WEBVTT

1
00:01:07.644 --> 00:01:16.224
Okay, good afternoon. Everyone, this is quantum computer programming, class, 20 and November nights.

2
00:01:16.500 --> 00:01:19.890
2020, so what we have.

3
00:01:19.890 --> 00:01:24.719
Today, let me just share stuff.

4
00:01:30.239 --> 00:01:34.500
And, uh.

5
00:01:34.500 --> 00:01:38.400
She is working and.

6
00:01:38.400 --> 00:01:43.500
I have the chat window open as usual, so.

7
00:01:43.500 --> 00:01:49.109
In type messages a synchronously, I encourage you to ask questions make comments.

8
00:01:49.109 --> 00:01:52.140
And as a fall back, I got my.

9
00:01:52.140 --> 00:02:02.519
Cell phone here, so he can message me if you have any other communication medium fails. So what do we have today?

10
00:02:02.519 --> 00:02:09.900
1st, I've got another homework up and this is a really open ended homework here.

11
00:02:09.900 --> 00:02:15.120
Idea is for you to teach yourself about Amazon bracket.

12
00:02:15.120 --> 00:02:20.340
amazon's computing platform, so browse around things here.

13
00:02:20.340 --> 00:02:28.349
And set things up and run something, find an interesting example program. It run something on.

14
00:02:28.349 --> 00:02:33.180
Their simulator later will work with actual machines now.

15
00:02:33.444 --> 00:02:42.775
I'm not certain that I actually have the right to require you to do this because this will probably cost you a few dollars not a lot of dollars.

16
00:02:43.194 --> 00:02:53.034
However, if you feel that I'm being unfair to ask you to do this, then write to me and I'll get you a replacement homework, which.

17
00:02:53.520 --> 00:03:00.990
Well, you know, just as easy to get 100% grade, but I think that.

18
00:03:00.990 --> 00:03:06.990
The balances, I'm asking you to pay a few dollars, but what you're getting is you're playing around.

19
00:03:06.990 --> 00:03:16.110
With quantum computing on Amazon, so if you'd like to actually express opinions on this now or later.

20
00:03:16.110 --> 00:03:19.110
In the class and, you know.

21
00:03:19.110 --> 00:03:24.569
Then I'm, I'm interested in hearing what you have to say.

22
00:03:24.569 --> 00:03:38.219
By the way we're starting to think a little less kindly about since they're putting up ads now and they're thinking a ways to.

23
00:03:38.219 --> 00:03:50.520
Charge us money so we're an faculty were we were investigating possible alternatives. This would be another thing to think of. I may post it on the.

24
00:03:51.870 --> 00:03:58.169
You know, on the next item on the blog for Thursday.

25
00:03:58.169 --> 00:04:09.330
Also other things coming up I have a couple of more I hope you enjoyed our guest lecture on Thursday and.

26
00:04:09.330 --> 00:04:17.879
What is coming up are we still have some more guest lectures coming up so various topics, more physics related because I may.

27
00:04:17.879 --> 00:04:23.160
You know, software type person, so this will balance the course out for you.

28
00:04:24.329 --> 00:04:31.709
The new stuff today is mostly to I have 2 things for you.

29
00:04:31.709 --> 00:04:42.988
1, the 1st thing, Aces enrichment material, which means, I'm not you don't get a grade for it. You spend as much or as little time or a 0T time. If you want.

30
00:04:42.988 --> 00:04:53.218
But I think it's relevant to the course. And I actually see, my most important role in this course, is to be a curator.

31
00:04:53.218 --> 00:05:01.528
For you, I E, eyebrows around and I find stuff which I think you will like, and I, I think will benefit to, you.

32
00:05:01.528 --> 00:05:07.348
And point you to it, and then as you're interested, you.

33
00:05:07.348 --> 00:05:20.369
You browse around it so, the 1 thing here there is a conference called the Super computing conference or formally the International Conference for high performance computing.

34
00:05:20.369 --> 00:05:23.459
Networking storage and analysis.

35
00:05:23.603 --> 00:05:36.233
And the speed here, it's virtual this year, like anything else. I've never actually gone to it. I've just browsed their website. It's occurring virtually over 2 weeks.

36
00:05:36.233 --> 00:05:39.564
Now it has various things happening.

37
00:05:39.838 --> 00:05:48.569
Now, the thing with the Super computing conference is it costs money to register, even in formally.

38
00:05:48.569 --> 00:05:53.968
So, I'm not suggesting you register for it.

39
00:05:53.968 --> 00:06:07.853
Over the 2 things 1st, given how much you're paying our Pi paying a few 100 dollars here and there to conferences would be rounding error or something but no, seriously.

40
00:06:08.903 --> 00:06:15.834
Okay. But you see, it has Harrison, your tutorials workshops on interesting topics and.

41
00:06:16.108 --> 00:06:23.908
1 or 2 of which are quantum computing so tie them to the course. Now, 1 thing, which is free, our exhibits.

42
00:06:23.908 --> 00:06:34.738
I believe so so you could browse around and enable JavaScript. I expect.

43
00:06:34.738 --> 00:06:40.949
Okay, you can browse around and see various things here. So you might actually enjoy it.

44
00:06:40.949 --> 00:06:46.168
Going around here and seeing very Super computing sorts of things.

45
00:06:48.449 --> 00:06:55.379
And the exhibit stuff can be seen, you can register for free.

46
00:06:55.379 --> 00:06:59.428
Because the exhibitors are paying to show stuff to you are basically.

47
00:06:59.428 --> 00:07:05.009
Now, something at the Super computing 20 that is specifically.

48
00:07:05.009 --> 00:07:09.209
Quantum computing is here come on click.

49
00:07:10.889 --> 00:07:14.819
Going on here.

50
00:07:20.519 --> 00:07:23.999
Gang annoying.

51
00:07:27.478 --> 00:07:32.608
Transferring data.

52
00:07:37.228 --> 00:07:40.858
Still transferring data.

53
00:07:40.858 --> 00:07:44.399
Okay, let me go back and tell you.

54
00:07:44.399 --> 00:07:48.389
In a background thing here.

55
00:07:48.389 --> 00:07:56.218
Hey, 2nd, time at work. Okay so we have here the good have a quantum future section.

56
00:07:56.218 --> 00:07:59.608
Okay, so.

57
00:07:59.608 --> 00:08:05.218
So this would cost money, but let's suppose you didn't want to pay money.

58
00:08:05.218 --> 00:08:15.389
Like me. Okay, so what I do and I actually did for today is I saw this interesting talk here building google's quantum computer.

59
00:08:15.389 --> 00:08:19.858
And did invited talk on.

60
00:08:19.858 --> 00:08:30.209
And so on, so, what I did is, I Googled for Marissa, Christina, and this, and I found a free version of her talk.

61
00:08:30.209 --> 00:08:39.208
And quantum future as you could possibly do that. So the people that are giving these talks frequently, have free versions.

62
00:08:39.208 --> 00:08:45.058
Of their talks, so what you can do is you can use the Super computing 20 page.

63
00:08:45.058 --> 00:08:48.568
You get pointers to interesting people.

64
00:08:48.568 --> 00:08:52.739
So, you go to a conference like this, you find out what.

65
00:08:52.739 --> 00:09:00.599
People think is important in Super computing and now you go to a job interview. You can then just.

66
00:09:00.774 --> 00:09:12.474
Idly drop a few comment, you know, like, after a week, you'll be able to say oh, yes, when I programmed amazon's quantum computer and class.

67
00:09:12.504 --> 00:09:16.043
I learned that to compare you get the idea.

68
00:09:16.349 --> 00:09:19.438
And the idea is to make yourself marketable.

69
00:09:19.438 --> 00:09:24.899
Okay, so this is a Super computing 20 here.

70
00:09:26.999 --> 00:09:41.788
Just optional, but and apart from the, I mentioned it, they got the quantum computing section, but they have a whole pile of other cool stuff here, relating to Super computing. So I have fun technical program.

71
00:09:41.788 --> 00:09:48.178
Okay, when I go to a conference, I particularly like things like keynotes and invited talks.

72
00:09:48.178 --> 00:09:53.999
Okay, access scale, climate change to the Max Institute, for example.

73
00:09:53.999 --> 00:10:05.514
Could be interesting keynote talk and the thing is, keynote talks they're not based on papers, like normal academic talks so you can't read the Speaker's paper perhaps.

74
00:10:05.903 --> 00:10:10.464
Okay so that I've done the keynote talk go for the invited talks.

75
00:10:13.198 --> 00:10:16.979
Chair of a I, and now you can see.

76
00:10:18.208 --> 00:10:21.389
And you could learn a lot of interesting stuff here.

77
00:10:21.389 --> 00:10:27.448
In future computing in the so called post.

78
00:10:27.448 --> 00:10:30.719
More slaw you're and I probably going to mention quantum there.

79
00:10:30.719 --> 00:10:37.619
And so on you to see interesting important people from leaders in the field.

80
00:10:37.619 --> 00:10:44.068
So, you can have fun with that. So you go the keynote talk invited talks.

81
00:10:45.833 --> 00:10:58.524
Primary session as a planner recession that means it's not split up into separate sections. High performance. Computing could be something. There.

82
00:10:58.974 --> 00:11:01.104
What else would you want to look at?

83
00:11:02.938 --> 00:11:06.749
Panels can be interesting. People just be about stuff.

84
00:11:06.749 --> 00:11:11.519
Exotic.

85
00:11:11.519 --> 00:11:17.969
Yeah, so if you had done these sorts of things you're going to have to pay to register, but still.

86
00:11:18.594 --> 00:11:31.433
They can be and it sounds you. Oh, Jack on Kara. Okay. He's a leader. He jacked on Garrick. He's the guy that does the top 500 website, which has the 500, fastest known computers, which he updates twice a year.

87
00:11:31.433 --> 00:11:44.303
He has come to our API several times to give talks. For example, you browse here. You can see names of leading people and tutorials. Tutorials might be quite interesting. Where do we have.

88
00:11:45.058 --> 00:11:50.158
These will charge.

89
00:11:51.448 --> 00:11:59.038
Yeah, see, you might have fun here. Okay. They're not all that incredibly expensive, but they would cost you a few dollars.

90
00:12:00.024 --> 00:12:14.214
Okay, so that was okay back to the course. The main technical thing that I want to do today is talk about Google and quantum computing and what I'm going to do it by presenting.

91
00:12:14.908 --> 00:12:18.269
Lots of different videos about Google.

92
00:12:18.269 --> 00:12:24.058
And for the long ones, I may just do the start of it.

93
00:12:24.058 --> 00:12:37.974
Depending, on interesting, I, to the keynote address looks quite entering so I may watch the whole thing if you fail, that you'd rather move on to something else and you can tell me, but it's an interesting talk. I think any case.

94
00:12:37.974 --> 00:12:43.854
So, Google, the leaders are all involved in quantum computing machine company.

