Mikhail Lukin and Dolev Bluvstein explain how they used trapped atoms to create 48 logical qubits

1
00:00:07,759 --> 00:00:10,800
Hello, and welcome to the Physics World Weekly

2
00:00:10,800 --> 00:00:12,820
podcast. I'm Hamish Johnston.

3
00:00:13,484 --> 00:00:16,125
This week, we're celebrating the winners of the

4
00:00:16,125 --> 00:00:17,105
Physics World

5
00:00:17,484 --> 00:00:20,385
breakthrough of the year award for 2024,

6
00:00:21,244 --> 00:00:22,945
and we have 2 podcasts

7
00:00:23,404 --> 00:00:25,024
for your listening pleasure.

8
00:00:25,910 --> 00:00:29,210
This episode is supported by the journal Reports

9
00:00:29,429 --> 00:00:31,289
on Progress in Physics,

10
00:00:31,829 --> 00:00:32,649
which offers

11
00:00:33,030 --> 00:00:33,530
unparalleled

12
00:00:34,070 --> 00:00:34,570
visibility

13
00:00:35,030 --> 00:00:36,250
for your groundbreaking

14
00:00:36,869 --> 00:00:37,369
research.

15
00:00:38,515 --> 00:00:41,814
This year's award is all about error correction

16
00:00:42,114 --> 00:00:43,575
in quantum computing,

17
00:00:44,034 --> 00:00:46,454
and we are honoring 2 independent

18
00:00:46,835 --> 00:00:47,335
teams.

19
00:00:48,114 --> 00:00:50,215
I've spoken to the lead researchers

20
00:00:50,730 --> 00:00:53,870
of both groups, and we're presenting those conversations

21
00:00:54,570 --> 00:00:56,750
in 2 different episodes.

22
00:00:57,450 --> 00:01:00,490
In this podcast, I'm in conversation with Mikhail

23
00:01:00,490 --> 00:01:02,750
Lukin and Dolev Blufstein

24
00:01:03,370 --> 00:01:04,670
at Harvard University,

25
00:01:05,245 --> 00:01:07,984
who, along with their colleagues, have implemented

26
00:01:08,525 --> 00:01:10,064
quantum error correction

27
00:01:10,444 --> 00:01:12,385
on an array of trapped

28
00:01:12,844 --> 00:01:14,144
atomic cubits.

29
00:01:14,924 --> 00:01:16,305
In a second podcast,

30
00:01:16,765 --> 00:01:18,144
I chat with Google's

31
00:01:18,444 --> 00:01:19,744
Hartmut Nevin,

32
00:01:20,420 --> 00:01:22,420
who leads a team that has made a

33
00:01:22,420 --> 00:01:23,479
major breakthrough

34
00:01:23,859 --> 00:01:24,759
in implementing

35
00:01:25,140 --> 00:01:26,680
quantum error correction

36
00:01:27,060 --> 00:01:28,119
in a processor

37
00:01:28,659 --> 00:01:29,479
that uses

38
00:01:29,780 --> 00:01:30,280
superconducting

39
00:01:30,739 --> 00:01:31,239
qubits.

40
00:01:32,259 --> 00:01:36,125
In principle, quantum computers can solve some problems

41
00:01:36,185 --> 00:01:37,564
that cannot be computed

42
00:01:37,864 --> 00:01:39,405
on conventional processors.

43
00:01:40,424 --> 00:01:42,524
However, the quantum processors

44
00:01:42,825 --> 00:01:43,884
available today

45
00:01:44,265 --> 00:01:45,405
are very susceptible

46
00:01:45,944 --> 00:01:47,724
to disruption by environmental

47
00:01:48,104 --> 00:01:48,604
noise,

48
00:01:49,010 --> 00:01:50,150
and this destroys

49
00:01:50,450 --> 00:01:53,810
the delicate quantum states that are used to

50
00:01:53,810 --> 00:01:55,270
store and process

51
00:01:55,810 --> 00:01:56,310
information.

52
00:01:57,010 --> 00:01:59,350
When quantum computing was first proposed,

53
00:01:59,730 --> 00:02:02,709
some physicists thought that this problem was insurmountable.

54
00:02:03,614 --> 00:02:06,655
But thanks to the development of quantum error

55
00:02:06,655 --> 00:02:07,155
correction,

56
00:02:07,775 --> 00:02:08,275
practical

57
00:02:08,655 --> 00:02:09,715
quantum computers

58
00:02:10,014 --> 00:02:12,115
that can solve useful problems

59
00:02:12,495 --> 00:02:14,354
could soon be a reality.

60
00:02:15,639 --> 00:02:17,819
This year, we awarded half

61
00:02:18,120 --> 00:02:19,180
of the 2024

62
00:02:19,959 --> 00:02:23,479
breakthrough of the year to Mikhail Lukin and

63
00:02:23,479 --> 00:02:23,979
Dolev

64
00:02:24,360 --> 00:02:24,860
Blufstein,

65
00:02:25,719 --> 00:02:27,900
and colleagues at Harvard University,

66
00:02:28,439 --> 00:02:29,180
the Massachusetts

67
00:02:29,560 --> 00:02:30,780
Institute of Technology,

68
00:02:31,775 --> 00:02:33,074
and CUERA Computing.

69
00:02:34,094 --> 00:02:35,474
And that's for demonstrating

70
00:02:35,935 --> 00:02:37,555
quantum error correction

71
00:02:37,935 --> 00:02:39,715
on an atomic processor

72
00:02:40,094 --> 00:02:40,915
with 48

73
00:02:41,455 --> 00:02:42,675
logical cubits.

74
00:02:43,455 --> 00:02:46,354
I'm very pleased to have Dolev and Mikhail

75
00:02:46,860 --> 00:02:49,280
on the line from Cambridge, Massachusetts.

76
00:02:50,459 --> 00:02:52,879
Welcome to the podcast, and congratulations

77
00:02:53,500 --> 00:02:54,239
on achieving

78
00:02:54,620 --> 00:02:55,360
a remarkable

79
00:02:55,739 --> 00:02:57,840
breakthrough in quantum computing.

80
00:02:58,459 --> 00:02:58,959
Hello.

81
00:02:59,944 --> 00:03:02,264
So, Mikhail, I think my first question is

82
00:03:02,264 --> 00:03:03,004
for you.

83
00:03:03,625 --> 00:03:05,884
What is quantum error correction,

84
00:03:06,344 --> 00:03:07,965
and why is it necessary?

85
00:03:09,784 --> 00:03:11,639
Thank you, for having us.

86
00:03:12,840 --> 00:03:13,340
So

87
00:03:14,039 --> 00:03:15,719
I have to answer this question, I will

88
00:03:15,719 --> 00:03:17,240
maybe go back a little bit to a

89
00:03:17,240 --> 00:03:20,300
history of, quantum computing and quantum information,

90
00:03:21,080 --> 00:03:21,580
which,

91
00:03:23,080 --> 00:03:25,159
kind of the early ideas are now by

92
00:03:25,159 --> 00:03:27,615
now, maybe 40 years old or so.

93
00:03:28,094 --> 00:03:30,754
And, when people started realizing that,

94
00:03:31,775 --> 00:03:34,514
you can use the ideas of quantum superposition

95
00:03:35,135 --> 00:03:37,555
and quantum entanglement to build new

96
00:03:38,175 --> 00:03:40,435
quantum information processing systems,

97
00:03:44,000 --> 00:03:46,180
From the very beginning, there was a concern

98
00:03:46,319 --> 00:03:48,879
whether you can actually realize such systems in

99
00:03:48,879 --> 00:03:49,539
in practice.

