.

36
00:05:31.048 --> 00:05:40.228
But say with artificial Cubics, you're doing these macro size processes that still exhibit mechanics, but are actually manipulable and better.

37
00:05:40.228 --> 00:05:44.009
And that's what's going on save for the junction.

38
00:05:44.009 --> 00:05:49.139
Where the quantum dots, and with the assumption on side now.

39
00:05:49.139 --> 00:05:52.978
It's the cubic can actually have multiple.

40
00:05:52.978 --> 00:05:56.249
Um, ways to be measured and set up as a supervision.

41
00:05:56.249 --> 00:06:00.418
That's with respect to charge box space.

42
00:06:00.418 --> 00:06:06.119
It seems the current state of technology there is that.

43
00:06:06.119 --> 00:06:10.499
It's not used as the medium for the cube.

44
00:06:10.499 --> 00:06:15.629
But rather that this phenomenon exhibits different wavelengths.

45
00:06:15.629 --> 00:06:19.048
Which is why it's part of like, a lot of vision technology.

46
00:06:20.548 --> 00:06:25.468
So, the actual cube, it it is that we're using.

47
00:06:25.468 --> 00:06:28.798
Specifically an ion house.

48
00:06:28.798 --> 00:06:31.889
40 C, plus 1 electron is.

49
00:06:31.889 --> 00:06:37.199
And it's superposition is not a factor of spin.

50
00:06:37.199 --> 00:06:40.288
But it's the factor optics so it's an optical Cuban.

51
00:06:41.548 --> 00:06:49.649
So, with this with this paradigm and check when use, but different wavelengths of energy.

52
00:06:49.649 --> 00:06:55.348
Laser light onto this ion, it's free electron space.

53
00:06:55.348 --> 00:06:58.499
Um, orbitals interact in a certain way and.

54
00:06:58.499 --> 00:07:03.358
Okay to other orbitals? No, we'll flip the pointer.

55
00:07:03.358 --> 00:07:06.749
We have over here is, um.

56
00:07:06.749 --> 00:07:11.699
Diagram of how the energy levels are being interacted with.

57
00:07:11.699 --> 00:07:16.918
And with respect to wavelengths, so there's 2 sets the main set.

58
00:07:16.918 --> 00:07:22.139
That is being used, has the advantage of being more stable so.

59
00:07:22.139 --> 00:07:27.749
That's that's noted here and P3 2 the orbital up here.

60
00:07:27.749 --> 00:07:34.978
When it's using that energy level, it's energy level to set the superposition.

61
00:07:34.978 --> 00:07:38.369
So that there is a certain probability.

62
00:07:38.369 --> 00:07:42.269
And then ending up, ending up in the 1 state.

63
00:07:42.269 --> 00:07:48.569
Which is D5 2 and it has another probability of ending in the 0T state.

64
00:07:48.569 --> 00:07:51.718
It's 1 app.

65
00:07:51.718 --> 00:08:01.199
That's that's exactly how superposition is set up for this event and the way it's measured.

66
00:08:01.199 --> 00:08:06.178
We can kind of go into that.

67
00:08:06.178 --> 00:08:11.699
That's part of setting up this queue how to measure and use it everything is that.

68
00:08:11.699 --> 00:08:16.139
These events have to be held in place.

69
00:08:16.139 --> 00:08:19.499
And the way it's set up for.

70
00:08:19.499 --> 00:08:27.389
Now, calcium is that there's 4 electrodes set up in a square configuration.

71
00:08:27.389 --> 00:08:30.809
Well, that's Paul's all strap. So this geometry.

72
00:08:30.809 --> 00:08:35.578
With these pair of charged rods electrical.

73
00:08:35.578 --> 00:08:38.609
Is electrical kind of going through the rods and.

74
00:08:38.609 --> 00:08:43.499
One's posing a negative charge in one's posing a positive charge.

75
00:08:43.499 --> 00:08:49.168
And the pairs, and then they switch so it's constant switching.

76
00:08:49.168 --> 00:08:58.109
Is on the order of, like, radio frequencies so they're switching to positive, negative, positive negative. This is exhibiting electrical field.

77
00:08:58.109 --> 00:09:05.938
Which forces on average the ion, stay in the center of that geometry.

78
00:09:05.938 --> 00:09:12.538
That's that's what keeps the ion stable.

79
00:09:12.538 --> 00:09:15.599
It's not necessary to.

80
00:09:15.599 --> 00:09:20.489
The lasers are for the, with the measurement and the.

81
00:09:20.489 --> 00:09:25.678
The setting up the state, but it's not not for the holding the.

82
00:09:25.678 --> 00:09:34.139
The ions in place now, in order to set all this up in order to create the ion and then bring it into Paul.

83
00:09:34.139 --> 00:09:38.068
On the 1st, it's superheated create.

84
00:09:38.068 --> 00:09:41.099
The calcium.

85
00:09:41.099 --> 00:09:44.668
Nothing and then ways are cool.

86
00:09:44.668 --> 00:09:48.149
Held in place, and then brought over.

87
00:09:48.149 --> 00:09:54.568
Kind of fired is it as it's kind of shown fired over into Paul's trap where, if you're going to be caught.

88
00:09:54.568 --> 00:09:59.038
And the numbers that are going to be caught.

89
00:09:59.038 --> 00:10:03.058
Going to be used as the number of candidates in this whole system.

90
00:10:07.283 --> 00:10:21.234
Okay, so so once our ions are loaded into our trap, uh, we then have to prepare the state into into something that's useful for quantum computation. Uh, so.

91
00:10:22.134 --> 00:10:35.543
As mentioned before we have all these atomic orbitals and different states that are present within our eye on and each of those transitions between states can be uniquely addressed with a different,

92
00:10:35.813 --> 00:10:36.413
um,

93
00:10:36.714 --> 00:10:38.183
wavelength laser polls.

94
00:10:40.073 --> 00:10:40.673
So.

95
00:10:41.129 --> 00:10:47.428
The 1st thing we want to do is prepare all of the ions into a known state. So.