95
00:12:44.158 --> 00:12:53.428
Amazon Google, and there are several startup companies that I pointed you to websites and we'll see more of them in the future.

96
00:12:53.428 --> 00:13:00.269
I on cue and so on David trap guy on quantum computer, because the technology, I'll try to tell you something about.

97
00:13:00.269 --> 00:13:05.908
Okay, so we've got Google here. Now. Google has a, um.

98
00:13:07.889 --> 00:13:15.749
This is like their leading website. This is actually the landing page you might say. So you can have fun.

99
00:13:15.749 --> 00:13:19.469
Browsing around gotten google's website here.

100
00:13:21.509 --> 00:13:27.629
Lots of quantum cats at different levels. This is where I found a few of the things I'd like to show you today.

101
00:13:27.629 --> 00:13:33.359
Opening keynote, Google, quantum summer symposium.

102
00:13:33.359 --> 00:13:39.629
I think it's worth watching actually if you disagree with me, then you can tell me.

103
00:13:39.629 --> 00:13:43.198
But we'll see what's happening with quantum computing at Google.

104
00:13:51.149 --> 00:14:00.839
So, yeah, 1, to start us off with some updates to tell you, what's the.

105
00:14:00.839 --> 00:14:07.139
The lab has been up to since last year's symposium.

106
00:14:08.158 --> 00:14:14.129
And, um, I have the privilege to, um, talk about the achievements, um, and resides of our team.

107
00:14:14.129 --> 00:14:22.078
But, of course, I want to acknowledge the fact that we have a hardworking and a very talented team.

108
00:14:22.078 --> 00:14:27.808
So, it's hard to give credit a, every place with you. So here you see, um.

109
00:14:27.808 --> 00:14:31.649
The list of people we are working with.

110
00:14:31.649 --> 00:14:34.948
And our team is steadily growing.

111
00:14:34.948 --> 00:14:40.828
And we are always eager to increase our intellectual horsepower and our intellectual diversity.

112
00:14:40.828 --> 00:14:50.339
So, if you like our work, and, um, you think you would like to contribute to have something to contribute don't be shy a contact us.

113
00:14:51.749 --> 00:14:57.719
So, I have a few messages that I wanted to share. So the 1st is we are opening.

114
00:14:57.719 --> 00:15:03.178
And when I wrote down this line, I didn't expect it would have such a broad meaning.

115
00:15:03.178 --> 00:15:07.198
But it's rather innocent. What I mean, in our case.

116
00:15:07.198 --> 00:15:13.469
Um, many people on the video call. Sure. I have visited our laboratories in Santa Barbara.

117
00:15:13.469 --> 00:15:20.278
And the place we call lovingly 1, stands for Google 1 laboratories.

118
00:15:20.278 --> 00:15:25.619
But they are now being joined by a much larger facility called.

119
00:15:25.619 --> 00:15:32.428
And you may have seen this architectural drawing a few times, and we showed you this before.

120
00:15:32.428 --> 00:15:38.068
But what's exciting to us as you can see in this picture, which was just taking a few days ago.

121
00:15:38.068 --> 00:15:41.068
The facility is, uh, almost ready to go.

122
00:15:41.068 --> 00:15:46.109
Actually, if it wouldn't have been for a covert 19, we would be in there right now.

123
00:15:46.109 --> 00:15:51.899
Actually, um, a number of our dilution refrigerators preceded us already installed.

124
00:15:51.899 --> 00:15:55.139
We are taking already data out of this, um.

125
00:15:55.139 --> 00:16:00.658
Facility in a remote fashion, so if you wanted to use.

126
00:16:00.658 --> 00:16:06.359
Focus upcoming band computing service it would do this the following way. You would.

127
00:16:06.359 --> 00:16:10.499
Somebody on, let's say your favorite, um.

128
00:16:10.499 --> 00:16:20.849
Quarantine Hangout, send you a download a search from a supplier, develop an environment where you would formulate a barbarism.

129
00:16:20.849 --> 00:16:25.349
Then you send it to our service or service name is engine.

130
00:16:25.349 --> 00:16:30.839
And then you have a choice probably initially you would just go with 2 simulators.

131
00:16:30.839 --> 00:16:39.688
So those can be used central W, check whether there are any box, or, at least in a noise free environment. Your code works.

132
00:16:39.688 --> 00:16:44.818
And then you transfer it over to the actual processes.

133
00:16:44.818 --> 00:16:54.028
So we have an increasing a fleet of processes, um, available in production, the order processes, such as have 20 Cubics.

134
00:16:54.028 --> 00:17:04.769
But our fabrication team, and led by Tony, Megan, I have actually braved the comic conditions and started to fabricate a 72 cubic processes.

135
00:17:04.769 --> 00:17:09.598
And they hopefully will join the fleet, um, in production. So.

136
00:17:09.598 --> 00:17:13.138
You also have for additional scalability.

137
00:17:13.138 --> 00:17:16.499
We have added machines in Google cloud.

138
00:17:16.499 --> 00:17:19.888
As, as a high memory machines where you can.

139
00:17:19.888 --> 00:17:23.249
Simulations up to 38 kilobits.

140
00:17:23.249 --> 00:17:32.098
Sorry, I give it a little bit a sweet spot. If you go beyond this, it gets expensive on our side. We have, we have to add quite a bit of machines.

141
00:17:32.098 --> 00:17:37.169
But in my experience, I have yet to see assimilation the 40 years.

142
00:17:37.169 --> 00:17:45.179
Bits where we would see a phenomenon, we haven't seen at 38. so it's a good testing ground.

143
00:17:46.229 --> 00:17:51.989
Then there is some fest act, show many on the call already familiar with it.

144
00:17:51.989 --> 00:17:55.949
So, beyond the engine and the circle programming environment.

145
00:17:55.949 --> 00:18:03.419
You find things such as open from a library for mental simulations, particularly a chemistry.

146
00:18:03.419 --> 00:18:12.719
Electronic structure calculations, if you happen to be interested in machine, learning, burgeoning field, then you may like, 10 Central.

147
00:18:12.719 --> 00:18:16.528
Print them as a library, um, to work with.

148
00:18:16.528 --> 00:18:24.148
And then there are various other environments. Eventually there would be a vertical libraries that would support.

149
00:18:24.148 --> 00:18:28.618
Get it installed past such as certifiable random number generation.

150
00:18:28.618 --> 00:18:33.778
So, this afternoon or later, this morning, you were here.

151
00:18:33.778 --> 00:18:38.608
Talk from the true needs of this effort from the hardware side. Eric Sarah.

152
00:18:38.608 --> 00:18:41.729
And then they've bacon from the software side.

153
00:18:43.048 --> 00:18:46.169
So, the next messages that we have new toys on, let's.

154
00:18:46.169 --> 00:18:50.009
Cooperate about it, we have need towards and.

155
00:18:50.009 --> 00:18:53.249
The various, but what I wanted to show is, um.

156
00:18:53.249 --> 00:18:59.338
Tool, I think is a piece of art, and actually it was done by our.

157
00:18:59.338 --> 00:19:05.308
Physics team left 5 things here in collaboration with you.

158
00:19:05.308 --> 00:19:12.058
And what is this toy? Does anybody who has ever run fibrous on a process at it?

159
00:19:12.058 --> 00:19:18.388
No, is that you have to tune it up you have to calibrate your process. Well, before you start.

160
00:19:18.388 --> 00:19:23.338
And, for example, you have a gate operation, let's say you have a gate.

161
00:19:23.338 --> 00:19:26.368
And it's characterized by an anger feta.

162
00:19:26.368 --> 00:19:32.159
And now you need to know our put this Saturday exactly at the known value.

163
00:19:32.159 --> 00:19:35.848
So, with this new tool, we can.

164
00:19:35.848 --> 00:19:40.169
Estimates parameters the coherent parameters in our circuit.

165
00:19:40.169 --> 00:19:47.068
Better than 3 times 10 to the minus 5. moreover, as you can see in this figure, we understand the.

166
00:19:47.068 --> 00:19:54.598
Siri, um, very well. So there is Siri and experimental resides for this to line up. Really?

167
00:19:54.598 --> 00:20:01.169
Well, and it's not only a high precision state of the art decisions also. Very fast.

168
00:20:01.169 --> 00:20:09.868
So, you can actually out run calibration drift, which is also quite important. So you can see that. Um, you stay below, um.

169
00:20:09.868 --> 00:20:13.078
Rather low arrow levels during, um.

170
00:20:13.078 --> 00:20:24.028
Duration of extended computations if you invoke it in between, it is also very scalable so you can simulate painlessly jewel up all the parameters.

171
00:20:24.028 --> 00:20:27.568
In your circuit and.

172
00:20:27.568 --> 00:20:31.558
Sure, that you get on sure. Have a good.

173
00:20:31.558 --> 00:20:34.919
Probabilities that you get at the end high fidelity resides.

174
00:20:37.048 --> 00:20:46.378
So, the next message is that we, and maybe on the verge, we may be able to implement commercially or scientifically. Interesting. Um.

175
00:20:46.378 --> 00:20:50.398
Applications at this point and.

176
00:20:50.398 --> 00:20:54.898
Save it internally we have the notion of a gold standard for in this.

177
00:20:54.898 --> 00:20:58.919
Um, application, so this called stand that essentially what.

178
00:20:58.919 --> 00:21:02.878
Systems the following features, obviously you want to.

179
00:21:02.878 --> 00:21:08.398
Compute something of commercial and scientific interest, but 2nd to ideally you want to.

180
00:21:08.398 --> 00:21:12.179
Has this computational beyond the reach of classical machines?

181
00:21:12.179 --> 00:21:16.618
And said we would like to require that a circle.

182
00:21:16.618 --> 00:21:20.759
Is open sauced as this will need to.

183
00:21:20.759 --> 00:21:26.548
A growing library of good examples that as a researchers fellow researchers can build a.

184
00:21:26.548 --> 00:21:33.538
Now, even though we Baptist ago standards, that doesn't mean there isn't any other good words that can be done in the era.

185
00:21:33.538 --> 00:21:40.348
Of course, important proof of concept stepping stones 2 weeks ago and stand. That can be excellent. Work.

186
00:21:40.348 --> 00:21:45.778
Or if you have ideas for era, mitigation, and 1 tries those out, these are all on.

187
00:21:45.778 --> 00:21:55.259
Pieces of work, we support as well and let's say we support as well because we're here throughout the conference or various funding the, because we offer.

188
00:21:55.259 --> 00:22:02.278
Those who want to have a suggestion for an 1 to run it on our machines.

189
00:22:03.358 --> 00:22:09.598
So you're right if you're a little laundry list of, um, different, um.

190
00:22:09.598 --> 00:22:12.929
We have already run and.

191
00:22:12.929 --> 00:22:16.858
Actually, is the service launch and I showed you.

