WEBVTT

1
00:00:13.079 --> 00:00:22.230
Okay, let's see.

2
00:00:24.960 --> 00:00:28.769
Silence.

3
00:00:30.629 --> 00:00:39.600
Good afternoon class.

4
00:00:39.600 --> 00:00:42.689
Assuming people can hear him.

5
00:00:42.689 --> 00:00:49.649
If someone can you hear me if you could just post a comment on chat telling me that you can hear me.

6
00:00:49.649 --> 00:00:58.710
Thank you. Thanks, Joseph. So cool. So what is happening today? Let me just pull up the web page.

7
00:03:07.080 --> 00:03:11.400
Okay, well, give me a minute while I rebuilt my laptop.

8
00:03:11.400 --> 00:03:15.810
Assuming that this is coming through and.

9
00:03:16.979 --> 00:03:22.259
Here so we have a refresher, she giving us a 10 or 15 minute talk.

10
00:03:22.259 --> 00:03:25.259
At 5 o'clock on some physics aspect of.

11
00:03:25.259 --> 00:03:28.500
Quantum computing, and then for the rest of.

12
00:03:28.500 --> 00:03:35.069
Today or 1st, another thing is that the next 3 classes are going to be.

13
00:03:35.069 --> 00:03:41.699
You guys talking presentation I put up the on the website for today, the schedule. So.

14
00:03:41.699 --> 00:03:50.879
Equally divided the class in 3 days on Thursday, the next Monday and next Thursday, and satisfies everyone's requests.

15
00:03:50.879 --> 00:03:56.400
So, for the rest of today, after professor, she's talk.

16
00:03:56.400 --> 00:04:00.780
Just so just take a 2nd or 2.

17
00:04:10.050 --> 00:04:19.110
Just trying to boot in and.

18
00:04:20.279 --> 00:04:24.629
Because it's more useful if I can manage to get the.

19
00:04:24.629 --> 00:04:30.689
My main laptop running, then I can do a better job. What I'm projecting now from is from my.

20
00:04:30.689 --> 00:04:33.899
iPad, but.

21
00:04:33.899 --> 00:04:39.298
I can do it better from the main laptop or making it started when professor.

22
00:04:39.298 --> 00:04:45.478
Um, professor, she is talking okay for the rest of today. I've got 2 sets of videos for you.

23
00:04:45.478 --> 00:04:51.718
Show as much as time allows the 1st, 1 gets down at the low level, because we've been talking about quantum computing.

24
00:04:51.718 --> 00:04:59.668
All semester, we haven't gone into too much detail about how quantum computers actually worked down at the very low level.

25
00:05:00.324 --> 00:05:14.694
So, I've got several videos from something called the Q tech Academy, which is actually available on. I'm talking about a couple of different architectures, the trends on Cuba in more detail, because we've talked around and we haven't gotten down to some videos on that. Some videos on.

26
00:05:14.968 --> 00:05:20.608
Spin quantum computer, so you can you tell you more down to reality.

27
00:05:20.608 --> 00:05:28.139
And then I got a call in the middle of video browsing to the Web site, which talks about our quantum Internet.

28
00:05:28.704 --> 00:05:43.613
Where you the idea is that you've got computers in different places and you and tangled the 2 quantum computers. So they're working as a unit and of course, you're getting exponentially more powers is something IBM is working on. So, I got a little video on all these videos, like 3 to 10 minutes long.

29
00:05:46.168 --> 00:05:50.278
Okay, um, let's see.

30
00:05:52.949 --> 00:05:57.389
Oh, good. Then you will like the 1st set of Cubics then. Thanks keep the feedback coming.

31
00:05:57.389 --> 00:06:05.459
And then the last couple of videos I'll be talking about a high level applications, um, some quantum computing for.

32
00:06:05.459 --> 00:06:09.028
Some chemistry, and also for.

33
00:06:10.199 --> 00:06:21.928
Printer that's and so on. So I hit you at the very low end and at the very high end with something in the middle and let me just take a minute or 2. I like I said, this will just take.

34
00:06:21.928 --> 00:06:26.639
To get my real computer going.

35
00:06:27.899 --> 00:06:31.559
So, I can show that I needed to show the videos that Chile. So.

36
00:06:32.819 --> 00:06:45.569
And.

37
00:07:22.439 --> 00:07:25.439
The problem is that.

38
00:07:25.439 --> 00:07:31.829
Mounting my system is not completely automatic, so.

39
00:08:12.899 --> 00:08:16.588
Annoying.

40
00:08:18.209 --> 00:08:28.889
Silence.

41
00:08:30.149 --> 00:08:34.558
Me a 2nd here. Okay.

42
00:08:36.359 --> 00:08:41.458
We're almost there.

43
00:08:47.639 --> 00:09:00.778
The problem is that my mount my booting process is complicated enough. I haven't fully automated that it's because.

44
00:09:00.778 --> 00:09:04.649
It's interacting with new things, you know.

45
00:09:04.649 --> 00:09:08.129
And then when you've got my system set up, so.

46
00:09:11.369 --> 00:09:14.399
The issue, so.

47
00:09:24.443 --> 00:09:47.423
Eva,

48
00:09:47.423 --> 00:09:48.144
okay.

49
00:09:53.489 --> 00:10:01.558
Start the meeting.

50
00:10:10.109 --> 00:10:14.879
Silence.

51
00:10:14.879 --> 00:10:18.958
Sure.

52
00:10:20.158 --> 00:10:24.389
Entire screen.

53
00:10:24.389 --> 00:10:34.499
Silence.

54
00:10:47.038 --> 00:10:58.408
Okay, good. Okay. So I have a link here to refresh your shoes page here. Welcome to look at that. So, those of you who, like, low level things.

55
00:10:58.408 --> 00:11:01.438
Let's scale up to electronics.

56
00:11:01.438 --> 00:11:07.678
Great. And until he.

57
00:11:07.678 --> 00:11:14.009
Connects in, but I would like to show you. I'll pick 1 of the shorter videos.

58
00:11:14.009 --> 00:11:21.899
Hey, rough. I mean, I'll beautiful Thank you. So, if you can, I'll stop sharing my screen and you can.

59
00:11:21.899 --> 00:11:26.249
Share your screen and then the floor is yours.

60
00:11:26.249 --> 00:11:31.469
So, okay, let me try, uh.

61
00:11:31.469 --> 00:11:38.489
I'm curious to my screen now.

62
00:11:38.489 --> 00:11:41.519
Yes, I do. Okay, great.

63
00:11:43.078 --> 00:11:48.629
And on my iPad, its full screen, so you are beautiful can hear you your screen.

64
00:11:48.629 --> 00:12:01.078
Great. Great. Great. Hi everyone I'm really honored have opportunity to describe to you what I do in and, uh, the connection to contact information and technology.

65
00:12:01.078 --> 00:12:15.418
I know everyone is mostly looking for this relevance to quantum computing, but I will also take a little bit time to describe what I do here on and how we make.

66
00:12:15.418 --> 00:12:20.009
Connect the thought to eventually, you know, continue permission to acknowledge.

67
00:12:20.009 --> 00:12:25.739
So, my name is and assistant professor in the chemical and biological engineering department.

68
00:12:25.739 --> 00:12:32.969
My lab is mostly focused on optics and electronics in 2 dimension.

69
00:12:32.969 --> 00:12:37.229
In particular that interesting is still called a 2 dimensional material.

70
00:12:37.229 --> 00:12:50.754
And what is the dimension material? It's some material that you can think down to a extremely thing is, you know, sub pedometers the famously long coughing and probably everyone's know that by now.

71
00:12:50.994 --> 00:12:56.693
So it's a 1 model. A carbon atoms are arranged in this Honeycomb structure.

72
00:12:58.313 --> 00:13:12.024
Our favorite material is this so called M X2 it's a traditional call tonight. Typically people label it as a m. X to the end could be a Mulligan, Adam or counseling. Adam the X could be a software or stadium.

73
00:13:12.958 --> 00:13:27.119
So, it's a beautiful build material. Long of the important thing our mentioned later is that this is a semi conductor, unlike roughing does not have a bank gap. This semi conductor actually can be used for.

74
00:13:27.119 --> 00:13:32.698
Factor of this kind of device we also have BOD may try, which is later.

75
00:13:32.698 --> 00:13:40.918
So what's surprising is that in 2 dimension, you can almost have everything you have in bulk material.

76
00:13:40.918 --> 00:13:45.509
And, uh, there's a certain amount of manage, come ways to reduce the dimension.

77
00:13:45.509 --> 00:13:56.668
For example, with this Max to this is a 2 dimensional semiconductor the atomic structure is shown here the bulk material come with this later structure.

78
00:13:56.668 --> 00:14:09.203
But if you can think down this to this 1 model layer, meaning that to these 3 atoms here, that's a 1 layer. Melissa Adam capital, sandwiched between 2 South atoms.

79
00:14:09.504 --> 00:14:20.724
We have this nice monitor assembly, come back to, which we can easily making 2 factor and this type of device using standard profit techniques.

80
00:14:21.114 --> 00:14:27.203
And what do you have is a typical, only off ratio as shown here.

81
00:14:27.538 --> 00:14:30.719
Uh, for this, Ryan can be as high as tend to the age.

82
00:14:30.719 --> 00:14:35.099
Uh, versus, you know, uh, tend to the 1.

83
00:14:35.099 --> 00:14:49.379
In grasping, and this large 1 off ratio is important to keep the efficiency of the factor of this device. And as you may know the frictionless team, essentially what to make your.

84
00:14:50.634 --> 00:15:04.283
But, for me, uh, my, my f, t's, uh, we also, uh, study electronics, but we are particularly interesting. The so called a photo opt to electronics meaning how does the material impact with a light.

85
00:15:05.303 --> 00:15:12.203
This material is also very special in term it's interaction with light meaning if you're shine light on this material,

86
00:15:12.714 --> 00:15:17.724
normally in traditional semiconductor temporary to see a natural full pair and they associate,

87
00:15:17.724 --> 00:15:18.443
you have the actual,

88
00:15:18.443 --> 00:15:20.573
the whole it get turned out.

89
00:15:20.604 --> 00:15:22.073
That's how it get sold ourselves.

90
00:15:22.408 --> 00:15:32.698
But thing to mention, because it's so thing, the electric field lines are mostly outside the material, even vacuum that means the screening is reduced.

91
00:15:32.698 --> 00:15:45.984
Or the effect of that constant is more as a result, the taxing extremist job. So the lateral hole bound together from this lateral whole pair called X1 way to understand this is that's released.

92
00:15:45.984 --> 00:15:56.244
So called a hydrogen model where you have this negative charge, the particle circling around passive particle, but they stay together for a chat neutral calls hypothetical.

93
00:15:57.269 --> 00:16:05.009
And this has a significant consequence in applications this way you enable. So called this.