96
00:10:47.428 --> 00:10:51.119
In order to do that, we, we a couple a.

97
00:10:51.119 --> 00:10:56.369
What's called a global addressing beam to the.

98
00:10:56.484 --> 00:11:10.854
The individual ions themselves, which for all ions that are in what's labeled here as as the 0T state, these ions are excited into a short lived auxillary, excited state.

99
00:11:11.124 --> 00:11:20.994
And then that state rapidly mckays down into the lowest orbital, which we, in here, it's labeled as the 1 state. Although the, the choice of.

100
00:11:21.328 --> 00:11:25.259
Which orbitals form the computational basis is somewhat arbitrary.

101
00:11:25.259 --> 00:11:30.389
So, then this, this guarantees that.

102
00:11:30.389 --> 00:11:38.818
All ions, which are already in the 1 state, remain in the 1 state, and all ions, which are initially in the 0T state are now prepared into the 1 state.

103
00:11:38.818 --> 00:11:45.058
So that yeah, sounds so the.

104
00:11:45.058 --> 00:11:59.158
No, no, go back. So, um, so the choice of of which orbitals or which splitting of the orbitals are are used as the basis States dictates what? Pulse.

105
00:11:59.158 --> 00:12:08.818
Energy or pulse of frequency, we use to control the Ion. So typically here, we're, we're talking about optical.

106
00:12:08.818 --> 00:12:11.969
So, these are controlled with optical pulses, but, um.

107
00:12:11.969 --> 00:12:23.099
In general, in practice, the states can be selected such that we can either use megahertz waves or gigahertz waves or low terror hurts waves.

108
00:12:23.099 --> 00:12:26.369
Uh, depending on the energy level, splitting between the different states.

109
00:12:26.369 --> 00:12:34.798
And then and then the detection works on a similar principle. We again illuminate the, the ions with.

110
00:12:34.798 --> 00:12:38.129
A, uh, a global addressing beam this time tuned.

111
00:12:38.129 --> 00:12:42.269
To the wavelength of of this transition between the 1 state.

112
00:12:42.269 --> 00:12:49.739
And now it's an auxiliary excited state used for readout. And so this, this, if.

113
00:12:50.153 --> 00:13:02.604
If an ion is in the 1 state, it becomes excited into the auxillary state, which then rapidly to case back down into the 1 state, and that's emits a photon in the process. So they, they refer to this as a bright transition.

114
00:13:02.994 --> 00:13:06.714
So, if we, if we measure scattered photons from.

115
00:13:07.048 --> 00:13:10.288
The the eye on, then we know that it's in the 1 state.

116
00:13:10.288 --> 00:13:14.759
If we don't measure scattered photons from the ion, then we know that it's in the 0T state.

117
00:13:14.759 --> 00:13:19.859
And typically, both of these processes take approximately less than 1 millisecond.

118
00:13:20.514 --> 00:13:30.323
Next slide so then the actual control of the on itself of the computational basis States is a very similar process.

119
00:13:30.354 --> 00:13:36.683
Except this time we're using a a wavelength that's tuned to the transition between the 1 and the 0T state.

120
00:13:37.019 --> 00:13:49.828
So, whereas in the previous slide, where we're talking about this sort of light, red, global, addressing beam here, now we're talking about those darker red individual, addressing ion beams.

121
00:13:49.828 --> 00:13:53.578
And then by controlling either.

122
00:13:53.964 --> 00:14:03.144
Or some combination of the length, or the amplitude, the frequency, and the polarization of the pulse we control, what type of gate is implemented.

123
00:14:03.323 --> 00:14:09.774
So we control the amount of rotation and the, the angle of rotation between the 0T and 1 States.

124
00:14:10.344 --> 00:14:20.183
So, typically, these these single cube operations are very well defined, and typically, the fidelity is limited by the accuracy and precision of our sources.

125
00:14:20.754 --> 00:14:25.943
So, then maximum fidelity, which has been experimental demonstrated has been 99.999% with 8 times.

126
00:14:25.943 --> 00:14:39.323
Typically, in the range of 1 to 20 micro seconds, and while this 1 to 2000, micro seconds may sound like a very short time. If we compare that to superconducting, which take approximately nanoseconds to implement K times.

127
00:14:39.323 --> 00:14:43.764
It's actually very slow gate time.

128
00:14:43.764 --> 00:14:55.134
If we were to compare this to sort of the analogy of classical computers, superconducting cube, it would be classical computers with clock speeds on the order of gigahertz. Whereas.

129
00:14:58.798 --> 00:15:06.749
Trapped ion Cubics would be like classical computers with clock speeds on the order of hundreds of kilohertz. So it's significantly slower.

130
00:15:07.224 --> 00:15:22.193
And this gate time can be slowed down, but, or it can be shortened. But in general, there's a, there's an inverse correlation between daytime and fidelity. So if we shorten that gay time, we sacrifice some amount of fidelity now in the operation.

131
00:15:23.903 --> 00:15:37.193
Next slide so then the implementation of multi cubic Gates is is slightly more complex than than the implementation of single cubic Gates.

132
00:15:37.734 --> 00:15:45.413
And this complexity often leads to the multi QB Gates being lower fidelity than the single cubic Gates.

133
00:15:46.524 --> 00:15:59.484
So, the Max fidelity, which has been demonstrated with multi cubic Gates and trapped ions is as high as 99.3% and the gay times are typically much longer than the single cubic Gates ranging from 30 to 250 microseconds.

134
00:16:01.859 --> 00:16:07.918
In practice, we typically see fidelity, which is in the range of 96 to 99%.

135
00:16:07.918 --> 00:16:21.869
So the type of multi cubic gate that's implemented in trapped ions is called the gate. It's essentially a controlled Z gate, whereas the common C, not operation is a controlled escape.

136
00:16:21.869 --> 00:16:25.438
So so here we.

137
00:16:25.438 --> 00:16:37.288
In in this schematic, it's shown for just 2 bits of control and a target but because all of all of the ions in the chain can be entangled together.