192
00:22:16.858 --> 00:22:21.628
Picture of where it stands right now it's important to knows that.

193
00:22:21.628 --> 00:22:26.249
This service has already been launching before a while since last year.

194
00:22:26.249 --> 00:22:32.759
And that we have these cutesy names, because for fish food, phase, dark food, phase capital.

195
00:22:32.759 --> 00:22:37.828
And then eventually we would say, um.

196
00:22:37.828 --> 00:22:41.729
The fun ends and you become more corporate and call it. Now the.

197
00:22:41.729 --> 00:22:47.699
Early access program as the door opens the open to the external world.

198
00:22:47.699 --> 00:22:55.318
Fish food, this was a phase defined by just people from our own team. The client team.

199
00:22:55.318 --> 00:23:00.568
We're running, so of course, they have inside a knowledge, they know about all the marks and, um.

200
00:23:00.568 --> 00:23:07.493
Works of hardware, and they were able to run hybrids and such as the approximate optimization.

201
00:23:08.544 --> 00:23:15.834
They prepared heartrate fork States or did interesting things, uh, looking at, uh, out of time or a correlations.

202
00:23:15.898 --> 00:23:20.398
Actually is a target, right after me, where is the details of these experiments?

203
00:23:20.398 --> 00:23:25.318
But then in the next phase, we widen the circle, we go to alphabet as a whole.

204
00:23:25.318 --> 00:23:36.509
And invited our colleagues now, have less knowledge about the of our offering. Hey, please. Um, if he'll run some algorithms.

205
00:23:36.509 --> 00:23:42.209
And we got interesting work, for example, from X, um, you will hear boxes later today.

206
00:23:42.209 --> 00:23:47.159
Whereas a simulate quantum gravity, uh, situations on the actual.

207
00:23:47.159 --> 00:23:51.749
And some other work, I'm excited about that. We may be able to I'm.

208
00:23:51.749 --> 00:23:57.148
Run at some point is to compute an aspect drive for molecules.

209
00:23:57.148 --> 00:24:00.628
Thing of proteins bound to so phases of products.

210
00:24:00.628 --> 00:24:03.959
Found for membranes and then.

211
00:24:03.959 --> 00:24:09.088
The final phase is that's opening soon and we'll run through for.

212
00:24:09.088 --> 00:24:12.449
Yes, we, um, invited many of you, um.

213
00:24:12.449 --> 00:24:17.788
Anticipated the wider community to submit, um, proposals.

214
00:24:17.788 --> 00:24:23.219
And we were, we selected the cream of the crop and we have.

215
00:24:23.219 --> 00:24:34.528
My suggestions by companies such as face craft to do a simulation of how about martyr on Michelle looking from Harvard endless suggested to study.

216
00:24:34.528 --> 00:24:39.838
Phenomenon kind of critical points so there's a nice list of, um.

217
00:24:39.838 --> 00:24:43.648
Experiments we are looking forward to running.

218
00:24:43.648 --> 00:24:46.739
And also about some of semi, which here actually.

219
00:24:46.739 --> 00:24:53.729
Tomorrow tomorrow will be when we hear from not from Google that's today tomorrow we will collaborate.

220
00:24:55.169 --> 00:25:03.358
Then the areas that I'm hands on working myself and machine learning, wanted to quickly give you an update there.

221
00:25:03.358 --> 00:25:10.469
A, sort of a set of beliefs or convictions that's shaping up it needs more work and it's definitely not.

222
00:25:10.469 --> 00:25:15.328
Problem but the following things statements seem to be true.

223
00:25:15.328 --> 00:25:23.519
That we can do basic machine learning operations, such as preparing probability distribution, describe data distributions.

224
00:25:23.519 --> 00:25:27.868
Or we can do feature maps into a high dimensional space.

225
00:25:27.868 --> 00:25:31.288
That are expensive to produce and classical machines.

226
00:25:31.288 --> 00:25:37.798
Or I should maybe say that possibly given beyond the reach of classical machines.

227
00:25:37.798 --> 00:25:42.118
And then if you use those machine learning.

228
00:25:42.118 --> 00:25:46.769
Uh, primitives, and when we apply them to a data sets.

229
00:25:46.769 --> 00:25:53.489
Is that manifestly in the sense that say results a classical data sets?

230
00:25:53.489 --> 00:25:58.169
But they reside from measuring tangled system.

231
00:25:58.169 --> 00:26:02.368
Then we seem to have advantages, let's say, in terms of sample complexity.

232
00:26:02.368 --> 00:26:08.729
What is less clear is what sees primitives also help us when we go to a general.

233
00:26:08.729 --> 00:26:18.388
Machine learning a data set and so I have to say we got some slack here. For example, we ran phantom Gan, Phantom um.

234
00:26:18.388 --> 00:26:24.449
Generative, uh, adversarial network for the same vanilla.

235
00:26:24.449 --> 00:26:34.348
Character a data set, digital data set, and we found that indeed we could produce with small, fragile distance.

236
00:26:34.348 --> 00:26:38.459
This is sort of a metric on getting good quality. Um.

237
00:26:38.459 --> 00:26:44.219
Images synthesize, but we still need.

238
00:26:44.219 --> 00:26:49.288
Some more this might have had nothing to do with resources.

239
00:26:49.288 --> 00:26:54.298
Let's say we might just have been lucky and stumbled on an architecture that worked. Well.

240
00:26:54.298 --> 00:27:01.019
So, there's definitely some more ingenuity needed here, but it's ripe for doing something interesting.

241
00:27:02.068 --> 00:27:06.808
And on that note, I wanted to explain the following.

242
00:27:06.808 --> 00:27:12.659
When we tried to discover diamonds as.

243
00:27:12.659 --> 00:27:16.439
Bob, I call it interesting.

244
00:27:16.439 --> 00:27:22.318
And we do it with our scars resource that we have is that many people. We only have that many processes.

245
00:27:22.318 --> 00:27:26.939
Then we follow the principle at Google that has served will be well and said.

246
00:27:26.939 --> 00:27:33.749
Says develop for the Pro. 1st, so if it's a note in the family loves the software product.

247
00:27:33.749 --> 00:27:40.528
Then she will explain address sibling she will explain with her parents. And so typically a good way to just.

248
00:27:40.528 --> 00:27:44.699
Software, and we have done the same thing with, um.

249
00:27:44.699 --> 00:27:53.909
Finding all the spend computing maybe we reached out to tub institutions, researchers 1st and definitely this gets resides.

250
00:27:53.909 --> 00:27:57.419
But there are 2 events that recently happens that have.

251
00:27:57.419 --> 00:28:01.979
And my question was as this should be sold approach or.

252
00:28:01.979 --> 00:28:06.148
We should start earlier than we had planned to complement despite.

253
00:28:06.148 --> 00:28:11.189
A more broad outreach approach, and since the 2 events were.

254
00:28:11.189 --> 00:28:19.169
Was working with my intern, um, Alex and 1 day blooded out. Oh, man. It is so hard to think about them.

255
00:28:19.169 --> 00:28:24.148
Interference patterns on a Hypercube we need to start in our graders to think about this.

256
00:28:24.148 --> 00:28:28.709
And there's some, none of us are really trained well, in.

257
00:28:28.709 --> 00:28:36.148
The art of crafting a, which, as we know is really the art of crafting interesting interference. Pepper.

258
00:28:36.148 --> 00:28:41.308
And sort of bringing in fresh, unspoiled minds as something can be saved for that.

259
00:28:41.308 --> 00:28:45.659
And the 2nd, um, events that happen as we are all aware of.

260
00:28:45.659 --> 00:28:51.778
Discussions around diversity and inclusion have heated up over the last month's.

261
00:28:51.778 --> 00:28:57.239
Tremendously and diversity and inclusion is all about breaking down.

262
00:28:57.239 --> 00:29:01.199
Barriers opportunity, so with a new.

263
00:29:01.199 --> 00:29:08.159
Cloud based simulators, we can essentially provide Google scale worldwide scale of.

264
00:29:08.159 --> 00:29:16.588
Access to computing resources, at least to the simulators and is there is an opportunity to.

265
00:29:16.588 --> 00:29:21.388
In April of finds Romano, John of quantum computing many of you will know.

266
00:29:21.388 --> 00:29:24.509
Ramu John was born in the 19 century.

267
00:29:24.509 --> 00:29:32.308
And rather modest circumstances taught himself mass, but came up with these amazing formulas that amazed.

268
00:29:32.308 --> 00:29:35.638
The, um, contemporary, uh, top mathematicians.

269
00:29:35.638 --> 00:29:41.729
Of his time, so if you are a kid in a town in India or a kid in a.

270
00:29:41.729 --> 00:29:49.048
Village in Kenya, as long as you have a good enough Internet bandwidth to watch low bandwidth, YouTube videos.

271
00:29:49.048 --> 00:29:53.368
You have enough Internet connectivity to participate in our.

272
00:29:53.368 --> 00:29:56.878
So so this seems to be a nice win.

273
00:29:58.888 --> 00:30:07.288
So 1, more thing we spend a lot of time since we reach beyond classical computation resides.

274
00:30:07.288 --> 00:30:12.628
Now, sing ahead and we work hard was actually a large part of the team.

275
00:30:12.628 --> 00:30:16.048
On a plan to build an error corrected clinical computer.

276
00:30:16.048 --> 00:30:20.669
Does this work was separate that by.

277
00:30:20.669 --> 00:30:26.578
Or hardware team, and with the leads, I should acknowledge here, uh, Jodi and Kelly.

278
00:30:26.578 --> 00:30:33.269
And to an immigrant and 10 of causes much broader and larger teams underneath them.

279
00:30:33.269 --> 00:30:36.989
But they have worked on, he was the.

280
00:30:36.989 --> 00:30:45.449
On the road map that looks 10 years out, but we sync to the famous Kennedy about.

281
00:30:45.449 --> 00:30:48.568
We think we can do it before the deacon ends.

282
00:30:48.568 --> 00:30:54.719
What can we do? We think by then we can build a large error. Correct? Can computer.

283
00:30:54.719 --> 00:30:58.259
Was 10 to the 6000000.

284
00:30:58.259 --> 00:31:07.318
Physical pupils, and in essence, this would be an information architecture based on the surface code supported by a trans.

285
00:31:07.318 --> 00:31:14.009
Cubans and we developed a tight schedule of well defined. There's actually a telephone books.

286
00:31:14.009 --> 00:31:17.608
Take on calculations and the designs behind this.

287
00:31:17.608 --> 00:31:21.628
Um, so we have a tied to the sequence of.

288
00:31:21.628 --> 00:31:27.509
Difficult, but if you reach milestones and eventually take us to this end goal.

289
00:31:27.509 --> 00:31:33.118
So, let me talk to you about our next big milestone, which is a demonstration.

290
00:31:33.118 --> 00:31:37.288
Of a logical human, so.