100
00:03:50,400 --> 00:03:50,900
And

101
00:03:51,360 --> 00:03:52,020
in particular,

102
00:03:52,319 --> 00:03:52,819
maybe

103
00:03:53,759 --> 00:03:55,939
around, like, 30 years ago where

104
00:03:56,634 --> 00:03:58,955
some of these early ideas started solid to

105
00:03:58,955 --> 00:04:01,614
solidify, for example, ideas of quantum simulations,

106
00:04:02,474 --> 00:04:05,215
the ideas of Shor's algorithms were put forward,

107
00:04:06,235 --> 00:04:09,040
people really started asking, you know, can quantum

108
00:04:09,180 --> 00:04:12,620
computers be built practical quantum computers be built

109
00:04:12,620 --> 00:04:13,840
be built and how?

110
00:04:14,300 --> 00:04:16,540
And already at that time, it was very

111
00:04:16,540 --> 00:04:18,800
clear. It will be a very challenging task.

112
00:04:19,180 --> 00:04:19,680
And,

113
00:04:21,420 --> 00:04:22,800
in particular, one

114
00:04:23,245 --> 00:04:25,665
challenge, I would say, is conceptual is that,

115
00:04:27,404 --> 00:04:28,464
it's very hard

116
00:04:29,004 --> 00:04:29,745
to put,

117
00:04:30,685 --> 00:04:33,485
a big system in a quantum superposition state.

118
00:04:33,485 --> 00:04:35,564
So at the kind of microscopic level, like

119
00:04:35,564 --> 00:04:36,464
single electrons,

120
00:04:37,000 --> 00:04:39,399
you know, single, you know, spins, you know,

121
00:04:39,800 --> 00:04:43,019
can be put routinely in a superposition state.

122
00:04:43,319 --> 00:04:45,720
But the big objects around us, you know,

123
00:04:45,720 --> 00:04:48,120
like, you know, this this span or this

124
00:04:48,120 --> 00:04:50,600
table, you know, while they're composed from quantum

125
00:04:50,600 --> 00:04:53,685
mechanical particles, you know, there is nothing quantum

126
00:04:53,685 --> 00:04:56,004
about them. Right? So big systems lose quantum

127
00:04:56,004 --> 00:04:58,824
character, and this is really fundamental. So

128
00:04:59,365 --> 00:05:00,185
in the,

129
00:05:01,444 --> 00:05:04,185
field of quantum computation, this specific example,

130
00:05:04,740 --> 00:05:05,240
this,

131
00:05:05,779 --> 00:05:07,779
issue shows up is that if you start

132
00:05:07,779 --> 00:05:10,180
building, for example, quantum computer and you build

133
00:05:10,180 --> 00:05:11,319
it from some kind of

134
00:05:11,699 --> 00:05:13,000
quantum logic operations,

135
00:05:13,540 --> 00:05:16,279
inevitably, there will be a small errors that,

136
00:05:17,060 --> 00:05:19,620
you know, will happen during this quantum logic

137
00:05:19,620 --> 00:05:20,120
operations.

138
00:05:20,634 --> 00:05:23,194
And these small errors eventually accumulate to make

139
00:05:23,194 --> 00:05:25,935
this, you know, the output completely classical.

140
00:05:26,314 --> 00:05:28,574
So it basically loses all quantum features.

141
00:05:29,194 --> 00:05:32,415
And, because of that, kind of early on,

142
00:05:32,634 --> 00:05:33,375
there were,

143
00:05:34,795 --> 00:05:35,339
kind of,

144
00:05:36,060 --> 00:05:37,899
there was a lot of excitement on one

145
00:05:37,899 --> 00:05:40,379
hand about quantum computers, but there was also

146
00:05:40,379 --> 00:05:41,199
a lot of skepticism.

147
00:05:41,500 --> 00:05:43,979
Right? And so and in particular, if you

148
00:05:43,979 --> 00:05:46,079
start looking at what type of error rates

149
00:05:46,459 --> 00:05:49,095
you need to really implement, you know,

150
00:05:49,894 --> 00:05:53,095
some, you know, interesting quantum algorithms at scale.

151
00:05:53,095 --> 00:05:55,894
These error rates are, you know, extremely low.

152
00:05:55,894 --> 00:05:56,634
You know?

153
00:05:57,014 --> 00:05:59,675
Well below 1 part per billion. You know?

154
00:05:59,975 --> 00:06:00,475
So

155
00:06:00,855 --> 00:06:01,355
and,

156
00:06:01,975 --> 00:06:02,954
for this reason,

157
00:06:03,574 --> 00:06:05,500
I mean, there was a little you know,

158
00:06:05,500 --> 00:06:08,379
some of the skeptics were actually very kind

159
00:06:08,379 --> 00:06:10,459
of prominent, you know, I would say, leaders

160
00:06:10,459 --> 00:06:12,139
on the field. We want you know, thought

161
00:06:12,139 --> 00:06:14,480
that, you know, while these quantum

162
00:06:14,860 --> 00:06:18,240
computers is really a furious dream, it's experimentally's

163
00:06:18,540 --> 00:06:20,785
nightmare. You know, it's really literally kind of

164
00:06:20,785 --> 00:06:23,425
impossible. It's impossible to reach such low error

165
00:06:23,425 --> 00:06:24,245
rates. And,

166
00:06:24,944 --> 00:06:27,665
and, already early on, there was an idea.

167
00:06:27,665 --> 00:06:28,964
Oh, okay. So in classical,

168
00:06:31,504 --> 00:06:35,430
computers or quantum classical information processing systems, you

169
00:06:35,430 --> 00:06:36,490
can often use

170
00:06:36,870 --> 00:06:37,529
some redundancy

171
00:06:38,069 --> 00:06:38,569
to,

172
00:06:39,430 --> 00:06:42,310
protect quantum information. And, you know, people started

173
00:06:42,310 --> 00:06:43,990
asking a question. So can you use this

174
00:06:43,990 --> 00:06:45,129
redundancy to

175
00:06:47,414 --> 00:06:49,194
basically preserve quantum information?

176
00:06:49,654 --> 00:06:52,154
And the answer, even conceptually at the time,

177
00:06:52,375 --> 00:06:53,115
was really,

178
00:06:54,214 --> 00:06:54,654
non,

179
00:06:55,094 --> 00:06:55,594
obvious,

180
00:06:56,854 --> 00:06:57,354
because,

181
00:07:00,569 --> 00:07:01,629
quantum information

182
00:07:02,009 --> 00:07:03,069
cannot be copied.

183
00:07:04,649 --> 00:07:07,289
Also, if you measure so to, like, utilize

184
00:07:07,289 --> 00:07:09,449
it done to see classically, you basically need

185
00:07:09,449 --> 00:07:09,949
to,

186
00:07:12,009 --> 00:07:13,149
measure state

187
00:07:13,644 --> 00:07:15,985
to verify where errors happen or not.

188
00:07:17,004 --> 00:07:19,104
In quantum mechanics, you basically cannot

189
00:07:19,404 --> 00:07:21,985
measure the state without collapsing this.

190
00:07:23,884 --> 00:07:25,805
And, you know, for this reason, it kind

191
00:07:25,805 --> 00:07:28,550
of seemed challenging that even in principle, you

192
00:07:28,550 --> 00:07:29,290
could use

193
00:07:29,750 --> 00:07:31,850
error correction to protect quantum information.