138
00:16:37.288 --> 00:16:44.428
Uh, this can be extended to any number of Cubics. So, the general idea is that the.

139
00:16:44.428 --> 00:16:51.989
The internal ironic States of the ions themselves are are coupled via the emotional states.

140
00:16:51.989 --> 00:16:56.969
Of the quantum wells induced by the Paul trap so.

141
00:16:56.969 --> 00:17:08.278
We have certain selection rules which are allowed and disallowed based on both the emotional state and the Ionic state of of.

142
00:17:08.278 --> 00:17:11.519
The couple of the ions, so.

143
00:17:11.519 --> 00:17:15.538
By by controlling the length of the pulse.

144
00:17:15.538 --> 00:17:22.318
Uh, and the order that we apply these pulses, uh, we can create and tangling operations between multiple Cuba.

145
00:17:22.824 --> 00:17:29.963
And the net result of that gate is, is shown in the upper right corner.

146
00:17:31.193 --> 00:17:41.753
And so, like I said, this is this is a control Z operation where normally we're used to seeing the controlled not operation, but using well known a quantum circuit identity.

147
00:17:41.753 --> 00:17:51.294
We can actually implement the controlled, not operation using the control Z operation simply by adding to how to mark gates on either side of, of that control to the gate.

148
00:17:51.568 --> 00:17:59.939
Next slide so then it trap systems are.

149
00:17:59.939 --> 00:18:06.778
Most well known for, for being very stable quantum computing modality.

150
00:18:07.523 --> 00:18:21.834
Part of the reason for this is that we're using ions as physical CubeSats. So just to, by the laws of physics, all of these ions will be physically identical. So whereas with with we're.

151
00:18:23.634 --> 00:18:37.943
We have process variations, so each, each cube, it is slightly different and so that requires a very intensive calibration. So for trapped iron quantum computing this, these calibration requirements are are nearly as strict.

152
00:18:38.729 --> 00:18:46.979
But it does, uh, sort of as a double edged sword it, it places very strict requirements on our.

153
00:18:46.979 --> 00:18:51.088
Are sources and detectors we have to be able to very precisely control.

154
00:18:51.088 --> 00:18:58.078
Uh, the, and the stability of our sources, and all of the electric fields, which are used to control these ions.

155
00:18:59.094 --> 00:19:09.114
But the benefit of this system is that we have very long coherence times, so in this diagram on the right this is sort of the basic ironic structure or basic orbital structure of an eye on.

156
00:19:09.564 --> 00:19:17.784
And most commonly, this red quadruple transition is used between the ground state and the orbital. And, uh.

157
00:19:18.118 --> 00:19:28.138
This transition has a very long lifetime of of typically on the order of 3rd, but as long as several seconds, depending on the species used.

158
00:19:28.138 --> 00:19:34.888
Um, and so because of this very long coherence time, even with the longer gate time, we can still fit.

159
00:19:34.888 --> 00:19:49.104
Many more operations into the coherence time of the cube so we can achieve much greater circuit depth as much as some estimates put it at a 1000 times greater than the circuit depth of

160
00:19:49.104 --> 00:19:51.173
superconducting Cubans.

161
00:19:51.479 --> 00:20:03.749
But 1 additional drawback is that the systems are susceptible to unique arrow error channels. 1 of those being the leakage air, which would be.

162
00:20:03.749 --> 00:20:06.929
Uh, the case where we accidentally excite.

163
00:20:06.929 --> 00:20:18.148
The ion into a state, which isn't used as a computational basis. So if in this diagram on the left, if we say, accidentally excited into the P orbital or the f orbital this.

164
00:20:18.148 --> 00:20:23.729
Either of these states aren't used as our computational basis and would not be would not be.

165
00:20:23.729 --> 00:20:31.048
Accurately read out by the readout operations and so we would almost in a way, sort of, sort of lose the information on that. It.

166
00:20:31.673 --> 00:20:36.923
Similarly, we can also physically lose the eye on itself so the on can sort of escape the trap.

167
00:20:37.193 --> 00:20:47.063
If, um, if we don't have a very high quality vacuum on the trapped ions, then extraneous gas particles can collide with ion and essentially knock it out of the trap.

168
00:20:48.743 --> 00:21:02.753
Next slide, so we wanted to compare trapped ion systems to other commercially available superconducting transport systems. So, in order to do this.

169
00:21:03.749 --> 00:21:13.199
We researched 2 benchmarking algorithms, which are commonly used. 1, is the Bernstein algorithm and the other is the hidden shift algorithm.

170
00:21:13.199 --> 00:21:25.469
The Bernstein Rani algorithm, basically in a single shot implements some Oracle function, which is the product of X and some constant and in a single shot, it returns that constant.

171
00:21:25.824 --> 00:21:39.503
Similarly, the head and shift algorithm determines some shift s of some function and again, and returns it in in 1 shot. So both both algorithms have similar outputs.

172
00:21:39.503 --> 00:21:44.423
So the, the results of the 2 can sort of be compared, but they have different circuit requirements.

173
00:21:44.699 --> 00:21:58.193
Uh, so the Bernstein, Ronnie, uh, that we can see on the left requires a high degree of entanglement. So it requires high connectivity from our quantum processor. Whereas the hidden shift algorithm is more of a deeper circuit.

174
00:21:58.193 --> 00:22:02.544
So it requires larger circuit depth and not more operations.

175
00:22:03.023 --> 00:22:16.193
So the Bernstein algorithm favors quantum processors that have a high degree of con activity, whereas the hidden shift algorithm favors quantum processors, which have good circuit depth and that's high fidelity.

176
00:22:18.568 --> 00:22:21.659
Next.

177
00:22:21.659 --> 00:22:21.989
So,

178
00:22:21.983 --> 00:22:22.253
our,

179
00:22:22.253 --> 00:22:28.433
our goal in implementing this benchmarking was to reproduce results from a 2017 paper from ion queue,

180
00:22:28.433 --> 00:22:29.034
where they,

181
00:22:29.034 --> 00:22:36.983
they run these 2 algorithms on an IBM system and their system and and compare the results of the 2.