291
00:31:37.288 --> 00:31:43.259
All right, I said, rather say the demonstration of renews logical error.

292
00:31:44.729 --> 00:31:50.729
We all know, is that error correction at the end of the day works by introducing redundant.

293
00:31:50.729 --> 00:31:54.209
So you have to exit information and code in your logic to it.

294
00:31:54.209 --> 00:31:57.898
And distributed over a set of physical Cubans.

295
00:31:57.898 --> 00:32:04.048
For example, to the array of, um, I'd say, 3 by 3 data Cubans, and put forward.

296
00:32:04.048 --> 00:32:10.528
Kind of information, or you can go to larger sets of 5 by 5, 7 by 7 arrays.

297
00:32:10.528 --> 00:32:14.548
And those data and the information.

298
00:32:14.548 --> 00:32:21.628
And of course, you have to peak in circumspect way as to not to collect.

299
00:32:22.888 --> 00:32:26.638
State, so you would do this by introducing metro Cubans.

300
00:32:26.638 --> 00:32:29.969
Which essentially do everything.

301
00:32:29.969 --> 00:32:37.858
Checks and based on the measurements, you can get a sense of did anything go wrong in my circuit and correct.

302
00:32:39.419 --> 00:32:48.209
So, we can also do 2 compliments is ultimately, we know we have to correct against 2 arrows phase, flip and flip arrows.

303
00:32:48.209 --> 00:33:02.699
But to, um, work, we can also study 1 dimensional codes by sneaking in 1 the architecture into the 2 here. Right? And a study repetition code a facelift code as well.

304
00:33:02.699 --> 00:33:05.848
The advantages you can go out to higher.

305
00:33:05.848 --> 00:33:10.679
So, what we are aiming for is eventually we want to, um.

306
00:33:10.679 --> 00:33:13.828
Publish of papers that will have a figure.

307
00:33:13.828 --> 00:33:17.098
A little bit like this, and sort of, um.

308
00:33:17.098 --> 00:33:23.308
The actual figures, not actual data yet. Um, but just to show where we are going.

309
00:33:23.308 --> 00:33:28.709
Going to work so we want to show that as you increase.

310
00:33:28.709 --> 00:33:31.888
Your code this this see.

311
00:33:31.888 --> 00:33:37.348
Logic error rate comes down and so and you want to have a nice.

312
00:33:37.348 --> 00:33:42.298
Suppression factor meaning fraud as you go from 3 by 3 to 5 by 5.

313
00:33:42.298 --> 00:33:47.489
You want to see, let's say, affect the 10 reduction of your logic error rate.

314
00:33:47.489 --> 00:33:51.838
So this is what we would like to demonstrate next.

315
00:33:51.838 --> 00:33:55.949
And see how far away we have from this, we did some.

316
00:33:55.949 --> 00:34:01.288
Siri, modern to my dad always used to say is the most.

317
00:34:01.288 --> 00:34:04.739
Practical things there is a good Siri and.

318
00:34:04.739 --> 00:34:09.119
We developed a component modern.

319
00:34:09.119 --> 00:34:16.228
Which, essentially after every operations that you're using to surface code, you introduce a simple party channel.

320
00:34:17.608 --> 00:34:22.559
And then with, um, with this, um, arrow model, and, uh.

321
00:34:22.559 --> 00:34:28.048
Putting in the actual numbers, you have sort of a tool to chart.

322
00:34:28.048 --> 00:34:31.409
It's a route into the future for example, we would see is that if.

323
00:34:31.409 --> 00:34:41.128
Today, he's a different boss tell you. So, how bad is ours? Easy arrow. How bad is a hot error? Sorry? It's the other way round too small 1 out of my error.

324
00:34:41.128 --> 00:34:47.548
Susie error it is something we found that surprised us and was not properly.

325
00:34:47.548 --> 00:34:52.708
Accounted for in the literature, so far is idling arrows matter. Lot.

326
00:34:52.708 --> 00:34:56.369
And, of course, in hindsight, not surprising if your cube, it is sitting there.

327
00:34:56.369 --> 00:35:00.208
And just suffer from T1 processes then.

328
00:35:00.208 --> 00:35:03.869
This is your overall error accumulates.

329
00:35:03.869 --> 00:35:07.108
So, if it applies us a simple arrow model.

330
00:35:07.108 --> 00:35:10.139
And we would run it today.

331
00:35:10.139 --> 00:35:13.559
Actually, our, our suppression factor would be smaller than 1.

332
00:35:13.559 --> 00:35:17.639
That means like, your arrow would increase as we goes to larger race.

333
00:35:17.639 --> 00:35:22.409
But it also gives us Nadia how we can.

334
00:35:22.409 --> 00:35:25.409
Reduce the different Aero budgets.

335
00:35:25.409 --> 00:35:32.099
And then makes approval of principal work and then eventually, our road map target is to get, um.

336
00:35:32.099 --> 00:35:40.619
The overall suppression fact that you about 10, and we definitely do think this is feasible based on calculations. We made.

337
00:35:40.619 --> 00:35:44.248
And based on what I said about the idling error.

338
00:35:44.248 --> 00:35:47.639
Seem to be 2 basic roads forward.

339
00:35:47.639 --> 00:35:57.088
1, is you reduce your error, or you reduce your cycling time, which is cycle time, which is, of course, good in its own right? You have the faster computers this way.

340
00:35:57.088 --> 00:36:00.778
So, you focus on circuit design to make sense better.

341
00:36:00.778 --> 00:36:04.289
And for our teams a little bit in our D. N. A.

342
00:36:04.289 --> 00:36:08.188
Printed on us by John muddiness, but.

343
00:36:08.188 --> 00:36:14.369
You can also, um, complimentary approach. They both can lift next each other where you focus more.

344
00:36:14.369 --> 00:36:19.079
On materials research, let's as a school of thinking would look more into this.

345
00:36:19.079 --> 00:36:25.528
Direction so here you would aim at longer coherence times. Let's say hundreds of micro seconds.

346
00:36:25.528 --> 00:36:35.099
Where if you make a cycle time faster, you may get away with tens of microseconds. And of course, you can do both prospects.

347
00:36:36.298 --> 00:36:41.938
So, unfortunately, the simple model is not quite enough. There are extra.

348
00:36:41.938 --> 00:36:46.409
Additional channels as leakage that calibration drift.

349
00:36:46.409 --> 00:36:49.708
As crosstalk the correlated arrows.

350
00:36:49.708 --> 00:36:54.809
We understand that correlated arrows have always been a man of credit of error correction.

351
00:36:54.809 --> 00:36:58.619
They're not good threshold. serums known today.

352
00:36:58.619 --> 00:37:02.699
The physics team started to work on the executives.

353
00:37:02.699 --> 00:37:06.539
And correlated arrows are not sort of a theoretical thing.

354
00:37:06.539 --> 00:37:11.728
That easily your curve and, for example, the leakage state decays into.

355
00:37:11.728 --> 00:37:14.938
The states that are valid in your code space.

356
00:37:14.938 --> 00:37:19.409
Or a much more dramatic example of correlated errors.

357
00:37:19.409 --> 00:37:23.579
Is imprints the impact of cosmic raise.

358
00:37:23.579 --> 00:37:28.409
On our better raise and we do seasons. Actually, this is a trace here.

359
00:37:29.489 --> 00:37:33.929
25 microsecond duration where we actually just run.

360
00:37:33.929 --> 00:37:38.309
Repetition code experiments and we do see those burst.

361
00:37:38.309 --> 00:37:41.789
Of arrows happenings that happen in the in the.

362
00:37:41.789 --> 00:37:44.938
Basically and temporary correlated.

363
00:37:44.938 --> 00:37:51.329
Manner so now he's saying, oh, my God, we will have to take kind of computers down to your mind shaft.

364
00:37:51.329 --> 00:37:55.949
No, it's not quite that bad. Um, we just will have to install phone on tracks.

365
00:37:55.949 --> 00:38:01.648
But this is something you will have to take care of. So.

366
00:38:03.119 --> 00:38:07.378
This was attempt to say 2 milestones and of course, you would just.

367
00:38:07.378 --> 00:38:10.469
Make your code distance large enough is that you.

368
00:38:10.469 --> 00:38:17.550
Approach about a 1000 physical Cubans at that time if everything goes well, and you have a lot of suppression factor.

369
00:38:17.550 --> 00:38:21.179
You should have coherence times of years are what I like to say.

370
00:38:21.179 --> 00:38:25.530
Essentially your device should stay here and until you switch off your computer.

371
00:38:25.530 --> 00:38:31.860
Then there is a milestone that I should maybe point out it's important from an investment and financing perspective.

372
00:38:31.860 --> 00:38:35.190
Eventually you will have 2 logical Cubans at hand.

373
00:38:35.190 --> 00:38:41.130
And you have the 4 gates that, between that? Hopefully not just different Gates, but a complete.

374
00:38:42.539 --> 00:38:46.440
And at that point, you really have, but maybe call the.

375
00:38:46.440 --> 00:38:49.800
Integrated search of classical computing.

376
00:38:49.800 --> 00:38:53.550
You have the title it's the module.

377
00:38:53.550 --> 00:38:58.230
And to build a large machine, you would just takes us module and to replicate it.

378
00:38:58.230 --> 00:39:02.070
And this is not too far out 5 6.

379
00:39:02.070 --> 00:39:07.050
Yes, um, somewhere there hopefully and it's important because.

380
00:39:07.050 --> 00:39:12.989
Pass the points and risk profile for Ben computing comes way down. It becomes more of a risk profile.

381
00:39:12.989 --> 00:39:18.750
Building a high rise of freeway, rather than doing cutting edge electrons.

382
00:39:20.070 --> 00:39:23.429
And now fast forward to the final milestone.

383
00:39:23.429 --> 00:39:28.650
You know, want to give you a picture how we envision.

384
00:39:28.650 --> 00:39:33.239
1M Cubics error, corrected, panel computer to look like.

385
00:39:33.239 --> 00:39:36.690
Well, consist of these modules that we tire.

386
00:39:36.690 --> 00:39:45.389
And let's say there is like a quite a book of technical information that went into this, a design.

387
00:39:45.389 --> 00:39:48.449
And something you see here.

388
00:39:48.449 --> 00:39:52.829
Scale is about right with the figure, a human figure a place there.

389
00:39:52.829 --> 00:40:04.500
So, there's sometimes this misconception oh superconducting queue where they require these thick wires and you're willing eventually need a factory floor full of dilution refrigerators reviews.

390
00:40:04.500 --> 00:40:14.250
Expensive and really this is actually not true if you spend careful time designing and thinking about your control electronics about wiring solutions.

391
00:40:14.250 --> 00:40:20.429
Then you see, and what it would be able to deliver a product in a manageable size.

392
00:40:21.599 --> 00:40:25.050
So, with this, I want to summarize a little bit.