94
00:16:05.009 --> 00:16:05.399
Uh,

95
00:16:06.264 --> 00:16:10.224
a photo up to electronics application,

96
00:16:10.254 --> 00:16:12.803
based on X touch so this,

97
00:16:13.014 --> 00:16:13.283
all,

98
00:16:13.283 --> 00:16:18.533
it's a traditional opt electronics application that you can be more efficient,

99
00:16:18.563 --> 00:16:20.244
be more fast with this material.

100
00:16:20.668 --> 00:16:28.198
But it was fascinating, or what's relevant to with quantum computing, is that this material have another quantum degree freedom.

101
00:16:28.198 --> 00:16:32.879
Call a vat is being Here's I'll try to explain to here.

102
00:16:32.879 --> 00:16:45.808
So, you'll see this atomic structure of this material. You can use a honeycomb structure, right? If I project all the 3 atoms layered into 1 layer, you have this kind of structure.

103
00:16:45.808 --> 00:17:00.178
And he can go to this, so called reciprocal lattice, which is quite momentum space. You'll also see this a periodic structure with this symmetry. Right?

104
00:17:00.178 --> 00:17:03.629
So this is what we call a brilliant zone.

105
00:17:03.629 --> 00:17:11.999
Emily, you can calculate this bad structure of this material access corners.

106
00:17:11.999 --> 00:17:21.298
And it looks like these, it must be to complicate let me go to to dimension and explain. What does that mean?

107
00:17:21.298 --> 00:17:24.719
So this is your traditional conduction ban, right?

108
00:17:24.719 --> 00:17:29.308
As, you know, it's the semiconductor conductor, the man.

109
00:17:29.308 --> 00:17:40.588
But this is a span as you can see that they have different things because I speak up coupling. This material is very large as result. You have this large spinning gap here.

110
00:17:40.588 --> 00:17:47.729
Uh, so this 1 is being up, this was being done, but if you go to the add of Valley, it's that's being done. Speeding up. It's opposite.

111
00:17:47.729 --> 00:17:55.348
What does the valley mean? So you'll see this, uh, 6 corners of it says 6 caladis right?

112
00:17:55.348 --> 00:18:00.838
So the 3 of them are equivalent. The other 3 are equivalent.

113
00:18:00.838 --> 00:18:07.828
So, if the 2 adjacent 2, 1, these 2, or we call K plus K minus or K and Prime.

114
00:18:07.828 --> 00:18:17.578
They look very similar if you look at that structure, in terms of the bank gap, it's the exactly the same averages fade but if you're looking to details.

115
00:18:17.578 --> 00:18:24.719
This trying to speak up this 1 is being done, so that different in terms of spin configuration.

116
00:18:24.719 --> 00:18:39.239
And this resulting a different response to light, for example, this 1 can be excited that we so right so comprised lights that stick them up plus and this 1 can be prob, absorbed with a lot of surprise light.

117
00:18:39.239 --> 00:18:43.108
So that means with a different light.

118
00:18:43.108 --> 00:18:47.878
You can actually access whether it's from this Valley.

119
00:18:47.878 --> 00:18:55.979
Off from the neighboring value, I'm not going to get into details, but if you write down the habit, Tanya.

120
00:18:55.979 --> 00:19:04.528
The behavior of this valley is exactly like the speed of the electrons. That's why quite a valley spin off pseudo speed.

121
00:19:04.528 --> 00:19:13.048
Well, if it's being, you can't use that for a superposition of 2 state.

122
00:19:13.048 --> 00:19:20.159
If, uh, to speak are coherent and then that's your 0T and 1 stage right? But.

123
00:19:20.159 --> 00:19:30.868
If that speaks data coherent, we can basically using once being plus other along to construct any, any configuration you want.

124
00:19:30.868 --> 00:19:35.219
A manager to that this is so called a box field.

125
00:19:35.219 --> 00:19:40.318
Imagine if 1 of us being is a point up that's your last.

126
00:19:40.318 --> 00:19:47.278
If another thing is going down with a passport, if these 2 are same attitude, right?

127
00:19:47.278 --> 00:20:00.269
You can construct a, uh, a Cape speak up plus being a, uh, a spin down that's actually a spade in this, uh, equator state.

128
00:20:00.269 --> 00:20:06.778
It came, magic can do this any combination of this too. That's essentially your spin state, right?

129
00:20:06.778 --> 00:20:15.148
So, if you have such a coherent spring state, that's essentially your quantum cube.

130
00:20:15.148 --> 00:20:18.778
That's essentially what to make the quantum computing possible.

131
00:20:18.778 --> 00:20:24.148
Um, a lot of efforts in the field is actually tried to construct.

132
00:20:24.148 --> 00:20:29.398
And manipulate this a coherent test being States. The reason is that, um.

133
00:20:29.398 --> 00:20:37.229
For whatever reasons nature. Faders right. Uh, so they won't make the sprinkler in lifetime long enough.

134
00:20:38.453 --> 00:20:49.794
People figure out a way to do that if you do use some sort of adjustment talent Junction, we can go to extreme and low temperature 10 minute. Calvin, right?

135
00:20:50.124 --> 00:21:01.884
And that will make the cocreated lifetime long enough so that you can do some speed manipulation, but still generates a very, very difficult to get a spin coca interstate.

136
00:21:02.278 --> 00:21:15.628
Now, Here's come up with my research as I described before now we have 2 Valley speeds right? If you have electron the whole pair, which reflects towns in kale, Prime Valley.

137
00:21:15.628 --> 00:21:28.074
Uh, that behave very differently. They absorb light differently, the mid life differently and in principle is to state. If they're coherent, they looks like the spin.

138
00:21:28.193 --> 00:21:34.104
They can, uh, interfere with each other and the construct and estatey walked. That's essentially another way of doing.

139
00:21:35.249 --> 00:21:46.798
So, my research 1 of my research is to look for this long, coherent, long, a long time state of this valley speed.

140
00:21:46.798 --> 00:21:51.449
For example, like here, if we use extremely powerful laser.

141
00:21:52.253 --> 00:22:02.034
We can use and so cost effective to lift the generously between K and K, Prime Valley before I showed you that the average equivalent. Right?

142
00:22:02.034 --> 00:22:08.874
But if I use extremely strongly the power, I can turn this to state a separate to average of this to.

143
00:22:09.209 --> 00:22:19.858
Intuitively what you can imagine is that it's kind of like having a to spin state here, but now you apply play magnetic field.

144
00:22:19.858 --> 00:22:25.888
Which would break the time reversal symmetry and therefore break the image between this Tuesday.

145
00:22:25.888 --> 00:22:30.479
And if you apply magnetic fields, what happened to us being.

146
00:22:30.479 --> 00:22:37.078
Well, if the spin is a clinical along with the field, energy is going to shift.

147
00:22:37.078 --> 00:22:45.538
But in this case here, right? If you start ways, equal superposition of a speed up and down so you have a state.

148
00:22:45.538 --> 00:22:51.689
Along this equate to state now, if you apply magnificent field to that.

149
00:22:51.689 --> 00:22:56.278
What you have is you happen to have a big state rotating.

150
00:22:56.278 --> 00:23:00.628
Alone this equator, but the penny got to make a few.

151
00:23:00.628 --> 00:23:07.439
So that's another way to detect the coherence time of this.

152
00:23:08.394 --> 00:23:22.614
So, if you can measure the response of this, a rotation as a function of the magnetic, a few effective and nephew, which is the laser house, you can see how this rotation dictating type.

153
00:23:22.949 --> 00:23:28.439
And those a timescale give you a sense of how coherent to these 2 States are.

154
00:23:28.439 --> 00:23:32.969
So, unfortunately, this 2 States are still fairly short lived.

155
00:23:32.969 --> 00:23:43.739
Um, we go down to a 4, Calvin, it's on the order 1 nano 2nd for any meaningful, uh, operation. We probably wanted to be as long as the 2nd.

156
00:23:43.739 --> 00:23:50.038
Or, even 2nd, 1, possible way is to go down to even a lot of temperature, like, you know.

157
00:23:50.038 --> 00:23:56.459
What to IBM and Google does go down to mimic Kelvin alternative way is to use other Nob.

158
00:23:56.459 --> 00:24:01.798
To somehow propel a state or force them to a, to be face.

159
00:24:01.798 --> 00:24:08.969
And, uh, that's what we are working on in lab right now. There's a few candidate can do that.

160
00:24:08.969 --> 00:24:12.719
Um, but it's, uh, it's, uh, it's promising, but their ongoing.

161
00:24:13.798 --> 00:24:17.729
That's pretty much what I have.

162
00:24:17.729 --> 00:24:23.939
Do you have any questions? Oh, thank you. Very much.

163
00:24:23.939 --> 00:24:27.778
I'll lead all false classing your questions so.

164
00:24:27.778 --> 00:24:30.023
Who are the major funding agencies for this work?

165
00:24:31.733 --> 00:24:46.644
I mean, for us, or for this field in general field in general, if students want to say, go to grad school and this and oh, it's everyone's deal and not of course. Uh.

166
00:24:47.669 --> 00:24:55.798
Ever ever major, uh, agent of like a falls Navy army. They are all founding this, uh, field.

167
00:24:57.419 --> 00:25:05.939
All right, so the class jobs, when you get out yeah, I mean, it's, it's a, it's a very, uh, hot field.

168
00:25:05.939 --> 00:25:10.499
But it also me, it's a very competitive field. We are competing with, you know.

169
00:25:10.499 --> 00:25:14.848
The leading Institute leading groups in the world.

170
00:25:15.534 --> 00:25:25.163
So, it's you need really need to think hard to see what you can bring to the table. Like, what's making your research, uh, different.

171
00:25:25.463 --> 00:25:31.044
So, for us, our advantage is that we figure out to to propel this material.

172
00:25:31.348 --> 00:25:34.979
So that we can get extremely clear on material.

173
00:25:34.979 --> 00:25:39.808
So, we can show all this intrinsic properties instead of, uh, the.

174
00:25:39.808 --> 00:25:46.229
Effective from defects we have a way to manipulate to this state.

175
00:25:46.229 --> 00:25:55.048
But again, no, it's a very competitive field. Uh, there's a lot of opportunities and there's a lot of competition as well.

176
00:25:55.048 --> 00:26:00.778
Oh, thank you. Does anyone else have any questions? Just make you unmute yourself and ask.

177
00:26:00.778 --> 00:26:05.638
No.

178
00:26:05.638 --> 00:26:09.628
John, you like low level stuff do you have a question.

179
00:26:10.798 --> 00:26:14.098
Now.

180
00:26:14.098 --> 00:26:17.818
Okay, well thanks again and.

181
00:26:17.818 --> 00:26:20.818
See, not not sure yet. Okay.

182
00:26:20.818 --> 00:26:24.989
So, you the next meeting virtually I see. You have fun.

183
00:26:24.989 --> 00:26:30.959
Okay bye so.

184
00:26:31.979 --> 00:26:36.419
Let me just this.