182
00:22:36.983 --> 00:22:37.074
so.

183
00:22:38.128 --> 00:22:43.739
Sort of as expected the die on queue performs better than.

184
00:22:43.739 --> 00:22:52.828
Then IBM system in the lower left corner or lower right corner here, we can see that the Bernstein files or on a.

185
00:22:52.828 --> 00:23:06.749
Algorithm has around 73% success probability on IBM where it has around 85% success profitability on Q. a corresponding. The, the, the heading shift has around 35% success profitability on IBM.

186
00:23:06.749 --> 00:23:12.028
Whereas, uh, around 77% success probability on on Q.

187
00:23:14.243 --> 00:23:14.604
Excellent.

188
00:23:16.134 --> 00:23:16.673
So here,

189
00:23:17.394 --> 00:23:17.753
we,

190
00:23:18.084 --> 00:23:20.334
we tried to reproduce the results using IQ,

191
00:23:20.364 --> 00:23:28.074
and we actually found the success probability to be slightly better than that demonstrated in on q2017 paper,

192
00:23:28.074 --> 00:23:30.594
which we sort of expect because their,

193
00:23:30.594 --> 00:23:33.983
their hardware has improved over the past 3 years.

194
00:23:33.983 --> 00:23:35.304
So, here we ran.

195
00:23:38.608 --> 00:23:48.509
For the plot on the left is the Bernstein Ronnie output and then the plot on the right is the head and shift output and the X axis is is each different possible.

196
00:23:50.064 --> 00:24:04.253
In this case, it's a, it's a, for Cupid alibre algorithm so each possible Oracle C or each possible hitting shift. S. and so so, in the ideal case, we expect the diagonal matrix to be all ones.

197
00:24:04.733 --> 00:24:13.044
And in fact, we get something very close to that. So, we get around 92% success rate for the Bernstein Ronnie, and around 84% success rate for the head and shift.

198
00:24:14.038 --> 00:24:18.568
So, roughly 6% better than, uh, than what was demonstrated in 17.

199
00:24:19.584 --> 00:24:34.554
And in general, this, this benchmarking really favors on queue because with the trap dial system, we have full activity, whereas with IBM system, we have relatively limited connectivity.

200
00:24:35.548 --> 00:24:38.699
Next.

201
00:24:38.699 --> 00:24:45.269
So, I did the IBM half of the same testing and what happens.

202
00:24:45.269 --> 00:24:48.749
Is that IBM allows you to do more shots? So you can do.

203
00:24:48.749 --> 00:24:52.348
1024 shots I did that for each case as well.

204
00:24:52.348 --> 00:24:57.449
And what's going on for the Bernstein buzzer? Rodney?

205
00:24:57.449 --> 00:25:00.898
And is that the success rate is actually lower than.

206
00:25:00.898 --> 00:25:06.659
What was produced in the paper so that 31% where the paper was at 35%.

207
00:25:06.659 --> 00:25:10.409
And kind of see some of these.

208
00:25:10.409 --> 00:25:14.189
High successes are coordinating, it's like.

209
00:25:14.189 --> 00:25:19.469
The 2nd cubin, um, sometimes shifting so that's what.

210
00:25:19.469 --> 00:25:26.249
Might be might be making an error during the Bernstein or any.

211
00:25:26.249 --> 00:25:32.669
Calculation, um, what's produced on the right the hidden shift benchmarking.

212
00:25:32.669 --> 00:25:39.148
That's kind of par for what happened in the paper, in fact, and prove about 3%. So, this is that.

213
00:25:39.148 --> 00:25:42.778
80% success where the papers around 77.

214
00:25:42.778 --> 00:25:47.429
So it seems like.

215
00:25:47.429 --> 00:25:52.019
They're doing they are improving they are getting a.

216
00:25:52.019 --> 00:25:56.548
Coherent readings when you when you're putting a.

217
00:25:56.548 --> 00:26:00.689
You're putting these on a computer through the algorithm.

218
00:26:00.689 --> 00:26:04.078
But it seems like what we did with.

219
00:26:04.078 --> 00:26:09.838
Star configuration of their key quantum computer.

220
00:26:09.838 --> 00:26:13.648
That did not show significantly better results.

221
00:26:13.648 --> 00:26:17.759
We're putting it through this, the burn steam around.

222
00:26:21.683 --> 00:26:32.153
So, just looking forward to the future of in quantum computing, uh, recently released their roadmap for how they expect to scale their quantum computers.

223
00:26:32.634 --> 00:26:45.834
Um, so they're really looking forward to a fault tolerant error, corrected era of quantum computing. So recently they demonstrated that, um.

224
00:26:46.348 --> 00:26:57.118
They could implement a a single fault tolerant cubic, uh, using a 13 to 1 overhead ratio. So they have 13 physical ions representing 1 error corrected, keep it.

225
00:26:58.104 --> 00:26:58.554
So,

226
00:26:59.814 --> 00:27:00.263
part of this,

227
00:27:00.263 --> 00:27:03.144
is that the very high fidelity of the,

228
00:27:04.584 --> 00:27:09.144
and enables them to have a much lower overhead for error corrections so,

229
00:27:09.144 --> 00:27:13.973
whereas on superconducting where the fidelity is lower,

230
00:27:13.973 --> 00:27:17.903
it's expected that it could have at least a 1000 to 1,

231
00:27:17.933 --> 00:27:22.344
overhead of physical cube to a virtual cube.

232
00:27:22.888 --> 00:27:34.108
So, initially I on Q, expects to scale somewhat slowly and and here on the on this access for.

233
00:27:34.108 --> 00:27:44.848
For the numbers, this is actually the algorithm algorithmic Cubans. So this is sort of analogous to to the virtual key bits. Um.

234
00:27:44.848 --> 00:27:47.939
Represented by many more physical Cupid.