393
00:40:25.050 --> 00:40:28.409
And also a issue, I'm a car to.

394
00:40:28.409 --> 00:40:31.679
Community to help us on various aspects.

395
00:40:31.679 --> 00:40:36.030
So, we feel we have a good plan, a plan of record that says.

396
00:40:36.030 --> 00:40:43.349
To the array of turns on cube can be used to enable the surface code information. Architecture.

397
00:40:43.349 --> 00:40:47.309
You will be able to drop it lowers the arrows threshold.

398
00:40:47.309 --> 00:40:50.550
Require but.

399
00:40:50.550 --> 00:40:56.010
There are places where maybe worth the 2.

400
00:40:56.010 --> 00:40:59.429
Invest in parallel efforts and we.

401
00:40:59.429 --> 00:41:03.119
Names those efforts, renegade efforts.

402
00:41:03.119 --> 00:41:09.989
And, for example, 1, renegade effort, we are thinking seriously about actually we have already fund external groups to look into this.

403
00:41:09.989 --> 00:41:13.260
Would it make sense to build the more protected to.

404
00:41:13.260 --> 00:41:17.760
Cubic Isis, I'm not Obtainium that is still out of reach.

405
00:41:17.760 --> 00:41:21.659
Or, but it feels to us as this community that has been thinking.

406
00:41:25.019 --> 00:41:31.139
And such cube of costs to give us more in terms of.

407
00:41:32.280 --> 00:41:37.650
Cause you need your cycle times, will come down.

408
00:41:37.650 --> 00:41:42.719
So, it's definitely something we're looking very serious, um, to possibly have.

409
00:41:43.860 --> 00:41:48.090
In the house, but we also read on the community to.

410
00:41:48.090 --> 00:41:52.889
You look into such alternatives. Another area is.

411
00:41:52.889 --> 00:41:56.429
Ask the question is spent an error correction.

412
00:41:57.570 --> 00:42:12.119
Your surface code possibly and depending on you ask if you go to our own surface code expert, Austin, you will tell you now, forget it. You know, if you have to the lettuce of.

413
00:42:13.679 --> 00:42:18.179
As you are confined to surface codes, anything you do will look like surface.

414
00:42:18.179 --> 00:42:27.239
And I haven't heard this directly, but it was conveyed to me that John presco 1 side many people have tried to do something beyond the surface codes that have failed.

415
00:42:27.239 --> 00:42:32.309
So, maybe we're stuck with this.

416
00:42:32.309 --> 00:42:36.690
But we are not ready to give up yet and we, um.

417
00:42:36.690 --> 00:42:43.320
Definitely interested in how it's.

418
00:42:43.320 --> 00:42:47.940
Put more autonomous forms of error correct?

419
00:42:47.940 --> 00:42:52.920
And then the last before I should issue is.

420
00:42:52.920 --> 00:42:56.039
It's the scarcity of I.

421
00:42:56.039 --> 00:43:01.079
So, what would we do with this concrete machine? What would we do.

422
00:43:01.079 --> 00:43:09.030
Was a 1M cubic machine does is a little bit more concrete question than just saying oh, there's a scaling advantage.

423
00:43:09.030 --> 00:43:14.219
No, for the cycle times we have for the remaining error rates that would be there.

424
00:43:14.219 --> 00:43:21.929
What are attractive commercially relevant? Valuable or scientifically interesting.

425
00:43:21.929 --> 00:43:27.090
To run and they are definitely we rely on the community to really create.

426
00:43:27.090 --> 00:43:32.880
Any interesting ideas is directions and then was that I wanted to.

427
00:43:32.880 --> 00:43:36.239
Thank you all for dialing in and also is a little bit.

428
00:43:36.239 --> 00:43:42.000
You all know long video conferences are challenged, but we try to make it as fun as possible.

429
00:43:42.000 --> 00:43:46.079
And I really hope you're going to enjoy the summer supposed.

430
00:43:58.349 --> 00:44:05.159
Did you lose the video towards the end of that or not?

431
00:44:06.750 --> 00:44:17.820
Could you still see the video you did lose the video for the last part?

432
00:44:17.820 --> 00:44:22.199
Yes, same. Okay.

433
00:44:22.199 --> 00:44:31.199
Well, there wasn't much at the end so what was going on there I stopped sharing and sharing again.

434
00:44:33.119 --> 00:44:36.869
Silence.

435
00:44:42.119 --> 00:44:46.650
Silence.

436
00:44:51.599 --> 00:44:57.570
Okay, I love computers.

437
00:44:59.849 --> 00:45:03.239
Silence.

438
00:45:03.239 --> 00:45:07.710
Silence.

439
00:45:11.429 --> 00:45:16.920
Silence.

440
00:45:18.869 --> 00:45:22.380
Um.

441
00:45:23.550 --> 00:45:31.710
Okay, I'm thinking I'm going to have to leave and re enter so don't go away. It's.

442
00:45:32.730 --> 00:45:39.239
I don't know, I'd probably have to kill and restart this Webex session. So don't go away. I'll be back in a minute.

443
00:45:39.239 --> 00:45:58.679
Silence.

444
00:46:05.280 --> 00:46:11.579
I don't know what went wrong there, but the computer's involved so okay.

445
00:46:13.320 --> 00:46:21.269
Next okay, so I thought that was very interesting. You see google's.

446
00:46:22.500 --> 00:46:25.829
And you're not saying that okay.

447
00:46:25.829 --> 00:46:29.429
Just a 2nd, here, something's not sharing right?

448
00:46:39.719 --> 00:46:43.139
Good.

449
00:46:43.139 --> 00:46:49.139
Okay, so some takeaways from that keynote address.

450
00:46:49.139 --> 00:46:59.400
Is well, Google looks very confident. Error correcting is very important. I haven't talked about it that much. I should probably put in a blurb on that at some point.

451
00:47:00.480 --> 00:47:06.119
So, there's a lot of work on different types error correcting. That would be the big.

452
00:47:06.119 --> 00:47:11.309
Message from there. Okay. Um.

453
00:47:11.309 --> 00:47:18.510
Next more Google stuff so they have this quantum cast page here and you're free to wander around this the interesting stuff.

454
00:47:20.190 --> 00:47:27.360
But I have a few things from here quantum supremacy explained since that was in the news, a short 1.

455
00:47:27.360 --> 00:47:31.650
A longer 1 and fun sort of, think quantum chess and large here.

456
00:47:31.650 --> 00:47:38.250
Imagination something from Peter shore who did the.

457
00:47:38.250 --> 00:47:41.579
Factoring thing and.

458
00:47:41.579 --> 00:47:51.269
I'm mentioning, I want to show you something by Peter shore, because I want to show you stuff from famous people so you can see what he's having fun with. Now. It's extremely technical.

459
00:47:51.269 --> 00:48:03.179
Which means, I don't understand it either, but I'll show you the 1st, few minutes at least and then I'll finish off. I'll run to the end of the class showing you a deeper thing by Marissa. Just Dana.

460
00:48:03.179 --> 00:48:07.710
On getting down to more details on.

461
00:48:07.710 --> 00:48:15.510
Building their quantum computer us do quantum supremacy? Explain it.

462
00:48:21.264 --> 00:48:30.715
My team,

463
00:48:31.224 --> 00:48:34.764
and today we're going to talk about that experiment we're working on,

464
00:48:34.885 --> 00:48:36.804
which is known as quantum supremacy.

465
00:48:37.945 --> 00:48:46.735
The latest experimental quantum process are produced at Google com, has 72 Cubans or quantum beats. We're testing quantum. Very good.

466
00:48:46.735 --> 00:48:59.965
Some result with a goal of reducing errors by their nature quantum Gates, half a priority of errors and errors during cross quantum circuits. Recalibrate quantums. Very good. We're reading down the property of her.

467
00:49:01.074 --> 00:49:11.425
We simulate quantum circuits with traditional computers to benchmark uncanny rate quantum circuits as we work to reduce the probability of an error. Simulations gets exponentially harder.

468
00:49:11.934 --> 00:49:16.764
This means that it gets too computationally intensive, even for a supercomputer to keep up.

469
00:49:17.280 --> 00:49:21.420
From this, we get the name quantum supremacy for this experiment.

470
00:49:21.420 --> 00:49:30.510
This has to do with something called a strong church during thesis in computer science, traditional computers from the.

471
00:49:30.510 --> 00:49:39.025
To your laptop implement equivalent integrations or classic. Okay. Although I'm ordering computer is, of course, much much faster.

472
00:49:39.505 --> 00:49:47.875
This trial church during thesis is that all universal computers are equivalent in this way, I can simulate each other efficiently.

473
00:49:48.985 --> 00:49:59.695
But according to quantum computing, the strong church during thesis is false on quantum computers can solve some problems exponentially faster than other universal computers.

474
00:50:00.235 --> 00:50:05.155
So, what we're trying to do is kind of breaking the strong church thesis.

475
00:50:06.000 --> 00:50:10.260
You can think of it as an arrow pointing to some direction and my sphere.

476
00:50:10.260 --> 00:50:19.469
Quantum gates are operations and candidates similar to classical Nate. We often the quantum Gates as boxes. We, they put on 1 side.

477
00:50:19.469 --> 00:50:25.019
And they are put on the opposite side in a quantum circuit. We apply layers of Gates.

478
00:50:25.019 --> 00:50:31.590
1 per clock cycle, a measurement of the quantums very good, produces a string of bits.

479
00:50:31.590 --> 00:50:36.750
For the quantum supremacy experiment, we choose the quantum gates at random.

480
00:50:36.750 --> 00:50:41.219
This is a Hello world program for quantum computers.

481
00:50:41.219 --> 00:50:53.969
Crucially, in this case, we have the strongest theoretical evidence against the strong church during thesis. It takes exponential time to simulate a random quantum circuit with a classical computer.

482
00:50:53.969 --> 00:51:04.469
According to quantum mechanics, every party call can also act as a wave and this applies to kids from a state of a quantum computer.

483
00:51:04.469 --> 00:51:09.269
See, the video cut out yeah.

484
00:51:09.269 --> 00:51:15.840
I just saw it. Okay.

485
00:51:17.280 --> 00:51:20.670
Having networking issues today.

486
00:51:22.349 --> 00:51:26.880
Okay.

487
00:51:30.420 --> 00:51:35.849
The quickest thing here it seems to be if I leave and re, enter the meeting, I think.

488
00:51:43.920 --> 00:51:49.139
Sorry.

489
00:52:09.719 --> 00:52:14.909
Silence.

490
00:52:21.030 --> 00:52:30.030
Silence.

491
00:52:31.349 --> 00:52:35.489
Silence.

492
00:52:48.420 --> 00:52:51.570
Silence.

493
00:52:56.579 --> 00:53:05.639
Silence.

494
00:53:07.829 --> 00:53:11.639
Silence.

495
00:53:11.639 --> 00:53:19.949
Silence.