194
00:07:32,150 --> 00:07:32,650
Nevertheless,

195
00:07:34,710 --> 00:07:35,210
about,

196
00:07:35,910 --> 00:07:37,050
you know, 25,

197
00:07:38,550 --> 00:07:39,770
30 years ago,

198
00:07:42,514 --> 00:07:43,495
there was theoretically

199
00:07:43,954 --> 00:07:46,214
shown that you can actually use,

200
00:07:47,794 --> 00:07:50,134
this kind of redundancy. You can actually use

201
00:07:50,354 --> 00:07:50,854
entanglement

202
00:07:51,314 --> 00:07:51,814
to,

203
00:07:53,074 --> 00:07:54,935
store and protect quantum information.

204
00:07:55,629 --> 00:07:56,129
And

205
00:07:56,430 --> 00:07:58,449
I would say it was really this breakthrough

206
00:07:58,589 --> 00:07:59,569
that really

207
00:08:00,750 --> 00:08:02,050
kind of jump started

208
00:08:02,910 --> 00:08:05,330
this field as we know it now.

209
00:08:06,750 --> 00:08:08,370
I see. And, Dolev,

210
00:08:08,830 --> 00:08:10,564
an idea that's central

211
00:08:11,345 --> 00:08:12,884
to quantum error correction

212
00:08:13,904 --> 00:08:16,004
are the concepts of a physical

213
00:08:16,545 --> 00:08:17,444
and a logical

214
00:08:18,145 --> 00:08:18,645
cubit,

215
00:08:19,105 --> 00:08:21,504
and and that's something that was important in

216
00:08:21,504 --> 00:08:25,220
your research. Can you can you explain why

217
00:08:25,220 --> 00:08:26,680
quantum error correction

218
00:08:27,220 --> 00:08:27,720
considers

219
00:08:28,180 --> 00:08:28,680
physical

220
00:08:29,139 --> 00:08:31,699
and logical cubits? And and what are they?

221
00:08:31,699 --> 00:08:34,039
What what are the differences between the 2?

222
00:08:34,339 --> 00:08:37,000
Error correction is this really remarkable process,

223
00:08:39,065 --> 00:08:41,225
and, it uses physical cubits. And when we

224
00:08:41,225 --> 00:08:44,264
say physical cubits, we mean things like ions

225
00:08:44,264 --> 00:08:45,004
or superconductors

226
00:08:45,544 --> 00:08:46,044
or,

227
00:08:46,424 --> 00:08:48,904
you know, defects or spins, things that have

228
00:08:48,904 --> 00:08:49,500
two levels

229
00:08:50,299 --> 00:08:52,480
that are like a quantum two level system.

230
00:08:52,539 --> 00:08:54,220
And we have been studying these for roughly,

231
00:08:54,379 --> 00:08:56,879
like, almost a 100 years since the beginning

232
00:08:56,940 --> 00:08:57,679
of NMR,

233
00:08:58,299 --> 00:09:00,299
where people were using these for, you know,

234
00:09:00,299 --> 00:09:01,514
various different applications.

235
00:09:02,475 --> 00:09:02,715
And,

236
00:09:03,434 --> 00:09:05,754
the invention Misha mentioned from 30 years ago

237
00:09:05,754 --> 00:09:07,674
was the fact that you can actually put

238
00:09:07,674 --> 00:09:09,455
these together to make a logical qubit.

239
00:09:10,554 --> 00:09:11,054
And,

240
00:09:12,154 --> 00:09:13,915
the way that it works because you can't

241
00:09:13,915 --> 00:09:15,054
copy quantum information

242
00:09:15,870 --> 00:09:17,870
is what you do is you take this

243
00:09:17,870 --> 00:09:19,330
abstract unit of information,

244
00:09:19,710 --> 00:09:21,250
which is our logical qubit,

245
00:09:22,110 --> 00:09:23,649
and you use entanglement,

246
00:09:24,990 --> 00:09:27,570
these, you know, funny quantum correlations between particles

247
00:09:28,524 --> 00:09:31,165
to take this one logical cubit degree of

248
00:09:31,165 --> 00:09:34,225
freedom and delocalize it. You spread it across,

249
00:09:34,285 --> 00:09:36,384
you know, a large array of physical cubits.

250
00:09:37,165 --> 00:09:39,504
And now what happens is that this delocalized

251
00:09:39,644 --> 00:09:40,144
information

252
00:09:40,605 --> 00:09:41,345
is protected.

253
00:09:42,509 --> 00:09:45,169
Now, you know, if the environment comes in

254
00:09:45,629 --> 00:09:46,929
and tries to measure

255
00:09:47,470 --> 00:09:48,990
part of the system, if it looks at

256
00:09:48,990 --> 00:09:50,429
just one of the physical cubits in the

257
00:09:50,429 --> 00:09:52,990
system, it actually will not learn anything about

258
00:09:52,990 --> 00:09:55,409
the underlying stored delocalized state.

259
00:09:55,950 --> 00:09:57,009
And so

260
00:09:58,024 --> 00:10:01,144
that is what the physical mechanism is behind

261
00:10:01,144 --> 00:10:02,825
logical cubits, and that's how we take we

262
00:10:02,825 --> 00:10:05,404
use entanglement between physical cubits to make robust

263
00:10:05,785 --> 00:10:08,504
logical cubits, and it is truly remarkable that

264
00:10:08,504 --> 00:10:10,445
it is, you know, physically possible.

265
00:10:11,945 --> 00:10:14,639
That being said, it is clearly a completely

266
00:10:14,639 --> 00:10:15,539
different object

267
00:10:16,080 --> 00:10:18,799
than these physical cubits are. These physical cubits

268
00:10:18,799 --> 00:10:20,559
are these two level systems we've been working

269
00:10:20,559 --> 00:10:22,100
with for, you know,

270
00:10:22,399 --> 00:10:25,519
almost a century. These logical cubits are these,

271
00:10:25,519 --> 00:10:27,404
you know, highly entangled states,

272
00:10:28,605 --> 00:10:30,625
and is in in a sense an abstract

273
00:10:30,764 --> 00:10:33,004
unit of information. And now that we're starting

274
00:10:33,004 --> 00:10:34,845
to work with them in the lab, we

275
00:10:34,845 --> 00:10:36,524
see that there's many, many differences to the

276
00:10:36,524 --> 00:10:38,545
physical cubits that we're used to working with.

277
00:10:39,565 --> 00:10:40,065
And,

278
00:10:41,165 --> 00:10:43,049
perhaps one of the, you know, most

279
00:10:44,089 --> 00:10:46,089
key features there in terms of the difference

280
00:10:46,089 --> 00:10:48,350
between physical and logical cubits is that

281
00:10:48,730 --> 00:10:50,730
in a physical cubit, it's a quantum two

282
00:10:50,730 --> 00:10:51,470
level system.

283
00:10:52,009 --> 00:10:53,690
You can take this cubit and it can

284
00:10:53,690 --> 00:10:55,245
be in you know, you can just

285
00:10:55,804 --> 00:10:57,404
rotate it and put it in any state

286
00:10:57,404 --> 00:10:59,565
that you want. And that is, you know,

287
00:10:59,565 --> 00:11:01,245
for example, what people do when they do

288
00:11:01,245 --> 00:11:03,725
NMR is you have, you know, a spin

289
00:11:03,725 --> 00:11:05,565
and the spin can process, and it can

290
00:11:05,565 --> 00:11:07,725
be in any, you know, point of this,

291
00:11:07,725 --> 00:11:09,184
you know, qubit's

292
00:11:09,565 --> 00:11:10,384
phase space.