235
00:27:47.939 --> 00:27:51.598
Um, so.

236
00:27:53.213 --> 00:28:07.554
So, they define this algorithm as as sort of the log based to scale representation of the quantum volume. Because their argument is that the way that IBM has defined quantum volume.

237
00:28:07.858 --> 00:28:15.028
It's sort of it's exponentially related to the fidelity of the gates. So if.

238
00:28:15.028 --> 00:28:22.409
If I am keyword to continue using quantum volume very quickly, they would get to sort of absurd numbers of quantum volume there.

239
00:28:22.409 --> 00:28:26.759
They estimate that their upcoming 32 cubic processor.

240
00:28:26.759 --> 00:28:38.189
Would have a quantum volume of around 4M using metric so they, they think that that's sort of an unrealistic representation. So they determine they defined this algorithmic Cuba.

241
00:28:39.804 --> 00:28:52.614
The other thing that they project is to have sort of rack mounted sort of server style data center, quantum computers accessible by 2023 and they think that these rack amount style quantum computers could be used to implement.

242
00:28:54.298 --> 00:29:04.199
Um, sort of meaningful demonstrations of quantum machine learning, uh, as well as leading to a demonstration of quantum advantage or quantum supremacy by 2025.

243
00:29:05.814 --> 00:29:06.173
Excellent.

244
00:29:07.943 --> 00:29:15.503
So the key barriers to realizing the scalability are 1,

245
00:29:15.503 --> 00:29:18.114
obviously improving the quality of the,

246
00:29:19.104 --> 00:29:20.153
the themselves,

247
00:29:20.183 --> 00:29:21.443
even though the quality is very high,

248
00:29:21.443 --> 00:29:23.814
we would like to be as high as physically possible.

249
00:29:24.118 --> 00:29:32.189
So, this relates really to the choice of ions, use the system temperature used the way that the, the different.

250
00:29:32.189 --> 00:29:34.703
orbitals that are used as the computational basis,

251
00:29:34.973 --> 00:29:38.124
and the way that the gates are implemented as a series of pulses,

252
00:29:38.544 --> 00:29:41.183
and specifically relating to the gate implementation,

253
00:29:41.213 --> 00:29:49.884
improving that the gate time while not compromising too much on fidelity will be a a significant endeavor although.

254
00:29:50.189 --> 00:29:56.578
Some areas of research right now, relating to the use of ultra fast optics to control these.

255
00:29:56.578 --> 00:30:07.019
Iot Cubics have demonstrated, or at least suggest that they could get a gay times down to the tens of PCO. 2nd range using ultra fast optical pulses.

256
00:30:07.019 --> 00:30:19.078
And another method to increase the scalability will obviously be to reduce the size of this system. And this will come through the monolithic integration of sources and detectors basically into integrated circuits.

257
00:30:19.078 --> 00:30:33.983
So, they have to integrate all of the voltage and our sources used for the ion trapping as well as the photon sources used for the cooling, the preparation, the read out and the control. All of these would ideally be integrated onto 1, single chip.

258
00:30:35.003 --> 00:30:41.064
Whereas now they're all sort of external sources, which are then coupled into the monolithically integrated ion trap.

259
00:30:41.729 --> 00:30:50.818
Next time. Okay so this is the end of our presentation Thank you for listening and will now answer any questions at this time.

260
00:30:50.818 --> 00:30:54.868
Yeah, thank you very much. I'll start out, so this is.

261
00:30:54.868 --> 00:30:58.229
Sounds like, I am curious leading IBM in the dust. Is that accurate?

262
00:30:59.429 --> 00:31:10.288
Yeah, I mean, I think our results definitely show that 1 of the really strong advantages of or 2 of the really strong advantages of Q is both the high degree of connectivity.

263
00:31:10.288 --> 00:31:21.719
And the superior fidelity, so if you look at a lot of IBM apologies, the connectivity is very limited. I mean, with the superconducting transplants, they can really only get sort of.

264
00:31:21.719 --> 00:31:26.759
Nearest neighbor connectivity, whereas has has full connectivity.

265
00:31:28.409 --> 00:31:38.278
I think that IBM is sort of a little bit more aggressive in their predict projections, whereas ion Q is a very conservative, but.

266
00:31:38.278 --> 00:31:42.239
I think I on Q, at least from the, from the papers that we read.

267
00:31:42.239 --> 00:31:45.959
And the literature review that we did, I think they're.

268
00:31:45.959 --> 00:31:49.798
Their background and foundation foundation is is very solid.

269
00:31:49.798 --> 00:31:54.058
Thank you does anyone else have a question.

270
00:31:56.128 --> 00:32:03.449
Okay, well, thanks again. Oh, there was a question in the chat. I missed it. Connor.

271
00:32:06.538 --> 00:32:12.989
Yeah, I mean, the gate time is definitely a significant challenge.

272
00:32:12.989 --> 00:32:21.058
There are some areas of research. Like I said, in the last slide alter fast optics.

273
00:32:21.058 --> 00:32:24.449
Um, basically using sort of 2nd, Pulse, lasers.

274
00:32:24.449 --> 00:32:29.519
Is is 1 area of research to reduce the gay times.

275
00:32:29.519 --> 00:32:38.068
Um, it's sort of like, the main main pursuit in reducing the gay times, or, at least the only 1 that's really demonstrating results below 1 Micro. 2nd.

276
00:32:38.068 --> 00:32:44.459
So, it's still relatively new and and yet to really prove high fidelity Gates.

277
00:32:44.459 --> 00:32:48.868
But it's definitely it's definitely something that they want to do.

278
00:32:48.868 --> 00:32:54.719
Okay, what else? Okay, thanks again, John and G.

279
00:32:54.719 --> 00:33:00.868
So, are you ready to tell us what you're doing?

280
00:33:00.868 --> 00:33:03.959
Oh, yeah. So can you hear me.

281
00:33:03.959 --> 00:33:08.878
Yes, I can hear you. Okay. So if you can share my screen now.