496
00:53:21.059 --> 00:53:37.710
Silence.

497
00:53:37.710 --> 00:53:41.340
Silence.

498
00:53:45.449 --> 00:53:49.619
Okay.

499
00:53:50.639 --> 00:53:54.510
I don't know why, what I did is I left the meeting and rejoined to meeting and that.

500
00:53:54.510 --> 00:53:59.309
Got hit working again onward.

501
00:53:59.309 --> 00:54:06.719
Takes an exponential number of waves or computation or parts. This is the property that we're testing.

502
00:54:06.719 --> 00:54:09.960
Doubt but the state of around the circuit.

503
00:54:09.960 --> 00:54:13.440
Looks like the speckles of a laser.

504
00:54:13.440 --> 00:54:28.195
This is a fingerprint of the chronic concern for some beat the strings. The commendation of parts interfere constructively. An intensity of the output probability grows for others. The computational path interfere.

505
00:54:28.195 --> 00:54:28.974
Destructible.

506
00:54:29.610 --> 00:54:37.199
And the output probably increases simulating interference of this financial number of computational parts.

507
00:54:37.199 --> 00:54:40.349
In the quantum circuit thanks exponential time.

508
00:54:40.349 --> 00:54:44.099
We can check if we obtain the correct fingerprint in the experiment.

509
00:54:44.099 --> 00:54:50.130
Master of the property for her 1st we get around medium B, the streams from the 1. that won't be there.

510
00:54:50.130 --> 00:54:58.500
This takes a few seconds now, we use an expensive classical simulation to take if those bitter strings have high probability.

511
00:54:58.500 --> 00:55:03.659
If this is the case, their rate is low an experiment has to see that.

512
00:55:03.659 --> 00:55:10.650
Implication will be the condom. Computers seem to be breaking the strong church during.

513
00:55:10.650 --> 00:55:17.789
As we reduce our farther, we expect to see a seminar exponential speed up for practical product. So what's next.

514
00:55:17.789 --> 00:55:23.909
The other videos in this area is to learn more about how quantum computer works, how to permit.

515
00:55:23.909 --> 00:55:38.010
You can also visit the to learn more about how quantum computers can be used to solve problems in chemistry a material science for check out. The links included below. Thank you.

516
00:55:39.150 --> 00:55:46.650
Okay, quantum supremacy Here's a fun sort of thing. Different crazy things you can do with quantum computing.

517
00:55:48.719 --> 00:56:03.239
30 minutes, so Hi, everyone now we'll be talking about a new tool for quantum education and it also happens to be a fun game.

518
00:56:03.239 --> 00:56:09.030
So, quantum computing can be and very intimidating to 1st, get involved with.

519
00:56:09.030 --> 00:56:16.139
Now, more than ever, it's really critical that we figure out how to make this fields, and the opportunities that it can offer more accessible to everyone.

520
00:56:16.139 --> 00:56:22.289
And that's where quantum chess comes in. It's like regular chess, but with the ability to also make quantum moves.

521
00:56:22.289 --> 00:56:28.739
Yeah, no, I lost the video again.

522
00:56:30.869 --> 00:56:35.130
Silence.

523
00:56:44.519 --> 00:57:06.630
Silence.

524
00:57:12.510 --> 00:57:18.900
If this happens again, I think we give up for today, but we'll see.

525
00:57:18.900 --> 00:57:24.690
Last year Google partnered with Chris in the quantum outreach lead at Caltech.

526
00:57:24.690 --> 00:57:27.809
To build this out further we had 2 goals in mind.

527
00:57:27.809 --> 00:57:33.690
1st, we wanted to provide a fund and approachable way for anyone to learn the fundamental concepts of quantum mechanics.

528
00:57:33.690 --> 00:57:38.010
And 2nd, we wanted to create a case study to teach aspiring developers.

529
00:57:38.010 --> 00:57:44.190
How to build a project like this on google's open source Python framework for creating and executing quantum circuits.

530
00:57:44.190 --> 00:57:48.630
So, today we're going to cover how quantum chest works on a classical computer.

531
00:57:48.630 --> 00:57:53.519
Show some examples of kind of moves you can make and also show how we've adopted it to run on search.

532
00:57:53.519 --> 00:57:58.110
So, now I'll hand it over to Chris to explain in more detail.

533
00:57:59.400 --> 00:58:12.295
Thanks Megan. Hi, everyone, I'm Chris Kent while I was the original creator of quantum chess. I'm going to talk to you a little bit about the motivation behind the game, and some of the game play aspects of quantum chess.

534
00:58:12.594 --> 00:58:18.894
So, if you were here earlier, you heard, Hartnett mentioned the desire to teach interference patterns to 3rd graders.

535
00:58:19.139 --> 00:58:34.014
But, as he actually pointed out as humans, we aren't really built to easily grasp these concepts. I think that's because we don't really interact with quantum physics on a daily basis as far as we can observe.

536
00:58:34.014 --> 00:58:41.844
Right? I like to use the analogy of gravity a lot. In this case, I'll be talking about classical, grabbing that problem. Gravity.

537
00:58:43.469 --> 00:58:51.204
I think we all have sort of an intuitive understanding of how gravity works. I mean, we know if you drop a ball, it's going to fall and hit the floor.

538
00:58:51.534 --> 00:59:06.474
You can make pretty good guesses about if you throw something how far it's going to go before it touches down and you don't need to know all of the complex math behind gravity to be able to make these intuitive guesses about how it's going to work or what's going to happen in

539
00:59:06.474 --> 00:59:07.644
certain situations?

540
00:59:08.094 --> 00:59:19.704
So I thought that if we could give people a way to interact with quantum mechanics, any more tangible way, some of these intuitive understandings might develop.

541
00:59:20.065 --> 00:59:23.664
And I thought again would be a good way to do that.

542
00:59:24.030 --> 00:59:30.510
And so I set out to build a board game on top of a quantum simulation.

543
00:59:30.510 --> 00:59:34.260
I decided that the roles would enforce.

544
00:59:34.260 --> 00:59:44.429
All of movement needed to be implemented through unitary evolution and that the player would be able to create superposition.

545
00:59:44.429 --> 00:59:51.630
In the game, and I thought if you have those 2 requirements, then you might get other quantum effects.

546
00:59:51.630 --> 00:59:56.699
Entanglement and interference for free and players might be able to come up with.

547
00:59:56.699 --> 01:00:01.349
Gameplay strategies that actually used quantum mechanics.

548
01:00:01.349 --> 01:00:06.570
So, 1 way to think about movement in a.

549
01:00:06.570 --> 01:00:09.599
Or game is through.

550
01:00:09.599 --> 01:00:14.639
Swapping the occupancy state of.

551
01:00:14.639 --> 01:00:19.559
The the squares on the board, so if a piece is in the square, you can think of it as a 1.

552
01:00:19.559 --> 01:00:24.360
And if they piece isn't any Square, you can think of it as a 0T. So if you want to move a piece.

553
01:00:24.360 --> 01:00:28.679
From point a to point B, you swap the 1 0.

554
01:00:28.679 --> 01:00:42.085
And you've effectively moved where that piece is, this leads to a natural unitary for implementing movement on top of a quantum state. It's the swap unitary or the swap in the game.

555
01:00:42.085 --> 01:00:48.204
I chose the swap for a number of reasons, including possible future hardware implementations.

556
01:00:49.644 --> 01:00:57.324
Once you have that unitary, you also have a natural way for players to create superposition through the square root of a nice swap.

557
01:00:57.630 --> 01:01:02.159
Which effectively puts a piece in a superposition of having moved and not moved.

558
01:01:02.159 --> 01:01:08.610
At the same time. So with those 2 Unitarians, I was able to build up.

559
01:01:08.610 --> 01:01:17.489
The rules of quantum chess, or of chess, on top of the quantum simulation of 64 Cubans. The main difference is.

560
01:01:17.489 --> 01:01:21.780
He says now have access to a split move. So, here on the left, you see a board.

561
01:01:21.780 --> 01:01:26.369
I can try and split this night to occupy both Avery and C3.

562
01:01:26.369 --> 01:01:31.860
And what you're presented with is a probability distribution of finding the night on a free.

563
01:01:31.860 --> 01:01:38.460
Or finding it on C3 what's actually going on in the back end is we have a superposition of possible boards.

564
01:01:38.460 --> 01:01:42.210
1, where the night moves to April, you seen here on the right.

565
01:01:42.210 --> 01:01:46.050
And the other where, the night moved to C3 C on the left here.

566
01:01:46.050 --> 01:01:57.269
So, that's the primary way. Players can create superposition in the game. Another effect that you can get pretty easily out of the game is entanglement.

567
01:01:57.269 --> 01:02:06.750
So 1 of the moves you can do in chess, is this on here on C2 could normally move forward 2 squares if it's its 1st move.

568
01:02:06.750 --> 01:02:12.539
Now, on 1 board, there's a night, they're blocking the way. So if you're trying to do this move.

569
01:02:12.539 --> 01:02:17.070
You actually end up with a entangled state.

570
01:02:17.070 --> 01:02:20.969
Where the night, if the night was their blocking.

571
01:02:20.969 --> 01:02:29.550
Upon didn't complete the move and if the night wasn't there blocking the punted complete the move. This is done with a controlled. I swap.

572
01:02:29.550 --> 01:02:36.449
On the Cubans that represent those squares, so you can see here on the right that if the night is there.

573
01:02:36.449 --> 01:02:40.769
The pond did nothing if night wasn't there, the pump did move forward.

574
01:02:40.769 --> 01:02:46.139
And you also get any phase of AI because we call that the movement is accomplished with an AI swap.

575
01:02:46.139 --> 01:02:53.010
The final effect that is important in the game, at least for our discussion is measurement.

576
01:02:53.010 --> 01:03:01.679
So, there are times when 2 pieces that are different, might interact. So if I wanted to try and move this queen from 1 to see to.

577
01:03:01.679 --> 01:03:11.789
There's a board in the superposition where the pod is there and that wouldn't normally be a legal move and there's another board upon isn't there so, that is a legal move.

578
01:03:11.789 --> 01:03:18.510
The game enforces the rule that you always have to be able to save what type of cases on a square.

579
01:03:18.510 --> 01:03:22.710
And so what it will do is actually measure the C2 Square.

580
01:03:23.940 --> 01:03:29.820
And you can see that here and it finds that the pond is there.

581
01:03:29.820 --> 01:03:37.320
Which noticed collapses the superposition such that the night was here blocking its move. And then the queen wasn't able to complete that move.

582
01:03:37.320 --> 01:03:44.159
Measurements are non deterministic, so you can get the opposite outcome. If we tried this a few more times.

583
01:03:44.159 --> 01:03:51.389
We might see that opposite outcome might get lucky on, like, the 3rd try or.

584
01:03:51.389 --> 01:03:54.780
Maybe, we'll have to try it 5 times.