293
00:11:10,845 --> 00:11:13,250
But with logical cubits, it's very different. You

294
00:11:13,250 --> 00:11:15,350
can't do things like arbitrary

295
00:11:15,889 --> 00:11:17,830
rotations like we do in NMR.

296
00:11:18,210 --> 00:11:20,070
You can only do digital operations

297
00:11:20,929 --> 00:11:23,009
where the logical cubit and the still localized

298
00:11:23,009 --> 00:11:25,575
information can only do things that are exactly

299
00:11:25,575 --> 00:11:27,495
the operations that are allowed by this logical

300
00:11:27,495 --> 00:11:29,174
qubit. And in a sense, it starts to

301
00:11:29,174 --> 00:11:31,274
make quantum information processing more digital.

302
00:11:31,815 --> 00:11:34,375
And this also should parallel, you know, what

303
00:11:34,375 --> 00:11:36,154
we do with our classical computers,

304
00:11:36,870 --> 00:11:39,269
with classical computers, all of our information processing,

305
00:11:39,269 --> 00:11:40,950
the reason it's so robust is because it's

306
00:11:40,950 --> 00:11:43,830
digital. It works on bits, and the bits

307
00:11:43,830 --> 00:11:45,990
do, you know, precise logic operations that are

308
00:11:45,990 --> 00:11:48,090
insensitive to, for example, voltage fluctuations

309
00:11:48,629 --> 00:11:49,529
in your computer.

310
00:11:50,095 --> 00:11:53,294
And we have analog classical computers as well

311
00:11:53,294 --> 00:11:55,955
that are, you know, can do arbitrary rotations,

312
00:11:56,495 --> 00:11:59,134
and these things are much more similar to

313
00:11:59,134 --> 00:12:00,815
the physical cubits that we've been working with

314
00:12:00,815 --> 00:12:01,475
in the field,

315
00:12:02,174 --> 00:12:02,674
before.

316
00:12:03,230 --> 00:12:04,830
But now that we're working with logical cubits,

317
00:12:04,830 --> 00:12:07,470
we're really starting to explore digital processing, very

318
00:12:07,470 --> 00:12:08,990
similar to how we have digital processing with

319
00:12:08,990 --> 00:12:12,110
our classical computers. So there's various, you know,

320
00:12:12,110 --> 00:12:13,410
very important key differences,

321
00:12:13,870 --> 00:12:15,870
between these physical and logical cubits that we're

322
00:12:15,870 --> 00:12:17,170
starting to explore now.

323
00:12:17,495 --> 00:12:20,294
I see. And, Mikhail, can you can you

324
00:12:20,294 --> 00:12:20,794
explain

325
00:12:22,054 --> 00:12:24,714
or describe the the physical qubits

326
00:12:25,174 --> 00:12:27,414
that you use in your lab? Well, I

327
00:12:27,414 --> 00:12:30,294
suppose, in this specific bit of research that

328
00:12:30,294 --> 00:12:32,590
you and Dovlev have done. What what does

329
00:12:32,590 --> 00:12:34,129
the, what does the physical

330
00:12:34,670 --> 00:12:35,649
cubic comprise?

331
00:12:38,110 --> 00:12:39,009
These experiments

332
00:12:39,389 --> 00:12:39,889
utilize,

333
00:12:42,509 --> 00:12:43,009
isolated,

334
00:12:43,550 --> 00:12:46,690
trapped, neutral atoms, actually, rubidium atoms,

335
00:12:47,634 --> 00:12:49,654
which are individually trapped

336
00:12:50,034 --> 00:12:50,615
and held,

337
00:12:51,634 --> 00:12:54,514
in optical tweezers and tightly focused beams of

338
00:12:54,514 --> 00:12:55,495
beams of light.

339
00:12:56,274 --> 00:12:56,774
So

340
00:12:57,315 --> 00:12:58,995
to kind of take a step back, you

341
00:12:58,995 --> 00:13:00,674
know, this, of course, one of the many

342
00:13:00,674 --> 00:13:02,759
platforms that people are now exploring.

343
00:13:03,299 --> 00:13:04,600
So the reason why,

344
00:13:05,779 --> 00:13:08,679
neutral atoms is a promising platform is because,

345
00:13:10,419 --> 00:13:11,559
first of all, they

346
00:13:11,940 --> 00:13:15,684
have excellent coherence properties. So while isolated and

347
00:13:15,684 --> 00:13:16,664
held in tweezers,

348
00:13:17,205 --> 00:13:20,004
you can basically store quantum information for very

349
00:13:20,004 --> 00:13:22,725
long time for tens of seconds or kind

350
00:13:22,725 --> 00:13:24,825
of minutes. It's kind of almost like unlimited,

351
00:13:25,044 --> 00:13:25,625
you know,

352
00:13:26,125 --> 00:13:27,384
in increase in principle.

353
00:13:28,830 --> 00:13:30,529
And that's one important,

354
00:13:31,149 --> 00:13:32,370
feature. For example,

355
00:13:33,710 --> 00:13:34,210
atomic

356
00:13:34,910 --> 00:13:36,769
clocks, you know, some of the

357
00:13:37,149 --> 00:13:37,970
most precise

358
00:13:38,509 --> 00:13:39,649
precise instruments,

359
00:13:41,105 --> 00:13:43,825
scientific instruments that humankind ever built, you know,

360
00:13:43,825 --> 00:13:44,884
utilize now

361
00:13:45,345 --> 00:13:47,605
neutral atoms, trapped neutral atoms.

362
00:13:47,985 --> 00:13:50,304
So and the second thing which is also

363
00:13:50,304 --> 00:13:51,684
very important is that

364
00:13:52,449 --> 00:13:54,870
in print in principle and also in practice,

365
00:13:54,929 --> 00:13:55,909
you know, you can,

366
00:13:57,570 --> 00:13:58,070
create

367
00:13:58,449 --> 00:13:59,669
a very large,

368
00:14:01,730 --> 00:14:03,909
number of this kind of, you know,

369
00:14:04,610 --> 00:14:05,350
of the,

370
00:14:05,970 --> 00:14:07,454
atoms. And, basically,

371
00:14:08,554 --> 00:14:10,815
I should also point out that our experiments

372
00:14:11,834 --> 00:14:13,995
make use of this so called laser cooling

373
00:14:13,995 --> 00:14:16,414
and trapping techniques. So they're actually room temperature

374
00:14:16,634 --> 00:14:20,075
systems, but their atomic motion is slowed down

375
00:14:20,075 --> 00:14:21,929
by essentially buffing the atoms

376
00:14:22,250 --> 00:14:23,389
in a beam of,

377
00:14:24,490 --> 00:14:26,169
beams of light of certain color. And as

378
00:14:26,169 --> 00:14:28,589
a result, they basically come to the standstill.

379
00:14:29,129 --> 00:14:31,769
So each of our experiments starts with a

380
00:14:31,769 --> 00:14:35,450
cloud which contains, you know, basically many millions

381
00:14:35,450 --> 00:14:36,990
of tens of millions of

382
00:14:37,304 --> 00:14:39,945
motionless atoms, which are, of course, fantastic, you

383
00:14:39,945 --> 00:14:41,004
know, initial,

384
00:14:43,384 --> 00:14:44,845
step, you know, to create,

385
00:14:45,625 --> 00:14:46,605
a lot of qubits.