282
00:33:10.888 --> 00:33:14.459
So, uh.

283
00:33:15.263 --> 00:33:21.894
Good afternoon everyone and my name's you. Hi, and today I'm going to talk about spout more something about the content meaning.

284
00:33:22.374 --> 00:33:23.963
And here's the accountant,

285
00:33:23.993 --> 00:33:33.653
and the 1st part is the background and basic concepts and timelines some algorithms and also the conclusions and the perspectives,

286
00:33:34.134 --> 00:33:35.814
and in this presentation,

287
00:33:35.814 --> 00:33:36.804
both see,

288
00:33:37.314 --> 00:33:41.544
if we're in the high level of quantum willing will be covered.

289
00:33:41.909 --> 00:33:46.558
So, we can see that recently. There's a very fast development in the.

290
00:33:46.884 --> 00:33:52.403
Area of concern computing and the most important 1 is nastier.

291
00:33:52.463 --> 00:34:06.054
Google quantum group announced as he had reached the content supremacy and use the 53 cubic device and for some specific setups or.

292
00:34:06.598 --> 00:34:20.693
Algorithms, whereas are classic computer use, like, 100 years old, 10 sound ears to finish the task while the content computer would only use several minutes. So, it reached that.

293
00:34:20.724 --> 00:34:24.023
So called condom supremacy and also the also tried.

294
00:34:25.378 --> 00:34:28.739
Use the computer to do the chemistry.

295
00:34:28.739 --> 00:34:34.559
Which means to Motors, making themselves advising.

296
00:34:34.559 --> 00:34:36.923
And the successfully Motors process,

297
00:34:37.824 --> 00:34:52.373
and even more recently in this group from China announced and published paper in science saying that they have realized for tonic device contents

298
00:34:52.554 --> 00:34:52.974
missing.

299
00:34:53.724 --> 00:34:59.784
And although they still scramble for tonic device.

300
00:35:00.148 --> 00:35:04.889
But still is kind of a breakthrough.

301
00:35:04.889 --> 00:35:09.239
And for now, the realization of Cuba is are mostly.

302
00:35:09.239 --> 00:35:10.554
Joseph and Johnson,

303
00:35:11.454 --> 00:35:21.954
Joseph jumped in superconducting qbs or the so called transmen and also the chapter line Cupid and in order to realize all those qbrs,

304
00:35:22.764 --> 00:35:30.054
we need to make some control on those cumulus sometimes that we use a field or the electric field.

305
00:35:30.923 --> 00:35:40.434
And we try to manipulate the stadiums. cubeys is sometimes kind of difficult because you can.

306
00:35:40.943 --> 00:35:55.793
It's really hard to exactly control the face of the springs. So this could lead to the arrows in the content system. And this is 1 of the most important challenges for the clinic computing and.

307
00:35:58.014 --> 00:36:11.273
Especially when we try to scare opposite device or try to reach more and more curious, be more and more difficult. So, this mixed scaling of Cape, based a quantum computer parameter now.

308
00:36:11.813 --> 00:36:20.454
So now, scientists and researchers are trying to find some to do is difficulty.

309
00:36:20.514 --> 00:36:35.123
And 1 of them is contaminating and the basic idea of the content naming is to sacrifice University. And so the trait with the so they get the scaling.

310
00:36:35.844 --> 00:36:38.543
And the basic idea is to.

311
00:36:38.878 --> 00:36:50.364
Transform the target program into finding the global mileage state and the problem dealing with C. C. to solve some optimization problem.

312
00:36:50.393 --> 00:37:00.563
The QBO problems kubranetes the quadratic, unconstrained binary optimization problems. And also the.

313
00:37:00.989 --> 00:37:13.318
Content is really closely related to the adobatic quantum computing. So here's a very simple analogy, but explains how it works.

314
00:37:13.318 --> 00:37:19.289
So suppose that we, a person is standing on top of.

315
00:37:19.289 --> 00:37:26.398
Mountain and he's surrounded with valleys and peaks. So the goal of this person is to fight the lowest.

316
00:37:26.398 --> 00:37:29.818
Possible Valley in this area, which so.

317
00:37:29.818 --> 00:37:33.268
This corresponds to finding the minimum.

318
00:37:34.583 --> 00:37:45.534
And, of course, the classical ways to go through all those valleys and heels, and then measure all his rallies and get the final result.

319
00:37:46.463 --> 00:37:56.423
Although it seems to be very tidier space of course, the effective way. But if we have a very narrow problem, and it could cost a lot of time.

320
00:37:56.994 --> 00:38:11.873
So, in the quantum computer, we can make use of their stupid position internally. So, if we have the superposition state, this means that all those different 1 person can coexist in all of those piece.

321
00:38:12.204 --> 00:38:17.873
So, each person will just be responsible for marrying.

322
00:38:18.449 --> 00:38:23.130
Low the near near is valleys.

323
00:38:23.130 --> 00:38:32.070
So, and also we can take, you make use of the which means that a person does now need to just.

324
00:38:32.425 --> 00:38:45.925
Just, as a person can turn us through those heels, but instead of just typing it, so it saves saves time and also it will prevent being trapped into some no commitments.

325
00:38:46.465 --> 00:38:55.974
So, in this 1, in this way, theoretically, we can get content speed up compared to the classroom computer and the content speed up is.

326
00:38:56.039 --> 00:39:00.179
Always refer to a rift reference to the.

327
00:39:00.179 --> 00:39:15.150
Explains your speed up and here's some of the formulations and most of their content meaning devices will use so called model and here's them up. There's a characteristic timescale.

328
00:39:15.150 --> 00:39:21.929
So, Here's some of the key events for the content meaning and.

329
00:39:21.929 --> 00:39:28.440
The idea is proposed 1st, proposed in 9, 8, 8, and.

330
00:39:28.440 --> 00:39:39.389
Uh, the formalization is formed in 98 and the 1st experiment is down in the also in 99 9.

331
00:39:39.389 --> 00:39:47.400
And recently, and from 2007 and 2020, the wave counseling has.