585
01:03:54.780 --> 01:03:58.619
I've gotten lucky in the past, and it's always been on, like the 2nd or 3rd try.

586
01:03:59.789 --> 01:04:02.880
We'll try it 1 or 2 more times and then I'll just move on.

587
01:04:04.860 --> 01:04:11.550
There you go, so you see the other outcome as well where the night was over in C3 or it's found to me and, I mean, 80.

588
01:04:11.550 --> 01:04:16.139
The night is found to be a 3 and so this pawn did move forward.

589
01:04:16.139 --> 01:04:19.230
The state collapses to that state where.

590
01:04:19.230 --> 01:04:24.269
This queen can then successfully move to C2. So.

591
01:04:24.269 --> 01:04:33.210
This was a quick overview of the rules of the game and how it works. This project with Google has been to try and.

592
01:04:33.210 --> 01:04:38.159
Get the game to a point where it could export some of this quantum behavior to.

593
01:04:38.159 --> 01:04:42.480
An external resource that could maybe then run stuff using circ.

594
01:04:42.480 --> 01:04:46.380
Um, and so now it supports an API.

595
01:04:46.380 --> 01:04:49.679
That we can implement and we have implemented.

596
01:04:49.679 --> 01:04:55.829
With Cirque, and we'll be able to see some of that in effect. Now I'm going to pass it over to Doug.

597
01:04:55.829 --> 01:05:04.920
Thank you so we know that there are significant challenges from bringing an algorithm from theory into practice.

598
01:05:04.920 --> 01:05:16.679
And we've seen these challenges already in their earlier presentations, and we're going to continue to see them again. And again, as more experiments move from theory to using this hardware.

599
01:05:16.679 --> 01:05:20.099
And since quantum chess.

600
01:05:21.085 --> 01:05:33.355
Doesn't require a lot of specific domain knowledge. We think this is going to be a great proof of concept, to illustrate these challenges and really boil down the essentials of what is going to take to get an algorithm ready for the.

601
01:05:34.434 --> 01:05:41.125
So, to that end, we're going to be introducing a new tutorial insert that's going to educate users through all of these stats from.

602
01:05:41.369 --> 01:05:50.485
Constructing circuits from quantum chest moves all the way to preparing their experiments for noisy hardware. So, among the challenges we're going to look at are cubic mappings.

603
01:05:50.815 --> 01:06:05.755
So, each algorithm is going to need to map logical cubital in this case squares on a quantum chess board into physical Cubics that match the topology and the connectivity of the underlying hardware device. And it really has to do this dynamically.

604
01:06:05.755 --> 01:06:14.034
So your day is at rune, when you find out 1 of the cubex, you were using has dropped out. There is performing badly. So, sir can help with this. And the tutorial is going to show you.

605
01:06:16.525 --> 01:06:28.465
Every algorithm is going to need gate compiling and decomposition, so hardware doesn't support every gate. So for instance, this entanglement needs a controlled ice swap. So that's a 3 cubic gate that no hardware is going to support.

606
01:06:28.465 --> 01:06:34.434
So we need to translate that into gates hardware can support. So cert can help with this too.

607
01:06:35.005 --> 01:06:48.505
And lastly, and most importantly, if you want to run in a quantum computer in the next decade, you have to deal with noise and being able to run with a noisy sampler is 1 step from taking your algorithm from perfect simulation to actually running on hardware.

608
01:06:48.505 --> 01:06:56.605
And that's where we're going to show next so we're going to run through these same moves that Chris showed, but now we're going to run it through a noisy sampler concert.

609
01:06:56.940 --> 01:07:10.465
So the view changed a little bit. So, let me explain what you're looking at, and on the left side, you can see the quantum board as before and on the right side you can see a console output of what's going on behind the scenes.

610
01:07:10.494 --> 01:07:19.974
And a few things that it's doing are constructing the, the circuit from the moves. So you can see a text version of that circuit there.

611
01:07:19.974 --> 01:07:34.195
And that's something that comes with serve is then dynamically compiling this into hardware Gates and based on the moves. You've done mapping the squares on the chess board, into physical Cubans. Leslie is running it on a noisy sampler.

612
01:07:34.195 --> 01:07:48.985
That's going to add noise to each gate and then it's going to perform some basic error mitigation to cancel out some of that noise and take get rid of outcomes that don't make any sense. Like, if pieces appear or disappear for no reason.

613
01:07:50.275 --> 01:07:50.695
So,

614
01:07:51.085 --> 01:07:52.014
as this is running,

615
01:07:52.014 --> 01:07:59.364
the last thing I'm going to mention is that while cert does have support for adding noisy models to simulation this noisy sampler is,

616
01:07:59.364 --> 01:07:59.965
in fact,

617
01:08:00.385 --> 01:08:03.054
a little bit special because it's in fact,

618
01:08:03.054 --> 01:08:03.655
a noisy,

619
01:08:03.655 --> 01:08:04.795
intermediate scale,

620
01:08:04.795 --> 01:08:05.695
quantum device,

621
01:08:05.695 --> 01:08:10.735
and is actually running live on our quantum processor in the new data center insight,

622
01:08:10.735 --> 01:08:11.605
Santa Barbara,

623
01:08:11.905 --> 01:08:18.265
and it's using this the google's quantum computing service with serve to call the quantum engine API right now.

624
01:08:18.265 --> 01:08:26.005
So, what you're really seeing right now is quantum chest moves occurring in real time, using a real quantum computer.

625
01:08:26.845 --> 01:08:27.204
So,

626
01:08:27.204 --> 01:08:31.314
in order to generate all these probability statistics that are generating these nice,

627
01:08:31.314 --> 01:08:38.034
fiery animate Nations is requesting a 1000 samples more post selection is required,

628
01:08:38.274 --> 01:08:42.774
sending it to the actual quantum computer and getting results back and interpreting this.

629
01:08:43.074 --> 01:08:50.965
And all of this is happening, and within about 3 to 5 seconds per mode so you can kind of see this happening as you're watching.

630
01:08:52.045 --> 01:09:02.305
So, as well, as seeing a live demonstration of our quantum computing service, you're also probably witnessing what is probably 1 of the most expensive board games ever run as well.

631
01:09:02.725 --> 01:09:07.135
And with that, I'm going to hand it back to Megan to get some closing remarks.

632
01:09:07.380 --> 01:09:16.890
Amazing Thank you, Doug. So I hope you all enjoy this brief introduction to quantum chess. Now I'll quickly discuss some of our next steps.

633
01:09:16.890 --> 01:09:20.670
So, 1st, we want to apply quantum chess to education.

634
01:09:20.670 --> 01:09:27.210
We've been working on improving the user experience, like creating puzzles where you need to make quantum moves in order to win.

635
01:09:27.210 --> 01:09:32.100
And making the game as easy as possible for everyone to understand, including middle and high school students.

636
01:09:32.100 --> 01:09:37.289
As I've mentioned, we created a quantum chess circuit tutorial, so anyone can see start to finish.

637
01:09:37.289 --> 01:09:44.399
How to run this kind of project on quantum hardware you want to spread the word about the game and get it integrated into some courses and lesson plans.

638
01:09:44.399 --> 01:09:47.460
So, if you may be interested in trying this out, please let us know.

639
01:09:47.460 --> 01:09:51.930
2nd, we're really excited about the future applications to quantum.

640
01:09:51.930 --> 01:09:57.420
Games like, chess have historically been very important for providing a basis for development of classical, a.

641
01:09:57.420 --> 01:10:02.489
Quantum chess could play a very similar role, empowering the development of quantum a.

642
01:10:02.489 --> 01:10:07.710
And for quantum machine learning, there's a better chance of finding quantum advantage on quantum data.

643
01:10:07.710 --> 01:10:11.789
Quantum chest moves provide a natural way for us to generate this data.

644
01:10:11.789 --> 01:10:17.670
And 3rd, with the ability to make quantum moves, the game of chess can now become a lot more interesting.

645
01:10:17.670 --> 01:10:22.470
So, tomorrow we're hosting an optional activity where you can see quantum chess in action.

646
01:10:22.470 --> 01:10:26.760
1st, you can watch a chess game between Chris, the creator of quantum chess.

647
01:10:26.760 --> 01:10:30.119
And Conrad, a chess grand master, and then.

648
01:10:30.119 --> 01:10:33.390
We'll work in teams to solve a series of puzzles.

649
01:10:33.390 --> 01:10:38.039
So, that's it for our presentation. Thank you so much for your time today.

650
01:10:44.729 --> 01:10:51.449
Okay, I thought that was quite interesting.

651
01:10:51.449 --> 01:11:05.850
What I want to do is I'm going to show you the 1st, 5 minutes or whatever bit of Peter shore. So, if you talked to some of his previous talks were a little shallow. This will bring the average steps of today's class up a little.

652
01:11:05.850 --> 01:11:18.149
Got to have a little depth in things and so this is a new graphic.

653
01:11:18.149 --> 01:11:22.470
We've had quite a money protocols for about 10 years, but this is a.

654
01:11:22.470 --> 01:11:29.279
New kind of money protocol and the big, I guess, lie about the name's part of money.

655
01:11:29.279 --> 01:11:37.739
Is you couldn't actually use it for granted money and that's all the properties to be good for money. But since quantum states, only last for around.

656
01:11:37.739 --> 01:11:42.210
Maybe a minute or 2, you really don't want quite a money.

657
01:11:42.210 --> 01:11:46.979
Because you don't want your money to disappear after 2 or 3 minutes.

658
01:11:46.979 --> 01:11:51.750
But the outline of the talk, I'm going to talk about.

659
01:11:51.750 --> 01:11:55.140
The motivation for looking at quantum money and the history.

660
01:11:55.140 --> 01:12:00.390
And then the background, and then the work in progress, which isn't.

661
01:12:00.390 --> 01:12:08.520
Up yet, but it's, I'm writing so quite of money. Well, the big problem with money maybe is that you can make copies.

662
01:12:08.520 --> 01:12:12.359
So, if you have a physical piece of money.

663
01:12:12.359 --> 01:12:17.039
What you can do is you can be very clever in manufacturing it and put all sorts of.

664
01:12:17.039 --> 01:12:21.539
Things in it, which are hard to copy, but if then, of course, the counter.

665
01:12:21.539 --> 01:12:29.670
Get clever 2 and figure out how to copy it. So you have a race between the characters and the governments.

666
01:12:29.670 --> 01:12:35.279
So, if you actually had entirely digital money.

667
01:12:35.279 --> 01:12:42.239
It would be perfectly possible to make copy so that 1 now quantum states, there's something called a no cloning therapy.

668
01:12:42.239 --> 01:12:49.409
Which says, you cannot make a copy of an unknown quantum state. So it seems that that is would be perfect for money.