185
00:26:38.638 --> 00:26:47.519
Cool. Okay so I want to show the spin cube videos. What I'll do is, I'll show the.

186
00:26:47.519 --> 00:26:59.038
Cubic because we've talked about trends moms a lot, but we haven't got enough detail. And I actually think the stuff on the website is not as good as the academy website. So.

187
00:26:59.038 --> 00:27:02.278
Oh, wow.

188
00:27:02.278 --> 00:27:05.818
See.

189
00:27:07.348 --> 00:27:12.239
Sure.

190
00:27:15.749 --> 00:27:19.648
In this lecture, I will explain.

191
00:27:19.648 --> 00:27:25.288
How we can perform measurement of superconducting Cubans at the end of every quantum algorithm.

192
00:27:25.288 --> 00:27:29.818
The computation result has to be obtained by performing measurement of the Cuba.

193
00:27:29.818 --> 00:27:33.269
And that's dictated by the loss of quantum mechanics.

194
00:27:33.269 --> 00:27:37.828
Any superposition States are projected into well defined zeros and ones.

195
00:27:37.828 --> 00:27:41.189
A transport Cuba as the effective hair.

196
00:27:41.189 --> 00:27:44.548
Can be measured by I read a resume that is covered to it.

197
00:27:44.548 --> 00:27:49.949
The residence frequency of the resonator is quite far away from the cubic transition frequency.

198
00:27:49.949 --> 00:27:56.788
On the order of gigahertz, however, due to the coupling, there is a shift in the resonate. The frequency.

199
00:27:56.788 --> 00:28:03.088
Depending on the current state, the shift is evidence here by the measured transmission tips.

200
00:28:03.088 --> 00:28:07.108
Both the Cuban in the ground States and the excited state.

201
00:28:07.108 --> 00:28:11.848
The shift is typically on the order of a few megahertz.

202
00:28:11.848 --> 00:28:15.959
3 orders of magnitude smaller than the attorney.

203
00:28:15.959 --> 00:28:19.888
We can observe the shift and thereby the cubic States.

204
00:28:19.888 --> 00:28:23.909
By injecting the resonator with a pulse near the resonate to frequency.

205
00:28:23.909 --> 00:28:28.679
Propose a show in here is reflected by the resonate.

206
00:28:28.679 --> 00:28:34.888
Here we show the upper voltage as a function of time for the Cuba in the ground state, and the excited state.

207
00:28:34.888 --> 00:28:41.669
Which clearly look different and we are able to distinguish and measure the 2 different human States.

208
00:28:43.679 --> 00:28:48.058
The traces, I've just shown you look very clean and distinguishable.

209
00:28:48.058 --> 00:28:51.808
As they are average over thousands of measurements, however.

210
00:28:51.808 --> 00:28:57.868
For most real problem protocols we required to discern the Cuban state in just a single run.

211
00:28:57.868 --> 00:29:00.868
But to avoid disturbing, our precious quantum system.

212
00:29:00.868 --> 00:29:04.409
We can only use reinforces the consist of a couple of photos.

213
00:29:04.409 --> 00:29:08.909
At this power levels, in the order of just the few filter, what.

214
00:29:08.909 --> 00:29:14.429
We are struggling to distinguish the different quantum states as the signal is hampered by quantum.

215
00:29:14.429 --> 00:29:17.878
Depicted here to qualify.

216
00:29:17.878 --> 00:29:21.509
How well, Cupid measurement is performing in the presence of this noise.

217
00:29:21.509 --> 00:29:25.979
Record the integrated footage of thousands of individual traces.

218
00:29:25.979 --> 00:29:29.128
I brought in these individual shelves in.

219
00:29:29.128 --> 00:29:40.558
We extract the fidelity of the measurements. The fidelity expresses the probability that the measurement returns the right outcome average over to.

220
00:29:40.558 --> 00:29:45.419
Possible here with input States. Absolute 0, 1 expresses.

221
00:29:45.419 --> 00:29:49.919
Program to take for erroneously, getting outcome 1 for a ground site input.

222
00:29:49.919 --> 00:29:54.749
An, absolute 1 0T vice versa in this case.

223
00:29:54.749 --> 00:30:01.019
And by the world's lowest noise, amplifiers, we can achieve a fidelity close to 98%.

224
00:30:02.759 --> 00:30:06.058
I've just shown you read out of 1, kill it.

225
00:30:06.058 --> 00:30:09.148
Of course, a full quantum computer.

226
00:30:09.148 --> 00:30:14.429
Consist of many continents and these condiments all have to be read out at the same time.

227
00:30:14.429 --> 00:30:20.638
To allow the simultaneous readout of each Cuba to its own real resonator.

228
00:30:20.638 --> 00:30:26.068
Uh, choose the length of the resonate as such that they resonate all of their distinct frequency.

229
00:30:26.068 --> 00:30:29.939
Just like the different strengths in a piano.

230
00:30:32.159 --> 00:30:35.638
My next sending down a post that is the sum of multiple components.

231
00:30:35.638 --> 00:30:39.749
In this case to each to its own.

232
00:30:39.749 --> 00:30:43.138
We can probe multiple Richard resonators at the same time.

233
00:30:43.138 --> 00:30:46.348
Each component will only be picked up.

234
00:30:46.348 --> 00:30:50.068
By the resonate though, and the analysis of the output signals.

235
00:30:50.068 --> 00:30:53.519
We again separate the 2 different frequency components.

236
00:30:54.929 --> 00:30:58.229
Yeah, we showed up ourselves for Kim with 1.

237
00:30:58.229 --> 00:31:01.769
For each of the 4 possible to Cuba basis States.

238
00:31:01.769 --> 00:31:07.138
Indicating that the signal practically only depends on the state of the target and Cubans.

239
00:31:07.138 --> 00:31:10.558
Separate things 0, 0T and 1. 0.

240
00:31:10.558 --> 00:31:14.398
From 0, 0T and 1, 1, Julie.

241
00:31:14.398 --> 00:31:19.858
We here show rid of results for the other Cuba, which separates 0, 0T and 1. 0.

242
00:31:19.858 --> 00:31:31.378
From 0, 0T and 1, 1, you might now be able to imagine how this can be extended to measure ever increasing numbers of.

243
00:31:31.378 --> 00:31:34.679
The fact that the 17 device.

244
00:31:34.679 --> 00:31:41.669
Where we play the simultaneous using 17 divided over 3 different input and output.

245
00:31:41.669 --> 00:31:45.148
This might sound to you like a finished story.

246
00:31:45.148 --> 00:31:49.949
However, there are many approvals to be made still to get to a fully.

247
00:31:49.949 --> 00:31:55.499
Scalable quantum computer exemplary topics that are currently being addressed in research.

248
00:31:55.499 --> 00:31:59.519
And which might also be attractive for you for additional reading.

249
00:31:59.519 --> 00:32:04.469
Advanced reader for shaping to quickly populate and populate.

250
00:32:04.469 --> 00:32:09.058
The readout resonators to speed up the read out.

251
00:32:09.058 --> 00:32:13.858
Server connect the Apple fires to suppress noise in the rid of amplification chain.

252
00:32:13.858 --> 00:32:18.118
And methods to quantify that performance.

253
00:32:18.118 --> 00:32:21.628
We work on a phone screen apologies in this case.

254
00:32:21.628 --> 00:32:25.439
Using multiple cascade agreed read address per.

255
00:32:25.439 --> 00:32:29.878
Increase the flux of photons and furnish speed up the, that.

256
00:32:29.878 --> 00:32:38.249
Okay, well that was the measurement video.

257
00:32:38.249 --> 00:32:41.429
Chilly my.

258
00:32:41.429 --> 00:32:47.459
Should show the other 2nd or the trans more about how they work inside. So, this will still be interesting.

259
00:32:47.459 --> 00:32:51.838
Can you hear.

260
00:32:51.838 --> 00:32:56.459
Right.

261
00:32:56.459 --> 00:33:06.239
Today I will give you an introduction to quantum computing, the superconducting quantum circuits.

262
00:33:06.239 --> 00:33:13.499
More specifically quantum computing with transplant quantum bits or Cubics in the circuit quantum electric dynamics. Architecture.

263
00:33:15.028 --> 00:33:21.179
Superconducting Cubans differ in 2 important ways from truly quantum 2 level systems provided to us by nature.

264
00:33:21.179 --> 00:33:24.868
Such as the spin of an electron or the span of certain nuclei.

265
00:33:24.868 --> 00:33:29.159
Like, carbon, 13 or Silicon 30th.

266
00:33:29.159 --> 00:33:33.239
Their multi level systems, not all the electronic levels in atoms.

267
00:33:33.239 --> 00:33:44.009
Multi level upon them systems can be used effectively as Cuban by confining all dynamics to 2 quantum left. Usually the ground and the 1st, excited state of the system.

268
00:33:45.058 --> 00:33:48.568
2nd, these cubes are circuits that we fabricate ourselves.

269
00:33:48.568 --> 00:33:53.939
This has advantages and disadvantages on the bright side. We have freedom to design.

270
00:33:53.939 --> 00:33:57.838
In a sense, we can play God designing artificial atoms.

271
00:33:57.838 --> 00:34:02.759
However, fabrication, uncertainty keeps us from making any to Cuba. It's the same.

272
00:34:02.759 --> 00:34:07.348
For this reason, you'd like to say that Cubans have individual personality.

273
00:34:07.348 --> 00:34:12.268
Well, the form of the Hamilton describing them content mechanically is well known.

274
00:34:12.268 --> 00:34:17.039
The parameters of this Hamilton can now be perfectly targeted in fabrication.

275
00:34:17.039 --> 00:34:22.648
This is interesting implications for scaling up our quantum processors, which I will discuss in a later video.

276
00:34:22.648 --> 00:34:32.548
But back to design, superconducting Cubics, generally consist of superconducting electrodes or islands that are interconnected by Joseph's injections.

277
00:34:32.548 --> 00:34:36.628
Tony, and typically consists of 2 now commuting contributions.

278
00:34:36.628 --> 00:34:42.539
1, capacitive and 1 and the 1st turns to localize the Cooper pairs.

279
00:34:42.539 --> 00:34:47.338
And the 2nd, favorite's their tunneling from 1 island to another. In other words.

280
00:34:47.338 --> 00:34:52.079
They're hoping there exists numerous varieties of superconducting Cubans.

281
00:34:52.079 --> 00:34:56.909
The charged cubic flux, Cupid face, cubic sodium.

282
00:34:56.909 --> 00:35:02.338
And many more these Cubans differ in terms of the number of superconducting islands.

283
00:35:02.338 --> 00:35:08.938
The number of junctions, and also quite importantly, the relative energy scales of the capacitive and inductive terms.

284
00:35:08.938 --> 00:35:15.059
Typically superconducting Cuba, it's working the frequency range between foreign and gigahertz approximately.