332
00:39:48.025 --> 00:40:01.525
Made a lot of success in this in this area, and they used it created devices for now, the device is up to 5000 Cubans. So it's very amazing. And now we can.

333
00:40:02.670 --> 00:40:06.000
Results devices, so we can try some benchmark.

334
00:40:06.000 --> 00:40:11.909
Algorithms on it is any very convenient. So now we try stand up to.

335
00:40:13.255 --> 00:40:20.635
Algorithms the 1st one's maximum card problem. I think most of you have also tried this in the homework.

336
00:40:21.085 --> 00:40:30.534
So this is the idea is very easy to find a cut who size up top or the total items, at least as charging cards. So we can just.

337
00:40:31.409 --> 00:40:36.480
1st, in the in the bracket.

338
00:40:36.480 --> 00:40:40.260
Final form, we can just create those codes.

339
00:40:40.260 --> 00:40:53.034
And write those the right to use this some of the costs to create the geometry and trying to use both the seminary real device to find the minimal.

340
00:40:53.034 --> 00:41:05.065
And, as a result is, is not very difficult to get the read out. And also the 2nd problem is traveling salesmen problem and.

341
00:41:05.369 --> 00:41:14.909
The idea is to, is that we have a list of points or the subsidies and our goal is to.

342
00:41:15.324 --> 00:41:19.315
Find the shortest possible route that can visit all those cities,

343
00:41:19.315 --> 00:41:20.454
or no,

344
00:41:22.465 --> 00:41:31.344
and we only visit each point only 1 visit only once and we're finding we need to return to the original points.

345
00:41:32.250 --> 00:41:37.409
And Here's some of the formalizations and it, we can see that the.

346
00:41:37.855 --> 00:41:52.135
How many Tony and distant tonia or there is a term called this is this represents the distance between 2 nodes and also we need some constraints because we have to make sure that.

347
00:41:52.860 --> 00:41:56.184
This means that each city will be for polling.

348
00:41:56.184 --> 00:42:07.914
So I'll be visiting owning once and we need to make sure that each position up on the road should only be allocated to owning 1 city or points.

349
00:42:08.965 --> 00:42:23.125
So we can similarly, we can use some codes in best practice platform, a true cradle to cradle the geometry.

350
00:42:23.155 --> 00:42:26.905
And actually, the strong tree is directly obtained from the.

351
00:42:27.210 --> 00:42:30.480
Predefined graph. So.

352
00:42:30.480 --> 00:42:33.570
As here's a.

353
00:42:33.570 --> 00:42:42.030
And to solve this problem, and we can direct a call, the traveling salesman Cubans.

354
00:42:42.030 --> 00:42:50.309
Because it already has a solver when we solve these problems. So we can get a.

355
00:42:50.309 --> 00:43:05.094
For this geometry simulator gives this rays out from the 0T to 3 to 4 to 1 and 2 and we can see that this is exactly the optimal optimal result.

356
00:43:05.125 --> 00:43:20.005
And the total distance, including return is 19, and for the device will get and as a result, actually, 02143 is actually the same result as a simulator.

357
00:43:20.005 --> 00:43:25.014
Just just as a river. It's the reverse of orders. And.

358
00:43:28.530 --> 00:43:40.050
So, we can see that using the WS Bret bracket platform. We can easily program some of those benchmark algorithms.

359
00:43:40.050 --> 00:43:47.820
And it's and also for the simple for those simple questions and problems.

360
00:43:47.820 --> 00:43:56.969
The content only device matters simulator all the real device can always predict the right right. Answer. But however.

361
00:43:56.969 --> 00:44:03.900
It's really hard. We cannot tell you if it's if you have any speed up compared to.

362
00:44:03.900 --> 00:44:09.630
Classic computer, because we our.

363
00:44:09.630 --> 00:44:14.400
But the problem we're interest is very simple. So, in order to.

364
00:44:14.400 --> 00:44:19.440
Exam it, whether or not, we can get quantum speed up. We need to.

365
00:44:19.440 --> 00:44:32.250
And make try to find some more advanced algorithms, and also we need to make the geometry of our network to be much more and more difficult or more complex.

366
00:44:32.250 --> 00:44:43.199
Okay, so to sum up, we can see that in order to develop in order to make the our system.

367
00:44:43.764 --> 00:44:51.445
In order to make the content meaning to be more competitive, but we need to develop some more and Navajo Antonia formulation.

368
00:44:51.835 --> 00:45:02.215
So call now stochastic and also as being how field control masses, and also to improve the superconducting cubes.

369
00:45:02.639 --> 00:45:11.670
And here's some of the references. Okay, that's it. Thank you for your attention and don't hesitate to ask any questions and concerns.

370
00:45:11.670 --> 00:45:15.119
Yeah, thank you very much then.

371
00:45:15.119 --> 00:45:18.329
Laura is open any questions on this.

372
00:45:19.559 --> 00:45:26.099
So they got a lot of potential then D, wave with.

373
00:45:26.099 --> 00:45:35.219
See, how they do to see a contestant from Canada? No, I'm from Canada originally, so okay.

374
00:45:38.039 --> 00:45:51.030
So then the final presentation and, and new SRE, they uploaded it to YouTube and I'll show the YouTube video now.

375
00:45:51.030 --> 00:46:04.469
Silence.

376
00:47:01.224 --> 00:47:04.764
Okay, this is why I have 2 machines running.

377
00:47:05.039 --> 00:47:12.900
I think I'm talking to you on my iPad, because in my laptop just on so.

378
00:47:12.900 --> 00:47:19.320
Give me a minute to see if I can get it going again. So I can run.

379
00:47:19.320 --> 00:47:24.449
You 2.

380
00:47:27.059 --> 00:47:33.059
Key, um, I'm going to see if I can run things from.

381
00:47:33.059 --> 00:47:38.880
The projected from the iPad see, if this works here, just give me a 2nd.

382
00:47:56.190 --> 00:47:59.760
Silence.