386
00:14:48,745 --> 00:14:49,804
But the challenge

387
00:14:50,179 --> 00:14:52,339
now is not just to create these qubits,

388
00:14:52,339 --> 00:14:54,179
but also to control them. And that's why

389
00:14:54,179 --> 00:14:56,659
we use this kind of optical techniques, the

390
00:14:56,659 --> 00:14:57,799
techniques from holography,

391
00:14:58,740 --> 00:15:00,440
to basically create a large,

392
00:15:03,220 --> 00:15:04,039
number of

393
00:15:05,235 --> 00:15:08,195
tweezers and trap the atoms there. There is

394
00:15:08,195 --> 00:15:09,335
one other challenge,

395
00:15:10,035 --> 00:15:12,035
and that is the atoms in the gas

396
00:15:12,035 --> 00:15:12,535
phase

397
00:15:13,075 --> 00:15:15,715
basically don't interact with each other, don't talk

398
00:15:15,715 --> 00:15:17,394
to each other. So in order to entangle

399
00:15:17,394 --> 00:15:19,254
them, to make quantum logic,

400
00:15:19,649 --> 00:15:21,350
we actually use lasers

401
00:15:21,809 --> 00:15:23,970
to promote the atoms into the so called

402
00:15:23,970 --> 00:15:25,889
Rydberg states. So the Rydberg states are the

403
00:15:25,889 --> 00:15:29,649
states where, electrons orbit, you know, very far

404
00:15:29,649 --> 00:15:32,404
away from the nuclear. So so the atoms,

405
00:15:33,985 --> 00:15:36,464
in this state, say, basically, can be thought

406
00:15:36,464 --> 00:15:38,404
of having kind of a large size.

407
00:15:38,784 --> 00:15:40,384
And then as a result of that, they

408
00:15:40,384 --> 00:15:42,865
really start interacting very strongly to each other.

409
00:15:42,865 --> 00:15:43,365
And

410
00:15:43,759 --> 00:15:46,100
in particular, to entangle these atoms,

411
00:15:47,120 --> 00:15:47,620
and,

412
00:15:48,159 --> 00:15:50,480
do quantum logic, we utilize something which is

413
00:15:50,480 --> 00:15:52,720
called Rydberg blockade. So that's the idea, which

414
00:15:52,720 --> 00:15:53,220
is,

415
00:15:53,600 --> 00:15:55,600
by now almost 25 years old.

416
00:15:56,345 --> 00:15:58,184
But it kind of, in the recent years,

417
00:15:58,184 --> 00:16:00,845
really proved to be kind of remarkably fruitful.

418
00:16:01,465 --> 00:16:01,865
And,

419
00:16:02,745 --> 00:16:04,904
and the the key idea of this Lydberg

420
00:16:04,904 --> 00:16:06,125
blockade is that

421
00:16:06,904 --> 00:16:07,725
you can basically

422
00:16:08,105 --> 00:16:09,404
consider 2 atoms,

423
00:16:10,299 --> 00:16:12,139
which you excite to the Rydberg states. If

424
00:16:12,139 --> 00:16:14,779
they sit far away, then, essentially, you can

425
00:16:14,779 --> 00:16:17,360
excite them independently. But when you bring them

426
00:16:17,419 --> 00:16:19,579
close to each other, then what happens is

427
00:16:19,579 --> 00:16:21,179
if one of the atoms is excited to

428
00:16:21,179 --> 00:16:23,500
the Rydberg state, the excitation for the second

429
00:16:23,500 --> 00:16:24,315
atom is blocked.

430
00:16:25,274 --> 00:16:28,394
And what this mechanism does, it basically makes

431
00:16:28,394 --> 00:16:30,894
the interaction of the atoms between the atoms

432
00:16:31,195 --> 00:16:33,995
almost digital. So if they're far away, they

433
00:16:33,995 --> 00:16:35,835
don't talk to each other. If they're close

434
00:16:35,835 --> 00:16:37,855
to each other, the interaction is nearly infinity.

435
00:16:38,750 --> 00:16:40,690
And that actually allows us

436
00:16:41,629 --> 00:16:42,110
to,

437
00:16:43,629 --> 00:16:44,129
entangle,

438
00:16:44,990 --> 00:16:45,809
the atoms,

439
00:16:47,950 --> 00:16:49,250
with very low errors.

440
00:16:49,710 --> 00:16:53,085
And in particular, it also enables us to

441
00:16:53,085 --> 00:16:56,445
entangle many pairs of atoms in parallel. So

442
00:16:56,445 --> 00:16:58,125
that is something that we will maybe talk

443
00:16:58,125 --> 00:16:59,985
about a little bit later. So,

444
00:17:00,365 --> 00:17:04,204
basically, these optical tools combined with this Littburg

445
00:17:04,204 --> 00:17:04,704
blockade

446
00:17:05,759 --> 00:17:08,420
is what enables, you know, high fidelity

447
00:17:09,440 --> 00:17:11,779
or in other words, low error rate, you

448
00:17:11,920 --> 00:17:12,660
know, parallel,

449
00:17:13,279 --> 00:17:14,500
control and entanglement

450
00:17:15,279 --> 00:17:17,279
of 100 of atoms at a time. You

451
00:17:17,279 --> 00:17:19,234
know? And this is an essential

452
00:17:19,855 --> 00:17:22,494
ingredient which actually allows us to do this

453
00:17:22,494 --> 00:17:24,894
kind of logical you know, build these logical

454
00:17:24,894 --> 00:17:26,914
processors kind of in a very efficient,

455
00:17:28,095 --> 00:17:28,595
way.

456
00:17:29,695 --> 00:17:31,295
And can I just ask you,

457
00:17:31,695 --> 00:17:32,195
Mikael,

458
00:17:33,214 --> 00:17:35,630
how how big is this

459
00:17:36,009 --> 00:17:36,509
cloud

460
00:17:36,809 --> 00:17:38,809
of atoms? Just to give our listeners an

461
00:17:38,809 --> 00:17:40,350
idea of the size of

462
00:17:40,730 --> 00:17:42,890
of, I mean, I know you've probably got

463
00:17:42,890 --> 00:17:44,670
a huge vacuum chamber, etcetera,

464
00:17:45,049 --> 00:17:46,430
but the actual cloud.

465
00:17:46,765 --> 00:17:48,845
The actual you know, the size of the

466
00:17:48,845 --> 00:17:49,345
processor,

467
00:17:50,444 --> 00:17:52,765
you know, which traps these atoms, you know,

468
00:17:52,765 --> 00:17:55,424
and basically allows us to kind of manipulate,

469
00:17:57,804 --> 00:17:58,605
the this

470
00:17:59,240 --> 00:18:01,980
this atoms is a couple of 100 microns.

471
00:18:02,200 --> 00:18:04,619
You know, it's actually, you know, relatively small.

472
00:18:06,039 --> 00:18:08,299
But even this even this,

473
00:18:08,840 --> 00:18:12,279
zone is actually kind of divided into small

474
00:18:12,279 --> 00:18:12,734
parts,

475
00:18:13,134 --> 00:18:15,134
smaller parts, and one of them is like

476
00:18:15,134 --> 00:18:18,015
a storage zone. Another one is the zone

477
00:18:18,015 --> 00:18:20,015
where we do logic and entangling zone, and

478
00:18:20,015 --> 00:18:21,775
another one is on the zone where we

479
00:18:21,775 --> 00:18:22,994
do kind of readout.

480
00:18:23,454 --> 00:18:24,914
And, actually, one other

481
00:18:26,609 --> 00:18:29,410
innovation now going to 2020, which actually really

482
00:18:29,410 --> 00:18:30,789
fueled with this development,

483
00:18:31,170 --> 00:18:32,769
was actually led by the left,

484
00:18:35,410 --> 00:18:36,230
was to

485
00:18:38,875 --> 00:18:39,375
realize

486
00:18:39,755 --> 00:18:40,634
what we call,

487
00:18:42,154 --> 00:18:42,894
the configurable,

488
00:18:43,835 --> 00:18:44,335
architecture.