285
00:35:15.059 --> 00:35:18.389
This frequency, let's call it 0 1.

286
00:35:18.389 --> 00:35:22.648
Is related to via planck's constant to the energy difference. Easy 1.

287
00:35:22.648 --> 00:35:26.398
Between the quantum levels that we assign a stage 0T and 1.

288
00:35:27.628 --> 00:35:34.228
0, 1 is also the frequency of the micro process that we will need to induce comparing transitions between these levels.

289
00:35:35.579 --> 00:35:40.349
In this course, we will focus on the transport, a derivative of the charged cubic.

290
00:35:40.349 --> 00:35:44.759
The transplant is the superconducting Cuba that we specialize on the.

291
00:35:47.818 --> 00:35:52.768
In its simplest form, the transmen consists of 2 islands interconnected by 1 junction.

292
00:35:52.768 --> 00:36:03.208
Those of you familiar with circuits, particularly the electrical engineers will recognize that the transplant looks. Well, just like a parallel combination of 1 capacitor at 1 doctor.

293
00:36:03.208 --> 00:36:09.239
In other words, NLC also later this gets us most of the way there. So let's take a look.

294
00:36:12.208 --> 00:36:17.579
The Hamilton for the later consists of 2 terms that are each quite dramatic with respect to 1 variable.

295
00:36:17.579 --> 00:36:21.958
Capacitive term is quadratic on the charge accumulated on 1 island.

296
00:36:21.958 --> 00:36:28.498
The opposite charges accumulated India inductive term is quadratic on the flex to the doctor.

297
00:36:28.498 --> 00:36:34.079
Discharge and this flux do not commute. In fact, they're economically conjugate variables.

298
00:36:34.079 --> 00:36:41.278
A direct correspondence between this Hamilton can be made to that of the simple harmonic oscillator by mapping.

299
00:36:41.278 --> 00:36:44.818
Flux to the position of the mass and charge.

300
00:36:44.818 --> 00:36:56.159
To the masses momentum, physics students will not be surprised to learn that the spectrum of the month. I still see, oscillator is perfectly harmonic levels are equally spaced in energy.

301
00:36:56.159 --> 00:37:00.809
This equal spacing is given by a familiar formula. 1 overrule Elsie.

302
00:37:00.809 --> 00:37:05.278
But unfortunately a harmonic spectrum, there's not a good to make.

303
00:37:05.278 --> 00:37:09.119
It is very difficult to confine the dynamics to just 2 levels.

304
00:37:09.119 --> 00:37:13.108
So, leakage out of the cubic space is a permanent threat.

305
00:37:13.108 --> 00:37:18.148
That is why the transport differs from the LLC also later in a fundamental way.

306
00:37:18.148 --> 00:37:22.588
In the transplant inductance is provided by a Joseph's injunction.

307
00:37:22.588 --> 00:37:28.949
And not by a typical coil doctor inductive energy for the junction is not quadratic.

308
00:37:28.949 --> 00:37:32.639
But, of course, I can function of the generalized flux through it.

309
00:37:33.869 --> 00:37:36.989
This difference has important consequences for the spectrum.

310
00:37:36.989 --> 00:37:41.699
Crucially, it disrupts the harmonic spectrum in ways that are very practical.

311
00:37:41.699 --> 00:37:46.588
At typical parameter values deep in the so called.

312
00:37:46.588 --> 00:37:51.148
Where the energy scale of the inductive term is much larger than that of the charging term.

313
00:37:51.148 --> 00:37:55.228
And let's say for a frequency of 0, 1 of about 6 gigahertz.

314
00:37:55.228 --> 00:37:59.518
The transition point too, from the 1st, to the 2nd excited state.

315
00:37:59.518 --> 00:38:03.298
It's lower by approximately 300 megahertz.

316
00:38:03.298 --> 00:38:06.869
This difference is sufficient in practice to confine the dynamics.

317
00:38:06.869 --> 00:38:11.789
To the 2 lowest levels, our cubic phase when performing single cubic Gates.

318
00:38:11.789 --> 00:38:18.420
With policies of duration, about 20 seconds Brian will discuss implementation of single cubic in detail.

319
00:38:19.949 --> 00:38:23.309
As a final note, we build our transplant not from 1.

320
00:38:23.309 --> 00:38:31.559
But 2, Joseph's, conjunctions in parallel, this gives us the possibility to tune the inductive element. And thereby also the Cuba transition frequency.

321
00:38:31.559 --> 00:38:35.730
By threading a magnetic flux through the loop defined by the 2 junctions.

322
00:38:35.730 --> 00:38:40.500
We can do this independently for each Cuba and our 2nd time.

323
00:38:40.500 --> 00:38:44.219
This capability is the war of course, enabling to Cuba. Gates.

324
00:38:44.219 --> 00:38:47.670
Doesn't matter what percent these in the later video.

325
00:38:47.670 --> 00:38:51.329
Okay.

326
00:38:51.329 --> 00:38:57.179
So, I get a little more in trends, mind cube, but look at the 2 cubic Gates. Now.

327
00:38:59.849 --> 00:39:06.780
Sure.

328
00:39:07.829 --> 00:39:15.090
Interactions between Cubans are controls using Tokyo with gets.

329
00:39:15.090 --> 00:39:20.489
In transport Cubans, the temperature with gate that is used to form a universal gate set.

330
00:39:20.489 --> 00:39:25.019
Just the conditional basement also known as the C phase, or sees that.

331
00:39:25.019 --> 00:39:28.920
Applying a conditional the.

332
00:39:28.920 --> 00:39:32.369
Because the target queue to acquire by already interface.

333
00:39:32.369 --> 00:39:36.750
Based on a state of document, as an example.

334
00:39:36.750 --> 00:39:41.849
Let us consider a control Cuban starting and eater.

335
00:39:41.849 --> 00:39:46.079
The ground, or the excited states, and it's our acute.

336
00:39:46.079 --> 00:39:52.440
Team starting in the plus state when the control queue it is in a ground state.

337
00:39:52.440 --> 00:39:57.780
And we apply to see there is no effect on the.

338
00:39:57.780 --> 00:40:02.610
However, when the control Cuba is in the excited state.

339
00:40:02.610 --> 00:40:05.760
The targets Cuba will acquire.

340
00:40:05.760 --> 00:40:11.340
Via radiance, or face in transport C phase gate.

341
00:40:11.340 --> 00:40:15.030
Is implemented by tuning in and out of residence.

342
00:40:15.030 --> 00:40:18.869
With an interaction in a 2 exertation metaphor.

343
00:40:18.869 --> 00:40:22.769
To understand how to process a basket.

344
00:40:22.769 --> 00:40:25.860
You will look at what mediates interaction.

345
00:40:25.860 --> 00:40:30.119
How cute was commissioned in and out of residence wouldn't interaction.

346
00:40:30.119 --> 00:40:36.090
And why this causes the Cubans to accumulate phase how to use this interaction diplomacy phase gate.

347
00:40:36.090 --> 00:40:40.559
Well, minimizing leakage out of the sub space and finally.

348
00:40:40.559 --> 00:40:44.760
What the experimental challenges are, when implementing high fidelity kits.

349
00:40:47.400 --> 00:40:52.800
In superconducting with Gates are based on a transverse OkCupid cubic.

350
00:40:52.800 --> 00:40:56.909
This company is through a coupling resonator.

351
00:40:56.909 --> 00:41:00.510
Because the interaction is mediated bike, open resonator.

352
00:41:00.510 --> 00:41:04.679
It is possible to couple physically separated Cubans, making a room.

353
00:41:04.679 --> 00:41:08.039
For all our components on chip, such as resonators.

354
00:41:09.750 --> 00:41:12.840
To be able to jump in and out of residence with interaction.

355
00:41:12.840 --> 00:41:16.079
The transaction frequency of a transport human must be controlled.

356
00:41:16.079 --> 00:41:19.769
By applying a current to the flux bias, like.

357
00:41:19.769 --> 00:41:23.670
The amount of of the transport changes.

358
00:41:23.670 --> 00:41:26.849
Making it possible to control the QC frequency.

359
00:41:28.110 --> 00:41:33.090
When it goes is applied to the flux by his line here with this detune from its operating frequency.

360
00:41:33.090 --> 00:41:37.469
For the duration of the books while the Cuban is the children.

361
00:41:37.469 --> 00:41:43.949
Face accumulates to understand how this flux control can be used pharmacy phase Gates.

362
00:41:43.949 --> 00:41:49.170
Let us take a look at the level diagram of 2 transport curate.

363
00:41:49.170 --> 00:41:54.119
As a function of a flux true in this level diagram.

364
00:41:54.119 --> 00:41:58.349
Subscribes they know the number of extensions in the control and to it.

365
00:41:58.349 --> 00:42:03.389
Perspective interaction that is used to performance in face.

366
00:42:03.389 --> 00:42:07.110
Isn't avoided crossing between a 1, 1 and Tuesday.

367
00:42:08.460 --> 00:42:12.150
The amount of face that a term cute picks up is most evident.

368
00:42:12.150 --> 00:42:15.780
When to 1, 1 and oh, 1 level diagrams are outlet.

369
00:42:15.780 --> 00:42:19.679
By expressing them in terms of the tuning with respect to their maximum energy.

370
00:42:19.679 --> 00:42:23.039
It can be seen that the majority of the targets.

371
00:42:23.039 --> 00:42:28.440
It's different based on the state of this difference is Mark and red over.

372
00:42:29.849 --> 00:42:34.469
I did Shannon, the QC into the red region is possible to make the target to it.

373
00:42:34.469 --> 00:42:37.559
Acquire a face conditionals that of control.

374
00:42:38.670 --> 00:42:42.480
However, the interaction we are using is an interaction.

375
00:42:42.480 --> 00:42:48.900
Between the 1, 1, and the O2 state to state is not part of the computational subspace.

376
00:42:48.900 --> 00:42:52.440
We want to avoid energy transfer to a state.

377
00:42:53.550 --> 00:42:57.750
This is definitely done using a special spellcheck known as a fast.

378
00:42:57.750 --> 00:43:03.360
India medicals that minimizes the leakage. This spill shaped can be expressed.

379
00:43:03.360 --> 00:43:09.030
As a floria series in the frame of the interaction, which with knowledge of the system parameters.

380
00:43:09.030 --> 00:43:13.800
And we covered it 1st to a frequency and then into a voltage to be applied.

381
00:43:13.800 --> 00:43:19.019
To the flux biased, because to conditional face and the linkage.

382
00:43:19.019 --> 00:43:25.110
The bands on the exact trajectory of the Cuban base to commit gates are highly sensitive.

383
00:43:25.110 --> 00:43:30.659
2 distortions of the bullshit distortions can be caused by electrical bonus in the signal part.

384
00:43:30.659 --> 00:43:37.530
Between away from generator and acute, such as filters and cables, but even the old chip response calls to distortions.