383
00:48:20.070 --> 00:48:28.380
Silence.

384
00:48:28.380 --> 00:49:10.079
Silence.

385
00:49:32.099 --> 00:49:37.289
Simulation.

386
00:49:38.849 --> 00:49:44.489
Silence.

387
00:49:50.789 --> 00:50:02.219
Silence.

388
00:50:24.210 --> 00:50:28.079
Silence.

389
00:50:34.710 --> 00:50:51.929
Silence.

390
00:51:36.119 --> 00:51:39.360
Okay, so.

391
00:51:48.989 --> 00:51:54.119
My crime computer's locked up. I don't want to take your time. It would take several minutes to reboot it.

392
00:51:54.119 --> 00:52:00.869
That's my well think pad we're playing the YouTube video through my iPad.

393
00:52:00.869 --> 00:52:04.409
The volume is so low that I can hard to hear it.

394
00:52:04.409 --> 00:52:09.329
And I don't think you can hear it either. So.

395
00:52:09.329 --> 00:52:13.469
I have the link to it online. I'm thinking.

396
00:52:13.469 --> 00:52:18.599
If it's okay with you guys that you can watch it on your own.

397
00:52:18.599 --> 00:52:22.920
And watch it and rewatch it. If you wish.

398
00:52:22.920 --> 00:52:27.239
And that I not take, I not spend a lot of time trying to.

399
00:52:27.239 --> 00:52:32.280
Get it going right now if that sounds reasonable for people.

400
00:52:34.139 --> 00:52:41.429
And if so then this would be the end of the class.

401
00:52:41.429 --> 00:52:45.210
And the end of the course, I'm open, we have some time.

402
00:52:45.210 --> 00:52:51.630
I'm open if we have if anyone has any questions.

403
00:52:51.630 --> 00:52:57.179
And 1 question, I'd like to ask you is how much money did you spend on Amazon?

404
00:52:57.179 --> 00:53:01.019
Was it a trivial amount or a serious amount or.

405
00:53:11.670 --> 00:53:18.119
Okay, dollars photo for.

406
00:53:20.400 --> 00:53:25.289
So, by standards, it was not much okay.

407
00:53:25.289 --> 00:53:28.409
Thank you. I was worried.

408
00:53:29.849 --> 00:53:33.119
Other than that, I encourage you to watch the quantum.

409
00:53:34.590 --> 00:53:45.599
Okay, so please Expander, John 2006 dollars. Okay.

410
00:53:49.530 --> 00:53:53.489
That's like 3 meals or something on the meal plan.

411
00:53:53.489 --> 00:53:58.230
Oh, boy. Okay. That was more. Okay.

412
00:53:59.250 --> 00:54:03.480
Please 1.

413
00:54:03.480 --> 00:54:08.670
Please watch. Okay.

414
00:54:11.369 --> 00:54:14.429
How should we submit the paper?

415
00:54:15.539 --> 00:54:19.320
Well, if it's up to, I don't know.

416
00:54:19.320 --> 00:54:22.559
10 or 20 megabytes email me.

417
00:54:24.659 --> 00:54:28.650
You know, whatever, some 20 megabytes.

418
00:54:30.659 --> 00:54:41.579
Silence whatever.

419
00:54:46.860 --> 00:54:56.369
Silence.

420
00:54:58.920 --> 00:55:02.760
Silence.

421
00:55:07.260 --> 00:55:16.110
Oh.

422
00:55:16.110 --> 00:55:19.199
Let me it was fine. Sure.

423
00:55:22.320 --> 00:55:25.530
Grades go great scope is fine. Um.

424
00:55:26.940 --> 00:55:30.840
I just thought that people might have stuff that wasn't a PDF file or whatever.

425
00:55:30.840 --> 00:55:35.519
Yeah, PDF is preferable actually.

426
00:55:35.519 --> 00:55:41.639
Silence.

427
00:55:41.639 --> 00:55:46.679
Other things quantum computing I'm seeing it in the news now business press.

428
00:55:46.679 --> 00:55:50.550
The Fortune magazine or something, and an article on it. So.

429
00:55:50.550 --> 00:55:55.889
It's getting the attention of people outside the technology companies, so.

430
00:55:59.070 --> 00:56:07.199
Any any other questions or comments on the course.

431
00:56:07.199 --> 00:56:11.070
What would you like to see better next time? So.

432
00:56:11.070 --> 00:56:14.130
Silence.

433
00:56:20.670 --> 00:56:28.170
Any ideas.

434
00:56:30.030 --> 00:56:35.519
Nothing.

435
00:56:39.480 --> 00:56:45.119
Okay, yeah, I.

436
00:56:50.250 --> 00:56:55.409
I'm trying not to overwork you and so.

437
00:56:55.409 --> 00:57:02.309
Thank you. Okay.

438
00:57:04.409 --> 00:57:17.519
Or structure.

439
00:57:17.519 --> 00:57:23.820
On it.

440
00:57:25.619 --> 00:57:31.679
Okay.

441
00:57:40.889 --> 00:57:51.360
More programming site is the.

442
00:58:00.389 --> 00:58:06.869
Good.

443
00:58:15.960 --> 00:58:33.090
Silence.

444
00:58:49.590 --> 00:58:59.639
Silence.

445
00:59:01.409 --> 00:59:12.210
Silence.

446
00:59:18.389 --> 00:59:53.159
Silence.

447
01:00:07.320 --> 01:00:10.769
Silence.

448
01:00:13.199 --> 01:00:31.320
Silence.

449
01:00:39.090 --> 01:00:42.269
Silence.

450
01:01:40.860 --> 01:01:44.309
Silence.

451
01:01:55.619 --> 01:01:59.489
Silence.

452
01:02:04.739 --> 01:02:11.940
Silence.

453
01:03:53.579 --> 01:03:57.389
Silence.

454
01:04:07.289 --> 01:04:11.760
Silence.

455
01:04:13.469 --> 01:04:16.860
Silence.

456
01:04:16.860 --> 01:04:43.590
Silence.