489
00:18:44,634 --> 00:18:46,894
So and to explain it, you know,

490
00:18:47,515 --> 00:18:49,454
like, let's think about how

491
00:18:50,075 --> 00:18:50,815
the conventional,

492
00:18:51,434 --> 00:18:54,670
you know, chips, you know, semiconductor chips. So,

493
00:18:55,210 --> 00:18:56,970
you know, you what you do, you know,

494
00:18:56,970 --> 00:18:58,730
you design this chip, you design this, you

495
00:18:58,730 --> 00:19:01,150
know, transistors, you design that, you know, connectivity,

496
00:19:01,849 --> 00:19:04,250
and then you basically, you know, send it

497
00:19:04,250 --> 00:19:04,750
to,

498
00:19:07,404 --> 00:19:10,065
you know, to to the, you know, factory,

499
00:19:10,125 --> 00:19:11,184
you know, where basically,

500
00:19:12,284 --> 00:19:14,684
the state of the art technique uses optical

501
00:19:14,684 --> 00:19:17,884
lithography. Basically uses optical tools to define where

502
00:19:17,884 --> 00:19:18,284
this chip

503
00:19:19,004 --> 00:19:20,670
these transistors are going to be.

504
00:19:21,150 --> 00:19:22,609
And then eventually, this,

505
00:19:23,710 --> 00:19:24,210
this,

506
00:19:26,549 --> 00:19:27,230
you know,

507
00:19:27,710 --> 00:19:28,210
things,

508
00:19:29,710 --> 00:19:31,950
are made. But, basically, the point what I

509
00:19:31,950 --> 00:19:33,090
want to make is that

510
00:19:34,274 --> 00:19:34,934
the connectivity,

511
00:19:35,634 --> 00:19:36,294
the architecture

512
00:19:36,754 --> 00:19:37,734
of this chip

513
00:19:38,434 --> 00:19:40,454
is fixed at a design stage.

514
00:19:40,914 --> 00:19:43,335
So what happens is that using optical tweezers,

515
00:19:44,274 --> 00:19:46,454
we can actually move atoms around,

516
00:19:47,349 --> 00:19:50,490
and we move them while preserving the coherence,

517
00:19:50,630 --> 00:19:53,029
while preserving the stored qubit. So this is

518
00:19:53,029 --> 00:19:55,929
done by encoding qubits into so called hyperfine

519
00:19:55,990 --> 00:19:58,470
states, basically spin states of atoms where they

520
00:19:58,470 --> 00:20:00,704
can live for a very long time. And

521
00:20:00,944 --> 00:20:01,684
most importantly,

522
00:20:01,984 --> 00:20:03,984
you know, by moving the atoms around, you

523
00:20:03,984 --> 00:20:04,804
can basically

524
00:20:05,265 --> 00:20:06,404
create the,

525
00:20:08,224 --> 00:20:08,964
the architecture

526
00:20:09,664 --> 00:20:10,565
where the connectivity

527
00:20:10,865 --> 00:20:13,265
is like a living organism. It changes during

528
00:20:13,265 --> 00:20:15,670
the computation itself. And it, for example, allows

529
00:20:15,670 --> 00:20:18,410
us to move atoms between these different zones.

530
00:20:19,190 --> 00:20:20,470
It allows us to,

531
00:20:21,910 --> 00:20:24,470
you know, entangle atoms in parallel. You know?

532
00:20:24,470 --> 00:20:27,029
And it allows us to basically implement all

533
00:20:27,029 --> 00:20:28,009
necessary ingredients,

534
00:20:28,955 --> 00:20:30,174
for the logical cubits.

535
00:20:31,035 --> 00:20:33,695
So, Dov, in in your work, you created

536
00:20:33,914 --> 00:20:38,234
48 logical cubits using these physical cubits that

537
00:20:38,234 --> 00:20:39,214
Mikhail has,

538
00:20:39,914 --> 00:20:40,414
described.

539
00:20:41,035 --> 00:20:41,535
How

540
00:20:42,160 --> 00:20:44,000
how did you do this? I mean, is

541
00:20:44,000 --> 00:20:46,960
it possible to to describe it in simple

542
00:20:46,960 --> 00:20:50,420
terms? How how you take these physical cubits

543
00:20:50,480 --> 00:20:50,980
and

544
00:20:51,359 --> 00:20:54,019
sort of blend them together to make logical

545
00:20:54,079 --> 00:20:54,404
cubits?

546
00:20:55,845 --> 00:20:58,325
Yes. Absolutely. So maybe there's 2 two stages

547
00:20:58,325 --> 00:21:00,484
of answering your question. So one is, like,

548
00:21:00,484 --> 00:21:02,505
on a physical level, how does this even

549
00:21:02,644 --> 00:21:03,144
happen,

550
00:21:03,445 --> 00:21:05,285
like, on a quantum mechanical level? And then

551
00:21:05,285 --> 00:21:06,904
the other one is, how did we

552
00:21:07,285 --> 00:21:09,380
make how are we able to, you know,

553
00:21:09,380 --> 00:21:11,140
really simplify the problem to make it much

554
00:21:11,140 --> 00:21:13,059
easier than it has been historically in the

555
00:21:13,059 --> 00:21:13,559
field?

556
00:21:14,019 --> 00:21:14,500
And,

557
00:21:14,980 --> 00:21:16,920
the, to answer the first one,

558
00:21:17,460 --> 00:21:20,420
it's we're we're leveraging the fact that we

559
00:21:20,420 --> 00:21:21,559
can entangle particles

560
00:21:22,144 --> 00:21:24,704
by moving them around entangle these atomic cubits

561
00:21:24,704 --> 00:21:26,464
by moving them around and zapping them with

562
00:21:26,464 --> 00:21:27,825
the laser pulses when they're next to each

563
00:21:27,825 --> 00:21:29,044
other to entangle them.

564
00:21:29,585 --> 00:21:30,065
And,

565
00:21:30,384 --> 00:21:31,444
to create these,

566
00:21:32,544 --> 00:21:33,599
logical cubit states,

567
00:21:34,160 --> 00:21:36,160
similar to what I was describing earlier, we

568
00:21:36,160 --> 00:21:38,720
take this, you know, like, one qubit, for

569
00:21:38,720 --> 00:21:39,220
example,

570
00:21:39,759 --> 00:21:41,599
or a collection of qubits, and then we

571
00:21:41,599 --> 00:21:43,059
entangle it with its surrounding

572
00:21:43,519 --> 00:21:44,339
atomic qubits

573
00:21:44,720 --> 00:21:46,259
in a very structured way

574
00:21:46,640 --> 00:21:48,494
that spreads out this information.

575
00:21:49,194 --> 00:21:51,835
That creates this logical cubit once you create

576
00:21:51,835 --> 00:21:53,994
this entangled state. Now you have to do

577
00:21:53,994 --> 00:21:55,835
multiple important things to it that we were

578
00:21:55,835 --> 00:21:57,914
able to explore in our work. One is

579
00:21:57,914 --> 00:22:00,554
that you, you know, can now do logic

580
00:22:00,554 --> 00:22:01,054
operations

581
00:22:01,679 --> 00:22:03,139
between these logical cubits.

582
00:22:03,919 --> 00:22:04,419
And,

583
00:22:05,279 --> 00:22:06,880
that's actually one of the things that's the

584
00:22:06,880 --> 00:22:07,940
hardest to do.