385
00:43:37.530 --> 00:43:43.860
On the signal, the cute experiences these effects are typically corrected by reinserting.

386
00:43:43.860 --> 00:43:48.210
And the way from X with a filter designed to invert assertions.

387
00:43:48.210 --> 00:43:51.900
During into, sort of the queue and experiences.

388
00:43:51.900 --> 00:43:56.340
Hospitals why, why, until it is equal to the intended way.

389
00:43:56.340 --> 00:44:00.179
Challenge flux building best is advocate.

390
00:44:00.179 --> 00:44:03.360
Is to correct release assertions. This requires.

391
00:44:03.360 --> 00:44:06.960
Characterizing distortions that the Cuban experiences when cool down.

392
00:44:06.960 --> 00:44:11.489
And correcting these with sufficiently high precision at the same time.

393
00:44:11.489 --> 00:44:14.670
Efforts are underway to become more resilient against these effects.

394
00:44:14.670 --> 00:44:19.349
Bought back for a new bullshit techniques and true innovations in the hardware.

395
00:44:20.369 --> 00:44:23.519
Thank you for your.

396
00:44:23.519 --> 00:44:29.489
Okay, so we've seen a little more about how things are implemented.

397
00:44:29.489 --> 00:44:35.010
And to a short 1 on some.

398
00:44:35.010 --> 00:44:40.440
Here, so I can do different technology.

399
00:44:42.780 --> 00:44:46.500
All right.

400
00:44:51.750 --> 00:44:55.949
Let's discuss some real life implementations of these.

401
00:44:55.949 --> 00:44:59.789
The most personal platform consists of semiconductor.

402
00:44:59.789 --> 00:45:03.960
Attached to superconducting leads that my rant about states emerge.

403
00:45:03.960 --> 00:45:07.409
And we have the breathing statistics of these animals.

404
00:45:07.409 --> 00:45:12.090
Immediately when the additional pair of these states in the 1 time measure modifier.

405
00:45:12.090 --> 00:45:16.889
Encounter a problem you cannot exchange these 2 particles without colliding them.

406
00:45:16.889 --> 00:45:20.880
Of using them with the measures of quantum state.

407
00:45:20.880 --> 00:45:24.179
And has loses of contact coherence.

408
00:45:24.179 --> 00:45:27.570
The rate to perform breathing is a clever workaround.

409
00:45:27.570 --> 00:45:32.880
If you attach an segment to another wire, you can move this topological object around.

410
00:45:32.880 --> 00:45:39.989
Just so I can do a vital with your car. This allows for breathing of the 2 particles, orange and blue.

411
00:45:39.989 --> 00:45:50.699
Showing the animation since the percent of this additional arm is crucial for this research direction there's actually a lot of development effort going into creating these structures.

412
00:45:50.699 --> 00:45:54.599
A few of these published in scientific literature are shown here.

413
00:45:54.599 --> 00:45:58.949
Let's go on to the front of state measurement in this scheme.

414
00:45:58.949 --> 00:46:03.539
Remember that the power of the States include either 0T electron.

415
00:46:03.539 --> 00:46:07.139
Or 1, electron charge, and these states are.

416
00:46:07.139 --> 00:46:12.059
That is that they are at the same energy level, if you want to distinguish between them.

417
00:46:12.059 --> 00:46:16.739
You have to break this if we close the rest of the wire.

418
00:46:16.739 --> 00:46:20.699
The door stays with off because of the charging and edge of the system.

419
00:46:20.699 --> 00:46:24.030
That's why for content for increments.

420
00:46:24.030 --> 00:46:31.199
Similarly, the chart States can then be measured by charge detectors that are put close to the.

421
00:46:31.199 --> 00:46:36.239
This without scale, relying on the interaction between the electrons in another buyer.

422
00:46:36.239 --> 00:46:42.929
This call interaction based operation what you need for this operation is a set of words.

423
00:46:42.929 --> 00:46:47.760
Which we can open and close that field in real devices.

424
00:46:47.760 --> 00:46:53.969
These are the Gates, which can locally control the flow of electrons inside and on the.

425
00:46:53.969 --> 00:46:59.190
Have a negative voltage, you remove the electrons and has closed the most.

426
00:46:59.190 --> 00:47:03.420
On the other hand, then they apply a positive voltage on 1 of these gates.

427
00:47:03.420 --> 00:47:10.079
You open the 4th and let the electrons flow. You can also use this concept to perform breathing.

428
00:47:10.079 --> 00:47:13.199
So, that you need not to physically move the States.

429
00:47:13.199 --> 00:47:19.260
Yes, why did it before? It starts from a day device with a pair of in each segment.

430
00:47:19.260 --> 00:47:23.610
It's a horse you can control the coupling between each pair.

431
00:47:23.610 --> 00:47:27.630
Of the political segments enabling interaction base branding.

432
00:47:28.769 --> 00:47:33.119
The animation show us how to exchange or grade States 2 and 3.

433
00:47:34.559 --> 00:47:38.190
Importantly, each 1 around the state is always located.

434
00:47:38.190 --> 00:47:41.730
At the end of each segment, then opening of our.

435
00:47:41.730 --> 00:47:47.099
We create a longer topological segment, which settles which apples? The quantum state.

436
00:47:47.099 --> 00:47:50.250
Of the Mariana to the very end of this longer segment.

437
00:47:51.480 --> 00:47:54.690
With this device, we can perform breathing operations.

438
00:47:54.690 --> 00:48:00.719
And they can also perform state measurement. Therefore, you have just cashed prototype the political.

439
00:48:01.800 --> 00:48:06.449
As an Outlook, let's say a few proposed geometries of the political cabinets.

440
00:48:06.449 --> 00:48:10.530
Similar to the 1, we have just discussed as you can see.

441
00:48:10.530 --> 00:48:14.789
There are many different schemes showing that this is a very active research topic.

442
00:48:14.789 --> 00:48:18.900
It remains to be seen which lead to the very 1st.

443
00:48:18.900 --> 00:48:23.010
Experiment demonstration of the project or Cupid and the eventually.

444
00:48:23.010 --> 00:48:28.530
A scalable system of these integrating.

445
00:48:30.300 --> 00:48:38.429
You know, that reminds me of 1 of the apps on my phone 1 of these games or something. You open gates and whatever. Okay.

446
00:48:38.429 --> 00:48:45.449
So those are low level things go up to a higher level thing about connecting distant quantum machines together.

447
00:48:45.449 --> 00:48:50.489
Okay, so.

448
00:48:50.489 --> 00:48:54.929
Sure.

449
00:48:57.989 --> 00:49:03.269
What is the quantum Internet? A quantum Internet.

450
00:49:03.269 --> 00:49:06.719
Provides radically new Internet to technology.

451
00:49:06.719 --> 00:49:12.809
That allows us to solve tasks, which are impossible to accomplish on the classical.

452
00:49:14.070 --> 00:49:20.130
As with any radically new technology, the cannot yet foresee all applications of a comfortable.

453
00:49:20.130 --> 00:49:23.789
But it already has quite a number of exciting ones.

454
00:49:25.469 --> 00:49:30.480
For example, it allows us to do absolutely secure communication.

455
00:49:31.500 --> 00:49:36.150
Secure identification position verification.

456
00:49:36.150 --> 00:49:40.380
Secure dedicated quantum computing in the cloud.

457
00:49:40.380 --> 00:49:44.340
And many others we will visit in other videos.

458
00:49:44.340 --> 00:49:48.750
So, what does the quantum Internet look like?

459
00:49:48.750 --> 00:49:53.159
On a quantum Internet we don't send classical bits.

460
00:49:53.159 --> 00:49:58.349
Zeroes and wants, but I will transmit.

461
00:50:00.719 --> 00:50:07.530
But otherwise the basic elements of a quantum Internet do not look so different from the classical 1.

462
00:50:09.750 --> 00:50:13.320
The 1st, element of a Internet.

463
00:50:13.320 --> 00:50:18.389
Is what we call an end note end note.

464
00:50:18.389 --> 00:50:22.230
It's basically your computer or a laptop or phone.

465
00:50:22.230 --> 00:50:27.719
That is attached to the Internet and that you use in order to run applications.

466
00:50:29.489 --> 00:50:33.780
So, you moved in note in order to use the quantum.

467
00:50:36.329 --> 00:50:39.780
As the name suggests on a quantum Internet.

468
00:50:39.780 --> 00:50:44.010
They will not use normal laptops, cell phones or computers.

469
00:50:44.010 --> 00:50:47.159
But instead the venues, quantum computers.

470
00:50:48.869 --> 00:50:52.829
These quantum computers actually don't need to be very complicated.

471
00:50:54.449 --> 00:50:58.500
It turns out that most applications of Internet.

472
00:50:58.500 --> 00:51:02.340
Only require these notes quantum computers.

473
00:51:02.340 --> 00:51:05.820
To be very simple and have less than 10 cubes.

474
00:51:07.949 --> 00:51:13.199
Except for most applications, they only have 1 queued.

475
00:51:14.699 --> 00:51:18.269
The reason why we typically do not need many Cubans.

476
00:51:18.269 --> 00:51:23.190
Is because the quantum Internet draws its power from quantum entanglement.

477
00:51:24.329 --> 00:51:29.400
In already 1, Cupid at each end point is sufficient to have intent.

478
00:51:31.800 --> 00:51:36.750
In contrast on the quantum computer, we always need more Cubans.

479
00:51:36.750 --> 00:51:41.639
That can be simulated on a classical computer in order to do something new and interesting.

480
00:51:44.190 --> 00:51:50.309
The next element of a quantum Internet is that similar to the classical Internet.

481
00:51:50.309 --> 00:51:56.969
We have all kinds of elements that allows to maximize the use of existing infrastructure.

482
00:51:58.440 --> 00:52:04.380
On the classical Internet, not every computer on the Internet has a direct fiber connection.

483
00:52:04.380 --> 00:52:11.519
To every other computer on Internet, but instead fibrous run through central points.

484
00:52:12.719 --> 00:52:16.289
Where their switches that direct the bits in the right direction.

485
00:52:18.989 --> 00:52:23.940
If you want to build the call to Internet and similar to a classical Internet.

486
00:52:23.940 --> 00:52:28.710
For example, 1 switches that are capable of switching in cubes.

487
00:52:31.769 --> 00:52:36.269
Ideally, you would like to send a Cubics of a very long distances.

488
00:52:36.269 --> 00:52:40.170
For any point owners to any other point on.

489
00:52:42.210 --> 00:52:46.050
In order to achieve this, we're going to need something.

490
00:52:46.050 --> 00:52:49.320
That is capable of sending Cubans over notices.

491
00:52:49.320 --> 00:52:55.349
This requires a very special form of for Peter called a quantum repeated.

492
00:52:56.880 --> 00:53:00.960
And quantum repeated makes very differently than the classical repeat.

493
00:53:00.960 --> 00:53:05.820
The separate video, you will learn all about content.