669
01:12:49.409 --> 01:13:02.909
So, this was his idea, and he came up with the idea in 969 and wrote the manuscript and it took 14 years to get it published. And that would not have happened without Charlie Bennett.

670
01:13:02.909 --> 01:13:10.229
You know, sending it to someone he knew who worked on this newsletter for.

671
01:13:10.229 --> 01:13:15.149
I guess ACM.

672
01:13:15.149 --> 01:13:19.890
Especially when the stroke, which isn't even a real Journal, it's a.

673
01:13:19.890 --> 01:13:28.199
You know, it's whatever whatever the editor wants to put in it gets put in. So any way we have.

674
01:13:28.199 --> 01:13:35.970
This amount of money now, it turns out that we sustain add some real drawbacks.

675
01:13:35.970 --> 01:13:39.510
Namely, if you wanted to verify a piece of amount of money.

676
01:13:39.510 --> 01:13:45.300
You had to send it back to some central authority who knew the secret that was used to make it.

677
01:13:45.300 --> 01:13:49.439
So, that was really.

678
01:13:49.439 --> 01:13:53.670
The problem, so, these quantum money schemes.

679
01:13:53.670 --> 01:14:02.250
I'll talk about you do not do not have this back. You can verify them just with a piece of the money.

680
01:14:02.250 --> 01:14:05.640
So, now I want to talk to them.

681
01:14:05.640 --> 01:14:16.380
Now, cryptography background for many years until the 19 seventies cryptography was done with ad hoc, crypto systems and many of these turned out to be eventually broken.

682
01:14:16.380 --> 01:14:21.479
So, over the last few decades, cryptography has become much more mathematical.

683
01:14:21.479 --> 01:14:26.340
And theoretical computer scientists try to prove the security of systems.

684
01:14:27.359 --> 01:14:30.390
There are 2 kinds of proofs of security and cryptography.

685
01:14:30.390 --> 01:14:37.020
There's informationally secure, crypto systems and computational secure systems.

686
01:14:37.020 --> 01:14:47.220
And information, security system system has the property that no matter how powerful the computer and adversary stuff has, they will not be able to break the system.

687
01:14:47.220 --> 01:14:58.140
Because they don't have access to the information. The disadvantage with secure systems is that there was only a very few cryptographic protocols.

688
01:14:58.140 --> 01:15:09.930
That can be made informationally secure, the computationally secure systems. The security system relies on the difficulty of solving some computational and heart problem.

689
01:15:09.930 --> 01:15:13.260
Like, maybe prime factorization.

690
01:15:13.260 --> 01:15:22.199
And the difficulty with computationally secure systems is that theoretical computer scientist don't know how to prove that any of these.

691
01:15:22.199 --> 01:15:25.260
Problems, so they're based on are really sick here.

692
01:15:25.260 --> 01:15:28.680
So, the best guarantee of security.

693
01:15:28.680 --> 01:15:32.880
As to find some problem that a lot of people have tried to crack.

694
01:15:32.880 --> 01:15:35.970
Have failed.

695
01:15:35.970 --> 01:15:41.789
So, quantum cryptography, the BB, any 4 protocol for quantum key distribution.

696
01:15:41.789 --> 01:15:45.750
Which was actually inspired by waist nurses, 1 of money.

697
01:15:45.750 --> 01:15:50.460
Can be proved informationally secure, assuming the laws. So part of the Catholics.

698
01:15:50.460 --> 01:15:54.659
So this solved the task, which was impossible to perform.

699
01:15:54.659 --> 01:16:03.390
With the classical computer, so 1 of the motivations to the point of money research was thinking about whether there are any.

700
01:16:03.390 --> 01:16:09.119
Group that the tasks that quantum computers might perform with computational security.

701
01:16:09.119 --> 01:16:12.869
But we're impossible for digital computer.

702
01:16:12.869 --> 01:16:18.180
And we think why the money is 1 of these and there have been a bunch of.

703
01:16:18.180 --> 01:16:22.739
Previous proposals for front of money, starting with ours.

704
01:16:22.739 --> 01:16:29.430
Which was based on data dairies, we call that computationally secure by the money needs some hard problem.

705
01:16:29.430 --> 01:16:32.489
So each of these has a different hard problem.

706
01:16:32.489 --> 01:16:37.680
Of course, the problem with thesis, and some of the hard problems turned out not to be hard.

707
01:16:37.680 --> 01:16:41.729
So, what this coin of money protocol does.

708
01:16:41.729 --> 01:16:46.770
Is it basis it's difficulty on problems.

709
01:16:46.770 --> 01:16:50.279
I have a lot of people think Labs. Cryptography is hard.

710
01:16:50.279 --> 01:16:54.659
Because post quantum cryptography, so.

711
01:16:54.659 --> 01:16:58.500
The public key assistance, which is quantum computer cannot be broken.

712
01:16:58.500 --> 01:17:03.420
1 of the major contenders for the replacement of is based on.

713
01:17:03.420 --> 01:17:09.720
Lattice systems, so people think they're harder.

714
01:17:09.720 --> 01:17:13.920
So, what is point of money? Well.

715
01:17:13.920 --> 01:17:20.189
We would like 1 of the players in the protocol, and we'll call her the men who are able to make a state.

716
01:17:20.189 --> 01:17:24.119
Which will call the quantum money state and a verification of protocol.

717
01:17:24.119 --> 01:17:32.640
And note that both of these are dependent on AI, which will call the serial numbers. So each quantum money state has a serial number.

718
01:17:32.640 --> 01:17:36.750
And you need to input the serial number into the verification protocol.

719
01:17:36.750 --> 01:17:40.050
To verify that quantum money state.

720
01:17:40.050 --> 01:17:45.329
So, the, a, the quantum money state of passed the test of verification.

721
01:17:45.329 --> 01:17:50.010
The test does not destroy the amount of money stake.

722
01:17:50.010 --> 01:17:56.579
And see, and inspiring counterfeit or close, both the quantum money stake and knows the protocol.

723
01:17:56.579 --> 01:18:02.189
Or verifying, it cannot produce a state of 2 quantum systems that both pass the test.

724
01:18:02.189 --> 01:18:06.689
Lisa bye. So that's what we would like.

725
01:18:06.689 --> 01:18:10.170
So, how does the amount of money protocol.

726
01:18:10.170 --> 01:18:17.880
Well, we'll find that and what we're going to do is we're 1st, going to give a little bit of background about Madison's.

727
01:18:17.880 --> 01:18:24.840
Then we're going to sketch our 1st candidate for quite amount of money and then we'll explain why it doesn't work.

728
01:18:24.840 --> 01:18:28.739
And very, very briefly say haven't been to.

729
01:18:29.760 --> 01:18:33.270
So, what is last last.

730
01:18:33.270 --> 01:18:36.449
It's a set of all your combinations of.

731
01:18:36.449 --> 01:18:39.539
And doctors and and you mentioned.

732
01:18:39.539 --> 01:18:48.779
So, it looks something like this and these and vectors can either be long like these blue vectors. Are they short?

733
01:18:48.779 --> 01:18:52.770
Like, these 2 red factors, and these 2 blue vectors them to read doctors.

734
01:18:52.770 --> 01:18:55.859
Are supposed to get the same.

735
01:18:55.859 --> 01:19:00.899
So this is a basis of long vectors. This is a basis of sharp factors.

736
01:19:00.899 --> 01:19:06.390
And the hard lattice problem is given a basis of long vectors for out.

737
01:19:06.390 --> 01:19:10.260
Find the basis of reasonably short thunders.

738
01:19:10.260 --> 01:19:16.079
And the best we know how to do is essentially the L cube.

739
01:19:16.079 --> 01:19:20.880
And what that does is it finds a basis exponentially longer.

740
01:19:20.880 --> 01:19:25.619
Were exponential and the dimension in the shortest possible places.

741
01:19:27.779 --> 01:19:31.590
So, we need to say a couple more things you can do with us.

742
01:19:31.590 --> 01:19:36.840
There's something called bounded distance decoding. So suppose you have a vector X?

743
01:19:36.840 --> 01:19:45.420
That is very close to 1 of our lattice doctors then we can find that lot of sector polynomial time. And what is very close to me.

744
01:19:45.420 --> 01:19:49.890
Well, it means it's exponentially closer to the shortest Specter in the Congress.

745
01:19:49.890 --> 01:19:55.560
There's also calcium something if you have a big enough bow around some doctor.

746
01:19:57.000 --> 01:20:02.699
Okay, that's a reasonable place to.

747
01:20:02.699 --> 01:20:07.109
The stop, I think, and.

748
01:20:07.109 --> 01:20:18.090
Just starting to get quite deep. So what I've tried to show you today was an introduction to an introduction to Google computing and.

749
01:20:18.090 --> 01:20:29.670
Hope you had some fun and I didn't get to the last long thing building google's quantum computer. So you can look at it yourself on your own or I may show just started it on Thursday.

750
01:20:29.670 --> 01:20:36.029
So the point about this part of the courses, I'm showing you other people, other companies besides IBM.

751
01:20:36.029 --> 01:20:44.220
And and there's a homework where you get a chance to look at amazon's bracket on your own. I may do a little bit in class also.

752
01:20:45.175 --> 01:20:58.194
And then, next topic, we're getting into things like trap Di and quantum computer, which is a different technology than the transmission cube. But IBM uses kept on quite a popular idea. So, Wikipedia articles not so bad.

753
01:20:58.194 --> 01:21:04.015
I'll get you some videos and some stuff on that. And then a company came out of Maryland, University of Maryland.

754
01:21:04.319 --> 01:21:09.090
Oh, yeah, I Q sorry which implement some of this.

755
01:21:09.625 --> 01:21:22.854
So, you can have some fun. We're moving on in the course the next big thing I have to type up is at the end, everyone, every group of people, There'll be a term project to do something big in quantum computing.

756
01:21:22.854 --> 01:21:30.145
You get to decide what and give a talk on it probably to talks over the last 3 classes. I'll write up something about that later.

757
01:21:30.420 --> 01:21:38.909
Okay, so hope you had some fun today and I'm preparing actually a summary on these different quantum computing.

758
01:21:38.909 --> 01:21:49.140
Sites which I'll be giving to the VP of research on Wednesday Robert hall runs a BI, weekly, quantum computing seminar, trying to get into it more. So.

759
01:21:49.140 --> 01:21:53.880
With my presentation available to the class after I do it.

760
01:21:53.880 --> 01:22:02.189
It's interesting. Okay, so I have a good week and I'll stay around here a little longer give people a chance to.

761
01:22:02.189 --> 01:22:06.510
Ask any questions other than that.

762
01:22:06.510 --> 01:22:11.130
Have fun.

763
01:22:13.289 --> 01:22:25.439
You're welcome.

764
01:22:25.439 --> 01:22:30.840
Silence.

765
01:22:56.609 --> 01:23:03.689
Silence.

766
01:23:17.699 --> 01:23:21.689
So, there's no questions then.