585
00:22:08,480 --> 00:22:09,460
When we take,

586
00:22:10,159 --> 00:22:12,240
you know, this logical cubit and then spread

587
00:22:12,240 --> 00:22:14,835
it out across an array of physical cubits,

588
00:22:15,875 --> 00:22:17,474
And we might do this now on 2

589
00:22:17,474 --> 00:22:19,234
different blocks of qubits, and we'll have 2

590
00:22:19,234 --> 00:22:20,534
different, you know,

591
00:22:20,914 --> 00:22:21,414
delocalized,

592
00:22:23,234 --> 00:22:24,214
degrees of freedom.

593
00:22:24,595 --> 00:22:26,914
They're now protected from their environment because now

594
00:22:26,914 --> 00:22:29,019
the environment can't come in and measure this

595
00:22:29,019 --> 00:22:31,259
underlying state. But now it's also very hard

596
00:22:31,259 --> 00:22:32,240
to get them to interact.

597
00:22:32,779 --> 00:22:35,259
And actually in the field before, people had

598
00:22:35,259 --> 00:22:38,140
done quite, you know, nice work in creating

599
00:22:38,140 --> 00:22:39,039
logical cubits.

600
00:22:39,980 --> 00:22:42,220
But the really huge challenge was always getting

601
00:22:42,220 --> 00:22:43,904
them to interact Because now there are these,

602
00:22:43,904 --> 00:22:46,164
you know, just completely delocalized degrees of freedom.

603
00:22:46,384 --> 00:22:48,384
And for example, imagine that you have these

604
00:22:48,384 --> 00:22:50,544
2 delocalized degrees of freedom, and they're stuck

605
00:22:50,544 --> 00:22:52,464
next to each other on a chip or

606
00:22:52,464 --> 00:22:54,484
something with a fixed 2 d connectivity.

607
00:22:54,865 --> 00:22:56,224
But now it's very hard to get them

608
00:22:56,224 --> 00:22:57,904
to interact because they can only interact through

609
00:22:57,904 --> 00:23:00,700
some boundary, whereas they're, like, delocalized over space.

610
00:23:01,240 --> 00:23:03,880
So with our ability to move cubits around,

611
00:23:03,880 --> 00:23:05,240
what we can now do is we can

612
00:23:05,240 --> 00:23:06,299
actually pick up

613
00:23:06,680 --> 00:23:08,860
the 2 logical degrees of freedom,

614
00:23:09,240 --> 00:23:10,920
put them right on top of each other

615
00:23:10,920 --> 00:23:11,660
by interlacing

616
00:23:12,279 --> 00:23:14,234
the 2 grids of atomic cubits,

617
00:23:14,795 --> 00:23:16,894
And then by entangling all of the pairs

618
00:23:17,434 --> 00:23:19,755
of the, you know, underlying logical blocks, that

619
00:23:19,755 --> 00:23:22,335
realizes, like, a logical entangling operation.

620
00:23:23,434 --> 00:23:23,934
So

621
00:23:24,234 --> 00:23:26,234
that is how we do both the creation

622
00:23:26,234 --> 00:23:28,894
of the logical cubits as well as their

623
00:23:29,115 --> 00:23:29,615
operations.

624
00:23:30,720 --> 00:23:31,200
And,

625
00:23:31,680 --> 00:23:34,400
that is something that it was an extreme

626
00:23:34,400 --> 00:23:34,900
simplification.

627
00:23:35,680 --> 00:23:37,759
So one is it now by doing these

628
00:23:37,759 --> 00:23:39,619
gates in this way that we call transversal,

629
00:23:40,320 --> 00:23:41,759
where we can take these 2 degrees of

630
00:23:41,759 --> 00:23:44,025
freedom and interact them directly. It's an

631
00:23:44,744 --> 00:23:47,305
native logical operation that can just be directly

632
00:23:47,305 --> 00:23:47,805
done.

633
00:23:48,505 --> 00:23:50,345
The other thing that's really special that is

634
00:23:50,345 --> 00:23:52,184
related to what Misha said is it now

635
00:23:52,184 --> 00:23:54,345
allows us to start controlling things in much

636
00:23:54,345 --> 00:23:55,244
simpler ways.

637
00:23:55,865 --> 00:23:56,684
And in particular,

638
00:23:57,144 --> 00:24:00,679
you know, modern quantum processors have almost exclusively

639
00:24:00,740 --> 00:24:02,579
been built in this way where you have

640
00:24:02,579 --> 00:24:04,259
several cubits and you just add more and

641
00:24:04,259 --> 00:24:05,700
more cubits and you add more and more

642
00:24:05,700 --> 00:24:08,339
controls to control each cubit. But one of

643
00:24:08,339 --> 00:24:10,579
the big innovations in this work that made

644
00:24:10,579 --> 00:24:11,880
this so much simpler

645
00:24:12,259 --> 00:24:14,419
is that once we're starting to do error

646
00:24:14,419 --> 00:24:14,919
correction,

647
00:24:15,595 --> 00:24:17,595
all of the physical cubits within a logical

648
00:24:17,595 --> 00:24:18,095
cubit

649
00:24:18,714 --> 00:24:20,474
just need to do the exact same operation

650
00:24:20,474 --> 00:24:22,255
in order to realize a logical operation.

651
00:24:22,794 --> 00:24:24,894
So now in this zoned architecture

652
00:24:25,274 --> 00:24:27,515
that we're describing, we can also work with

653
00:24:27,515 --> 00:24:29,434
all these logical cubit blocks as if there's

654
00:24:29,434 --> 00:24:31,339
just one big atom essentially.

655
00:24:31,640 --> 00:24:33,480
And we take this one, you know, big

656
00:24:33,480 --> 00:24:35,000
atom and put it next to one other

657
00:24:35,000 --> 00:24:36,940
big atom and do this entangling

658
00:24:37,240 --> 00:24:39,880
logical operation in a single parallel step, and

659
00:24:39,880 --> 00:24:41,900
then can go and move these, you

660
00:24:42,279 --> 00:24:44,279
know, big qubits and interact them with other

661
00:24:44,279 --> 00:24:46,184
big qubits. And that was one of the

662
00:24:46,184 --> 00:24:47,644
things that was really central

663
00:24:47,944 --> 00:24:49,304
for us to be able to create such

664
00:24:49,304 --> 00:24:51,944
a large number of logical cubits and explore

665
00:24:51,944 --> 00:24:53,565
different types of interesting algorithms,

666
00:24:54,424 --> 00:24:56,505
with them. We were able to create 48

667
00:24:56,505 --> 00:24:59,619
of these small logical cubits using these approaches

668
00:24:59,619 --> 00:25:02,099
and do 100 of logical operations, whereas in

669
00:25:02,099 --> 00:25:04,259
the field, people had previously only done 1

670
00:25:04,259 --> 00:25:06,740
or 2. Before this work, we were also

671
00:25:06,740 --> 00:25:09,700
able to study things such as improving logic

672
00:25:09,700 --> 00:25:11,079
operations as we increase

673
00:25:11,380 --> 00:25:13,664
the size of the error correcting code and

674
00:25:13,664 --> 00:25:15,345
study a lot of really key features of

675
00:25:15,345 --> 00:25:18,085
what does error corrected quantum computation look like

676
00:25:18,144 --> 00:25:19,904
due to the fact that we can, you

677
00:25:19,904 --> 00:25:21,045
know, do this abstracted,

678
00:25:21,985 --> 00:25:24,144
control where we're working with logical cubits as

679
00:25:24,144 --> 00:25:25,680
the fundamental units of this processor.