494
00:53:07.590 --> 00:53:12.510
Realizing a quantum Internet interest, like, on the Internet.

495
00:53:12.510 --> 00:53:16.079
We will also need some control traffic.

496
00:53:16.079 --> 00:53:19.409
Basically, next to the quantum communication.

497
00:53:19.409 --> 00:53:22.769
I will also use classical communication.

498
00:53:22.769 --> 00:53:27.840
For example, to direct the Cubans to the right destination in the network.

499
00:53:30.750 --> 00:53:33.929
This is what a call from Ethernet it looks like.

500
00:53:35.550 --> 00:53:38.820
I've already mentioned that they call the Internet.

501
00:53:38.820 --> 00:53:42.030
Allows us to solve tasks that are impossible.

502
00:53:42.030 --> 00:53:47.400
To accomplish on the classical now, the question is.

503
00:53:47.400 --> 00:53:53.159
What makes a quantum Internet, or what makes a transmission of.

504
00:53:53.159 --> 00:53:56.429
So much more powerful than what we have today.

505
00:53:58.409 --> 00:54:03.420
Cubans has very special features, for example.

506
00:54:03.420 --> 00:54:08.039
They cannot be copied, making them ideal for security applications.

507
00:54:09.929 --> 00:54:13.769
2 can also be in a very special state.

508
00:54:13.769 --> 00:54:19.469
State and Tango state between.

509
00:54:19.469 --> 00:54:23.519
Is the essence of the power of the Internet.

510
00:54:25.230 --> 00:54:30.630
In order to understand entanglement why entanglement is so useful.

511
00:54:30.630 --> 00:54:36.300
It is sufficient to understand 2 very fundamental properties.

512
00:54:36.300 --> 00:54:42.989
Often Tangram, so, let me explain these 2 properties of entanglement.

513
00:54:42.989 --> 00:54:46.170
And why take is power to call to it.

514
00:54:48.030 --> 00:54:53.309
The 1st, feature of entertainment instead it allows maximum coordination.

515
00:54:53.309 --> 00:54:56.880
So, what is this.

516
00:54:58.590 --> 00:55:03.960
2 cubes can be entangled, even at very long distances.

517
00:55:05.429 --> 00:55:12.210
For example, I can have a Cuban in dealt with a cubic very far away for example, in China.

518
00:55:14.550 --> 00:55:17.849
Now, if I make a measurement of my Cupid here, adults.

519
00:55:17.849 --> 00:55:21.119
And a friend of mine would make the same measurement in China.

520
00:55:22.619 --> 00:55:26.099
Then it will turn out that we always gets the same outcome.

521
00:55:27.570 --> 00:55:31.619
You can think of a measurement as asking a question to acuity. It.

522
00:55:32.880 --> 00:55:37.139
For example, I might ask the Cuban, you're pointing left.

523
00:55:37.139 --> 00:55:45.510
Or you're pointing right? Maximum coordination rules that if I see the.

524
00:55:45.510 --> 00:55:48.630
Then immediately instantaneously.

525
00:55:48.630 --> 00:55:54.000
If my friend in China and makes the same measurement, the Cuban, but also be pointing to the left.

526
00:55:55.679 --> 00:56:00.960
If I see point to the right, and also in China, it will be pointing to the rights.

527
00:56:00.960 --> 00:56:05.159
Even, if this onset is not determined ahead of time.

528
00:56:07.349 --> 00:56:10.800
In fact, randomly, we will get left left.

529
00:56:10.800 --> 00:56:16.260
All right, right but the point is that the outcomes will always be the same.

530
00:56:17.309 --> 00:56:23.699
The amazing thing about entanglement is that this is true for any measurement.

531
00:56:23.699 --> 00:56:28.980
Or any question we might ask if I were to ask the Cuban.

532
00:56:28.980 --> 00:56:33.150
Kevin, are you ready and we want.

533
00:56:33.150 --> 00:56:37.320
All observe maximum coordination, regret.

534
00:56:37.320 --> 00:56:43.559
Open book, but never anything else. So the 1st, feature of entanglement.

535
00:56:43.559 --> 00:56:50.340
Is maximum coordination, and it is this feature that makes entanglement so suitable.

536
00:56:50.340 --> 00:56:54.059
For tasks that require synchronization or coordination.

537
00:56:54.059 --> 00:56:59.789
The 2nd feature from tanglement instead it's inherently private.

538
00:57:01.619 --> 00:57:10.260
Because, of course, you might have been wondering, given that entanglement is so powerful, allowing this instantaneous maximum coordination.

539
00:57:10.260 --> 00:57:14.280
Wouldn't that be great if manager which could be in time.

540
00:57:15.630 --> 00:57:21.179
Now, it turns out that only 2 Cubans can be maximally entangled with each other.

541
00:57:22.949 --> 00:57:26.099
Entanglement is inherently private.

542
00:57:26.099 --> 00:57:31.980
I have a, and the cubic that disentangle with someone in China.

543
00:57:31.980 --> 00:57:36.300
And you can think of this entanglement as a private connection.

544
00:57:36.300 --> 00:57:42.480
That nothing else can have part of it is not possible for any other cubic anywhere.

545
00:57:42.480 --> 00:57:47.610
To any between the delta and the Cuba in China.

546
00:57:50.130 --> 00:57:53.760
It is this feature that makes quantum communication.

547
00:57:53.760 --> 00:57:57.960
So fundamentally suitable for tasks that require privacy.

548
00:57:57.960 --> 00:58:05.219
Security, so, entanglement is power to a call to the Internet.

549
00:58:05.219 --> 00:58:10.500
Later video, you will see how to use this entanglement.

550
00:58:10.500 --> 00:58:14.639
To also send to cubes using quantum teleportation.

551
00:58:14.639 --> 00:58:20.219
Sure.

552
00:58:22.050 --> 00:58:29.489
Okay, I'll let you look up to quantum teleportation video on your own. If you'd like it. I'd like to do finish off with some higher level.

553
00:58:29.489 --> 00:58:36.119
Videos about applications of quantum computing now machine learning chemistry.

554
00:58:36.119 --> 00:58:44.699
Tenser flow stuff.

555
00:58:47.639 --> 00:58:50.789
Yard cycle bike and bicycle.

556
00:58:50.789 --> 00:58:54.960
Are all closely related, but how does the computer know.

557
00:58:56.099 --> 00:59:02.670
Humans are very good in recognizing your face, but for computers, such recognition is typically hard.

558
00:59:02.670 --> 00:59:07.349
The reason behind this is that the computer programing language of computers.

559
00:59:07.349 --> 00:59:14.159
Many of these problems.

560
00:59:14.159 --> 00:59:17.849
So, instead of the standard programming language.

561
00:59:17.849 --> 00:59:22.199
We can also take a different approach called machine learning.

562
00:59:23.429 --> 00:59:26.699
Machine learning algorithms are more flexible.

563
00:59:26.699 --> 00:59:31.440
When it comes to dealing with situations that are slightly different every time.

564
00:59:31.440 --> 00:59:35.670
Or when the data is noisy, they're also really good.

565
00:59:35.670 --> 00:59:38.940
And then patterns from large data sets.

566
00:59:38.940 --> 00:59:42.900
This makes them very useful for applications.

567
00:59:42.900 --> 00:59:46.409
Such a self driving cars or face recognition.

568
00:59:46.409 --> 00:59:51.389
So, machine learning is discipline.

569
00:59:51.389 --> 00:59:54.510
Of learning patterns from large amounts of data.

570
00:59:54.510 --> 00:59:59.309
Instead of coding, complicated algorithms into computer programs.

571
00:59:59.309 --> 01:00:06.300
The idea of machine learning is to show simple computer algorithm lots and lots of examples of correct.

572
01:00:06.300 --> 01:00:10.050
And wrong behavior, for example.

573
01:00:10.050 --> 01:00:14.159
If they want to train a computer to recognize a cats.

574
01:00:14.159 --> 01:00:20.880
We could show the computer many examples of cats. This is a cat. This is a cat. This is a, this is not a cat.

575
01:00:20.880 --> 01:00:25.769
In this way, the computer starts to learn from the data.

576
01:00:25.769 --> 01:00:33.539
What the expected behavior is the requirements for these kinds of algorithms are enormous computing power.

577
01:00:33.539 --> 01:00:36.989
And the ability to work with many data points.

578
01:00:36.989 --> 01:00:41.550
Thousands millions or billions of data points.

579
01:00:41.550 --> 01:00:46.230
This requirement of skill means that machine learning techniques.

580
01:00:46.230 --> 01:00:51.750
Even though they have been known for decades, have not been used in real world situations.

581
01:00:51.750 --> 01:00:56.820
Until very recently, how can quantum solve this.

582
01:00:58.559 --> 01:01:02.039
That's where Shane learning is a vision or an approach.

583
01:01:02.039 --> 01:01:06.000
Rather than a single algorithm, it is not immediately clear.

584
01:01:06.000 --> 01:01:09.630
What us quantum computing cloud could have for a machine learning.

585
01:01:09.630 --> 01:01:12.960
Harbor of 1 closer inspection.

586
01:01:12.960 --> 01:01:16.170
It becomes apparent that many machine learning techniques.

587
01:01:16.170 --> 01:01:21.269
Rely on more basic tasks, such as linear system solving for their functioning.

588
01:01:21.269 --> 01:01:24.570
Any questions can be efficiently solved.

589
01:01:24.570 --> 01:01:30.239
Using the algorithm named office inventors.

590
01:01:30.239 --> 01:01:33.750
However, as well, we'll discuss in a moment.

591
01:01:33.750 --> 01:01:36.900
There are significant challenges in the realizations.

592
01:01:36.900 --> 01:01:42.150
Of this approach cornerstone machine, learning technique.

593
01:01:42.150 --> 01:01:48.389
Principal component analysis and also we speed up significantly on a quantum computer.

594
01:01:49.530 --> 01:01:55.650
Principal component analysis can be understood as and identifying the most important parts.

595
01:01:55.650 --> 01:02:00.360
Or large Matrix, and using this to describe the metrics.

596
01:02:01.380 --> 01:02:04.679
If return for example, of recognizing cats.

597
01:02:04.679 --> 01:02:09.510
We could show a computer many examples of cats and us these images.

598
01:02:09.510 --> 01:02:16.860
To create an average cat to reconstruct the individual pictures we start by using the most important variable.

599
01:02:16.860 --> 01:02:22.320
Which is the 1 that gives the largest variants this could be, for example, the size of the cat.

600
01:02:22.320 --> 01:02:27.119
But it could also be a combination of factors, for example, a combination of weight and size.

601
01:02:27.119 --> 01:02:33.989
It would also be something else, but the computer will find out, like, comparing all images.

602
01:02:33.989 --> 01:02:39.480
Variable with the largest variance is the 1st principal component.

603
01:02:39.480 --> 01:02:44.099
After that's we can just take the 2nd, most important variable. The 1st.

604
01:02:44.099 --> 01:02:47.789
And so on in doing, so we end up.

605
01:02:47.789 --> 01:02:51.239
With a set of questions that we can reconstruct a cat.

606
01:02:51.239 --> 01:02:57.150
To decide curacy well, this is a powerful way for recognition.

607
01:02:57.150 --> 01:03:04.949
It is still hard for classical computers. The quantum algorithm for principal component analysis is called the algorithm.

608
01:03:04.949 --> 01:03:08.070
Named after enable cost.

609
01:03:08.070 --> 01:03:11.969
This algorithm can provide an exponential speed up.

610
01:03:11.969 --> 01:03:17.730
I would therefore be highly valuable. What are the caveats?

611
01:03:19.769 --> 01:03:23.460
Because the advantages of quantum machine learning.

612
01:03:23.460 --> 01:03:28.650
Mostly rely on the agent's algorithm that can be used to solve linear equations.

613
01:03:28.650 --> 01:03:33.659
Caveats related to solving linear system supply here as well.

614
01:03:33.659 --> 01:03:37.739
Perhaps the most significant challenge is the data input problem.

615
01:03:37.739 --> 01:03:42.030
This is especially challenging for machine learning algorithms.

616
01:03:42.030 --> 01:03:47.159
Since by design, we're dealing with very noisy, a difficult data.

617
01:03:47.159 --> 01:03:53.849
This means that finding an efficient way of storing this data in a quantum memory.

618
01:03:53.849 --> 01:04:00.329
For later reference is highly non trivial. A similar encoding problem holds 40 algorithm.

619
01:04:00.329 --> 01:04:03.329
Used for quantum principal component analysis.

620
01:04:03.329 --> 01:04:06.539
If we can overcome these challenges.

621
01:04:06.539 --> 01:04:09.690
There won't be many applications for quantum machine learning.

622
01:04:09.690 --> 01:04:13.170
Application areas can range from security.

623
01:04:13.170 --> 01:04:16.679
For fast recognition of persons in large crowds.

624
01:04:16.679 --> 01:04:22.559
Personal advertisements, medical applications, such as recognizing diseases.

625
01:04:22.559 --> 01:04:25.949
Or even for accurate prediction for.

626
01:04:25.949 --> 01:04:31.079
Sure.

627
01:04:32.940 --> 01:04:38.610
Okay now see something quantum chemistry for a few minutes.

628
01:04:45.389 --> 01:04:52.139
Oops.

629
01:04:54.630 --> 01:05:01.139
Computer simulation is widely used in technology today and really across a wide range of applications as well from.

630
01:05:01.139 --> 01:05:06.210
How air flows over the weight of an airplane to how traffic moves through a city.

631
01:05:06.210 --> 01:05:14.579
Or how an electrical circuit functions problems however, are extremely complex simulates. It cannot be solved today.

632
01:05:14.579 --> 01:05:18.719
Both accurately and quickly, even on the most powerful supercomputers.

633
01:05:18.719 --> 01:05:22.260
A good example is stimulation of quantum systems.

634
01:05:22.260 --> 01:05:28.019
So just molecules or materials that's actually very relevant such simulations.

635
01:05:28.019 --> 01:05:31.050
It can help us design more quickly new medication.

636
01:05:31.050 --> 01:05:35.820
It can help us design materials that don't have electrical resistance or.

637
01:05:35.820 --> 01:05:39.960
Come up with new chemical compounds to be used in industry or agriculture.

638
01:05:42.000 --> 01:05:49.860
In a computer simulation, you can quickly generate and check many different versions of a molecule and compute their properties.

639
01:05:49.860 --> 01:05:54.090
Avoiding their by the expressive process of having to synthesize.

640
01:05:54.090 --> 01:05:57.900
Many molecules 1 after another, the test in real life.

641
01:05:57.900 --> 01:06:02.159
What their properties are, let's have a look at the theory.

642
01:06:04.289 --> 01:06:11.489
The basis for any assimilation is a model model is for us a simple description that captures the essential features.

643
01:06:11.489 --> 01:06:16.500
Of the object, we're trying to simulate whether it's the airplane or the electrical circuits.

644
01:06:16.500 --> 01:06:20.849
In the case of quantum systems, the model is a Hamilton.

645
01:06:20.849 --> 01:06:24.690
Toenail that is for us large matrix.

646
01:06:24.690 --> 01:06:31.139
With lots of numbers, and it contains the information about the basic elements of the molecule or the material.

647
01:06:31.139 --> 01:06:35.550
A system, we're trying to signal this Tanya.

648
01:06:35.550 --> 01:06:39.900
Intercom curious will allow this computer to find all the relevant properties of the system.

649
01:06:39.900 --> 01:06:47.190
Without ever having to build it in the lab. The difficulty is that if you do this on a conventional computer.

650
01:06:47.190 --> 01:06:52.619
It will take that computer exponential time time that goes up as 2 to the power and.

651
01:06:52.619 --> 01:06:56.940
Where, and is the number of particles that your molecule or material.

652
01:06:56.940 --> 01:07:01.860
Is composed of how can quantum solve this.

653
01:07:03.150 --> 01:07:09.929
Quantum computers, on the other hand, have themselves an exponential complexity built in that matches the exponential complexity.

654
01:07:09.929 --> 01:07:13.079
Of the quantum systems, they were trying to simulate to computers.

655
01:07:13.079 --> 01:07:18.570
And as a result, a console computer requires only polynomial time, not exponential time.

656
01:07:18.570 --> 01:07:23.730
The simulate and not audience of molecules or materials.

657
01:07:23.730 --> 01:07:27.480
Specifically the time goes up as end to the power 3.

658
01:07:27.480 --> 01:07:33.869
Instead of 2 to the power and an exponential improvement and really the difference between being able to solve the problem.

659
01:07:33.869 --> 01:07:36.989
Or what are the caveats.

660
01:07:38.639 --> 01:07:42.000
A few final comments about the power and possibilities.

661
01:07:42.000 --> 01:07:46.769
Handle Tonya assimilation. 1st only certain properties can be actually.

662
01:07:46.769 --> 01:07:52.679
Simulated efficiently, for instance, the lowest energy of a molecule we can compute efficiently.

663
01:07:52.679 --> 01:07:55.739
But to describe the state.

664
01:07:55.739 --> 01:08:01.679
The lowest energy states in food is exponentially complex, even for the quantum computer.

665
01:08:02.820 --> 01:08:08.010
2nd, the result of simulation, any assimilation classical.

666
01:08:08.010 --> 01:08:11.639
Is always just as good as the information that you put it.

667
01:08:11.639 --> 01:08:14.699
So the more accurate, the information that you have.

668
01:08:14.699 --> 01:08:20.670
To enter your Hamilton into more accurate, the results about the properties of the material.

669
01:08:20.670 --> 01:08:25.770
Molecule that you're trying to assimilate so keeping in mind these caveats.

670
01:08:25.770 --> 01:08:30.479
Still, we think that the 1st, important application of quantum computing will be.

671
01:08:30.479 --> 01:08:33.600
Hello Tanya simulation.

672
01:08:33.600 --> 01:08:37.260
Sure.

673
01:08:39.600 --> 01:08:46.529
And then 1, final flow video, this 1 is not from Q tech Academy.

674
01:08:46.529 --> 01:08:50.550
6 quantum summary symposium that happens.

675
01:08:50.550 --> 01:08:53.789
Every so often.

676
01:08:56.250 --> 01:09:04.859
Hi, everyone I'm Murphy new from Google.

677
01:09:04.859 --> 01:09:11.850
It's an honor to speak on behalf of our content team and introduce you to an open source library.

678
01:09:11.850 --> 01:09:23.369
For quantum machine learning before I start sharing with you, the awesome is up to you. Let's go down the memory lane of how much I wish I had.

679
01:09:23.369 --> 01:09:29.640
3 years ago as a board rescue and FedEx from MIT, I came to Google for the summer internship.

680
01:09:29.640 --> 01:09:35.880
Started you gave me a very hard quantum control problem of finding the analog control wave form.

681
01:09:35.880 --> 01:09:40.319
For realizing any to pubic gate well, facing realistic imperfections.

682
01:09:40.319 --> 01:09:46.680
Such as the kitchen control certainties so, congressional control optimization methods of built.

683
01:09:46.680 --> 01:09:52.260
Pretty bad so I saw there is no reason not to try machine learning, but Google.

684
01:09:52.260 --> 01:10:05.819
But it wasn't easy at 1st, after struggling with various incompatible pillaging issues and guest stuck on your full time. I also have to write my own quantum simulator on hyper factor optimization.

685
01:10:05.819 --> 01:10:08.970
So, that experience was exhausting and.

686
01:10:08.970 --> 01:10:15.180
When I returns of Google for another summer, I decided to what machine learning software.

687
01:10:15.180 --> 01:10:18.989
I just instead write a custom network and Optimizer.

688
01:10:18.989 --> 01:10:23.640
Just with comment Tensor algebra I thought for my next project.

689
01:10:23.640 --> 01:10:27.659
The same summer so came up to me and asked.

690
01:10:27.659 --> 01:10:35.039
What, if I give you a platform to generate and interface data with classic machine learning infrastructures.

691
01:10:35.039 --> 01:10:39.869
Seamless seamlessly. I asked her to out. Yes when can I use it?

692
01:10:39.869 --> 01:10:47.250
So much for it, so next year I going to go for time and I'm still waiting.

693
01:10:47.250 --> 01:10:51.149
But Michael Brighton is giving us hope that we won't get there. So.

694
01:10:51.149 --> 01:10:54.420
And fast forward to March 2020.

695
01:10:54.420 --> 01:10:58.409
Tons of content is lunch. So now it takes a physicist.

696
01:10:58.409 --> 01:11:13.284
I just want a software TensorFlow quantum to do machine learning research is for quantum systems with this unified platform. We have process both classical data and using both classical and quantum.

697
01:11:13.560 --> 01:11:23.789
Um, machine learning, architectures so what do we mean by quantum data? 1 kinds of data present in the fast developing quantum communication networks.

698
01:11:23.789 --> 01:11:34.529
For the longest on key distribution transmits information carried by a single photo across the continent between Australia and China through satellite railing.

699
01:11:34.529 --> 01:11:44.310
Another kinds of data is obtained through various methods ranging from our center to read for.

700
01:11:44.310 --> 01:11:49.170
And many more for quantum imaging, and realizing optical memories.

701
01:11:49.170 --> 01:11:53.189
Lastly, we also have measurement and calibration results.

702
01:11:53.189 --> 01:11:58.739
Attend in computer was called the superconducting 1 at Google.

703
01:11:58.739 --> 01:12:05.399
It is super cool. Also we have analog forms sent in the microwave.

704
01:12:05.399 --> 01:12:10.170
That Larisa, others carefully ranked to control our computers.

705
01:12:10.170 --> 01:12:13.859
These diverse kinds of data coming fed into.

706
01:12:13.859 --> 01:12:20.640
Equally diverse machine learning architectures to extract useful information and predictions.

707
01:12:20.640 --> 01:12:28.890
I listed some examples of classical quantum machine learning architectures here, which I won't have time to go into detail.

708
01:12:28.890 --> 01:12:34.140
But the takeaway is this diverse data and machine learning architectures.

709
01:12:34.140 --> 01:12:38.310
And not be united on their web framework. So full quantum.

710
01:12:38.310 --> 01:12:42.960
It is a software framework for hybrid quantum, classical machine learning.

711
01:12:42.960 --> 01:12:48.600
Under tons of flow insert we aim at enabling fast prototyping.

712
01:12:48.600 --> 01:12:51.750
Training and for us and testing of models.

713
01:12:51.750 --> 01:13:01.350
For final data, so you actually facilitate the discovery of new hours on for the wise and error corrected on computers.

714
01:13:01.350 --> 01:13:06.810
So, what is the secret behind that allows such a unification?

715
01:13:06.810 --> 01:13:11.640
Fundamentally, speaking quantum circus are no more than.

716
01:13:11.640 --> 01:13:18.300
That's like classic on your an hour. So if we can seamlessly call word quantum circuits to.

717
01:13:18.300 --> 01:13:22.050
That are compatible with TensorFlow data structure.

718
01:13:22.050 --> 01:13:26.430
And corporate content, measurement outcomes to tenser contractions.

719
01:13:26.430 --> 01:13:30.029
Of Megan, the operators also specified by the.

720
01:13:30.029 --> 01:13:35.069
Or so Co Ops, then quantum data and quantum machine learning architecture.

721
01:13:35.069 --> 01:13:38.460
Become part of an overall computational graph.

722
01:13:38.460 --> 01:13:42.000
Easily integrable with classical machine learning agents.

723
01:13:42.000 --> 01:13:49.619
So, with this big picture in mind here is the more detailed look at the software stack.

724
01:13:49.619 --> 01:13:53.220
At the top is a humongous amount of data we need it.

725
01:13:53.220 --> 01:13:59.729
For machine, I mean, it can be classical or quantum data prepared by a quantum circuit.

726
01:13:59.729 --> 01:14:03.420
The data that into the machine, learning the architecture.

727
01:14:03.420 --> 01:14:06.720
Including boss, traditional tents are full layers.

728
01:14:06.720 --> 01:14:10.409
And the new content layer provided by.

729
01:14:10.409 --> 01:14:17.520
There were models the output specified by the 10th operators are passed down to the hardware back.

730
01:14:17.520 --> 01:14:22.050
Be in or.

731
01:14:22.050 --> 01:14:25.859
Thorough test simulator.

732
01:14:25.859 --> 01:14:29.369
Depending on whether acquire simulation, the quantum.

733
01:14:29.369 --> 01:14:33.239
Measurement or execution of the quantum circuit in the lab.

734
01:14:33.239 --> 01:14:38.880
So, now, let's see if I come back to the past what I would have.

735
01:14:38.880 --> 01:14:47.399
Well, my reinforcement learning project, so no more struggling with software compatibility issue are writing your own quantum.

736
01:14:47.399 --> 01:14:52.590
Over the choice of a, you can optimize quantum circuit.

737
01:14:52.590 --> 01:14:57.930
In real time now you might be eager to see how she works in action.

738
01:14:57.930 --> 01:15:01.529
Let me give you a test of it was real quantum computers.

739
01:15:01.529 --> 01:15:05.039
For all for learning 1, last noise are device.

740
01:15:05.039 --> 01:15:09.689
Noise is notoriously prevalent.

741
01:15:09.689 --> 01:15:17.369
Almost all the state Cubics, it manifests as a slow drifting term of error ends of cubic frequency.

742
01:15:17.369 --> 01:15:23.880
By this haven't telling him, whereas represent the amplitude of the noise.

743
01:15:23.880 --> 01:15:31.199
E, represent a specific time so the problem I want to learn how to define as follow.

744
01:15:31.199 --> 01:15:35.369
Given steak prepared in the X spaces.

745
01:15:35.369 --> 01:15:39.539
I vaulted under Z. I'm telling you. It's not.

746
01:15:39.539 --> 01:15:44.609
For a certain amount of time, and perform the measurement of the spaces.

747
01:15:44.609 --> 01:15:49.470
X expectations values so this is the early cover. Ramsey experiment.

748
01:15:49.470 --> 01:15:53.909
Given this experiment result merits for different.

749
01:15:53.909 --> 01:15:58.350
Amount of time up to T. I like to predict.

750
01:15:58.350 --> 01:16:04.949
The measurement for the future times, so a success so a prediction of this.

751
01:16:04.949 --> 01:16:09.029
Sequential data will also means a correct.

752
01:16:09.029 --> 01:16:16.590
Of the white noise parameters, it seems to be a simple problem, but it's difficult to the weak aptitude.

753
01:16:16.590 --> 01:16:19.859
And the long term dynamics of the warm up noise.

754
01:16:19.859 --> 01:16:23.279
So, now it's time to use the magic of machine learning.

755
01:16:23.279 --> 01:16:27.659
Recurrent neural network is particularly designed to represent.

756
01:16:27.659 --> 01:16:35.699
The time sequential data efficient 1 of the most widely used repairing your network is called long term, short term memory.

757
01:16:35.699 --> 01:16:42.930
Which explicitly CrowdRise the balance act of prioritizing the long term. Where's the short term memory?

758
01:16:42.930 --> 01:16:47.189
It's represented in the lower, right? It looks.

759
01:16:47.189 --> 01:16:53.100
Kind of complicated, but you don't have to worry about it. It will take care.

760
01:16:53.100 --> 01:16:56.159
All you need to do is touch a car and they.

761
01:16:56.159 --> 01:16:59.819
She tends to flow function, colored in this yellow box.

762
01:16:59.819 --> 01:17:08.069
As well specifying size uncertain the hyper parameter I input size of the recurring your network.

763
01:17:08.069 --> 01:17:15.479
And then to train this to learn a time sequential structure of your experimental data.

764
01:17:15.479 --> 01:17:20.489
What's left is to defy our cost function and to choose the Optimizer.

765
01:17:22.920 --> 01:17:30.659
So, you might be more ambitious instead of using a quantum simulator. You might want to swap in a real quantum computer.

766
01:17:30.659 --> 01:17:37.949
And this is also very simple, just replace an expectation layer with simple. The expectation.

767
01:17:37.949 --> 01:17:43.079
And don't forget to choose your best back and to be the amazing content.

768
01:17:43.079 --> 01:17:48.989
Which they can, we'll talk more about and probably gave yourself grateful for that. I asked.

769
01:17:48.989 --> 01:17:52.170
And specify the number of reputations.

770
01:17:52.170 --> 01:18:00.689
Of the measurement of each circuit, bam, this is the fresh data taken from yesterday our quantum engine.

771
01:18:00.689 --> 01:18:05.100
Whereas the upper right is the training performance.

772
01:18:05.100 --> 01:18:10.949
With full carpark about as the input data to the, and the aren't part.

773
01:18:10.949 --> 01:18:16.020
As a prediction, and the large part of the figure is the testing results.

774
01:18:17.789 --> 01:18:21.659
So, that summarized my lighting and talk for 10 simple.

775
01:18:21.659 --> 01:18:26.159
Although, I mean, is worth a 1000 words I hope that you will be.

776
01:18:26.159 --> 01:18:29.250
To try converting everything into.

777
01:18:29.250 --> 01:18:33.659
There TensorFlow quantum and go out there to solve big problems.

778
01:18:33.659 --> 01:18:40.949
And let us know how you like it. Finally, I'd like to Susie asked, they got car walking, tensive flow, quantum team.

779
01:18:40.949 --> 01:18:44.460
Was to start the project back in 2018.

780
01:18:44.460 --> 01:18:51.300
We probably would have lot of sense of economies on Michael, leading the engineering effort together. It was amazing how.

781
01:18:51.300 --> 01:18:54.539
From the student researchers to remodel internship.

782
01:18:54.539 --> 01:19:01.710
Over the past 2 years and LS amazing effort on deploying as intensive product.

783
01:19:01.710 --> 01:19:07.590
To share with a broader public, and it's still ever improving open source library.

784
01:19:07.590 --> 01:19:10.890
And that depends contributors like, you.

785
01:19:10.890 --> 01:19:20.100
So, thank you for listening for the next step. Please go ahead to check out for. Let's talk about as at TensorFlow summit.

786
01:19:20.100 --> 01:19:26.670
And our website, which have many amazing videos and demo tutorials as well as.

787
01:19:26.670 --> 01:19:33.000
On tensive flow white paper and if you have any question feel free to ask them and the dory.

788
01:19:33.000 --> 01:19:40.109
And I will give it back to Marissa.

789
01:19:41.130 --> 01:19:44.489
Well, okay, so.

790
01:19:44.935 --> 01:19:59.215
That Tom gave you some, some high level applications of quantum computing. Oh, 1 other thing some personal stuff. I passed a Pre registration period. If anyone is still looking.

791
01:19:59.760 --> 01:20:08.909
For a course in the spring, I'm teaching a parallel computing class, so we'll be using.

792
01:20:08.909 --> 01:20:12.869
And multi core machines involving.

793
01:20:14.305 --> 01:20:27.895
And then, and the focus will be on parallel computing tools that you can buy, you can afford yourself use in your lab, using your project. So I'm not competing with computer science.

794
01:20:27.895 --> 01:20:32.154
Course I'm more down not involving Super computing. So.

795
01:20:32.430 --> 01:20:36.180
You're welcome to look at that. If you wish.

796
01:20:36.895 --> 01:20:51.114
And then 1 other thing, I don't have a photo for it, but I'm trying to do applied electrical engineering in my own house. And what I had installed today were 2 Tesla power walls. So they were just installed a couple of hours ago.

797
01:20:51.390 --> 01:21:00.329
And what we'll have installed on Friday is a couple of Teslas, solar panels. I'll be playing with some Tesla toys, and can report.

798
01:21:00.329 --> 01:21:05.399
Report how they work. Okay so.

799
01:21:05.399 --> 01:21:09.270
See, you on Thursday and.

800
01:21:09.270 --> 01:21:15.359
We have on talking, we have our guest speakers on Thursday here.

801
01:21:15.359 --> 01:21:20.699
I'll stay around a few minutes if there's questions other than that have a good week.

802
01:21:20.699 --> 01:21:25.229
Silence.

803
01:21:25.229 --> 01:21:32.909
Silence.

804
01:21:33.989 --> 01:21:38.789
And thank you, Amanda.

805
01:21:38.789 --> 01:21:47.699
Silence.

806
01:21:57.449 --> 01:22:01.680
Silence.

807
01:22:06.090 --> 01:22:13.560
Silence.