680
00:25:26,240 --> 00:25:28,720
I see. And and, Mikhail, you you've got

681
00:25:28,720 --> 00:25:29,539
these 48

682
00:25:30,160 --> 00:25:33,360
logical cubits. Are you able to to actually

683
00:25:33,360 --> 00:25:34,900
do practical calculations

684
00:25:35,920 --> 00:25:38,740
with your system? Is there are there problems,

685
00:25:39,039 --> 00:25:40,500
you know, maybe even trivial

686
00:25:41,194 --> 00:25:43,994
computational problems that you can solve using it?

687
00:25:43,994 --> 00:25:45,835
Or is it very much a sort of

688
00:25:45,835 --> 00:25:46,815
proof of principle

689
00:25:47,755 --> 00:25:48,255
system?

690
00:25:49,194 --> 00:25:51,434
So maybe we'll answer this question in 2

691
00:25:51,434 --> 00:25:52,954
parts. I will start, and then I'll let

692
00:25:52,954 --> 00:25:55,159
Alef complete my answer. So,

693
00:25:55,700 --> 00:25:57,640
and, you know, to answer it, I maybe

694
00:25:57,779 --> 00:25:59,799
want to make a step back

695
00:26:00,500 --> 00:26:01,559
and, you know,

696
00:26:02,659 --> 00:26:04,119
mention that in addition

697
00:26:04,819 --> 00:26:07,315
to kind of building, you know, large scale

698
00:26:07,315 --> 00:26:10,054
quantum computer, another big challenge in the field

699
00:26:10,355 --> 00:26:12,934
is to identify what can we use these,

700
00:26:13,474 --> 00:26:15,315
you know, devices for. You know? How can

701
00:26:15,315 --> 00:26:16,615
they really help humankind?

702
00:26:18,125 --> 00:26:20,039
And you could say, well, I mean, it's

703
00:26:20,039 --> 00:26:21,579
kind of a, you know,

704
00:26:22,440 --> 00:26:24,519
funny question to ask for this field. So

705
00:26:24,519 --> 00:26:26,279
active and so on, but this is not

706
00:26:26,279 --> 00:26:28,919
unusual. So when I renew some new tool

707
00:26:28,919 --> 00:26:31,240
comes, you know, into play, you know, people

708
00:26:31,240 --> 00:26:34,365
often, you know, have hard time anticipating where

709
00:26:34,365 --> 00:26:36,625
it's, you know, going to be most,

710
00:26:37,085 --> 00:26:37,585
useful.

711
00:26:38,365 --> 00:26:40,125
But there is one area where it is

712
00:26:40,125 --> 00:26:42,305
very clear that this quantum computers,

713
00:26:43,644 --> 00:26:44,785
and quantum simulators,

714
00:26:46,045 --> 00:26:47,579
will have tremendous value,

715
00:26:47,899 --> 00:26:50,559
And that is in modeling and simulating,

716
00:26:53,099 --> 00:26:55,759
systems which have high degree of entanglement.

717
00:26:57,099 --> 00:26:59,039
And these type of,

718
00:27:00,140 --> 00:27:01,279
you know, systems

719
00:27:01,740 --> 00:27:04,914
occur in various areas of science, of physics

720
00:27:04,914 --> 00:27:05,575
in particular.

721
00:27:06,515 --> 00:27:07,975
Certainly, you know,

722
00:27:08,914 --> 00:27:11,255
many of the condensed matter models,

723
00:27:13,715 --> 00:27:16,434
feature, you know, so called strongly correlated systems.

724
00:27:16,434 --> 00:27:16,920
You know?

725
00:27:17,559 --> 00:27:18,220
You know,

726
00:27:18,759 --> 00:27:21,080
feature, you know, high degree of an or

727
00:27:21,080 --> 00:27:23,240
expect I expect it to feature feature high

728
00:27:23,240 --> 00:27:24,220
degree of entanglement.

729
00:27:25,160 --> 00:27:27,980
Another area which is actually also very exciting,

730
00:27:29,137 --> 00:27:29,355
is,

731
00:27:30,714 --> 00:27:31,214
involves,

732
00:27:31,835 --> 00:27:34,554
simulating system where you can build entanglement very

733
00:27:34,554 --> 00:27:35,054
quickly.

734
00:27:35,674 --> 00:27:36,174
And,

735
00:27:37,115 --> 00:27:38,095
this is actually,

736
00:27:39,194 --> 00:27:41,755
very interestingly connected to the physics of black

737
00:27:41,755 --> 00:27:44,670
holes. You know? So people believe that black

738
00:27:44,670 --> 00:27:46,190
holes, at least, you know, a kind of

739
00:27:46,589 --> 00:27:48,049
their quantum description, you know,

740
00:27:49,069 --> 00:27:49,569
involves

741
00:27:50,029 --> 00:27:52,910
this process, involve, you know, fast scrambling. You

742
00:27:52,910 --> 00:27:55,390
know? This is where you basically, you know,

743
00:27:55,390 --> 00:27:57,789
create entanglement kind of in the fastest way

744
00:27:57,789 --> 00:27:58,289
possible.

745
00:27:58,694 --> 00:28:01,494
And so one of the experiments we have

746
00:28:01,494 --> 00:28:04,214
done is actually exploring this kind of fast

747
00:28:04,214 --> 00:28:04,714
scrambling.

748
00:28:05,575 --> 00:28:07,734
And maybe I'll let Daleyf add because, you

749
00:28:07,734 --> 00:28:09,815
know, he's he was a mastermind of this,

750
00:28:09,815 --> 00:28:11,880
you know, specific. You know?

751
00:28:12,599 --> 00:28:15,480
Yeah. So following up on that. So maybe

752
00:28:15,480 --> 00:28:17,319
I will also take a step back about

753
00:28:17,319 --> 00:28:20,380
this quantum scrambling and say that one of

754
00:28:21,079 --> 00:28:23,720
the biggest open challenges in physics is we

755
00:28:23,720 --> 00:28:24,619
do not understand

756
00:28:25,000 --> 00:28:27,259
how quantum mechanics and gravity combine.

757
00:28:27,855 --> 00:28:29,855
It is, in my view as a physicist,

758
00:28:29,855 --> 00:28:31,934
one of the most interesting open questions of

759
00:28:31,934 --> 00:28:32,595
our time.

760
00:28:33,134 --> 00:28:35,234
And that is actually one of the places

761
00:28:35,375 --> 00:28:36,595
where quantum computers

762
00:28:37,054 --> 00:28:38,914
can almost certainly be very useful.

763
00:28:39,535 --> 00:28:40,035
And

764
00:28:40,539 --> 00:28:43,339
remarkably, we don't fully understand how quantum mechanics

765
00:28:43,339 --> 00:28:45,900
and gravity combine, but one of our best

766
00:28:45,900 --> 00:28:47,980
guesses in terms of how this might arise

767
00:28:47,980 --> 00:28:48,799
in our universe

768
00:28:49,180 --> 00:28:50,320
is that there's entanglement

769
00:28:51,180 --> 00:28:52,079
on some boundary

770
00:28:52,460 --> 00:28:53,355
in our universe,

771
00:28:53,755 --> 00:28:56,315
and the entanglement on this boundary gives an

772
00:28:56,315 --> 00:28:57,375
emergent gravitational

773
00:28:58,634 --> 00:29:01,054
description of the universe, which is amazing.

774
00:29:01,835 --> 00:29:02,335
And,

775
00:29:03,035 --> 00:29:06,075
however, it's really hard to make progress on

776
00:29:06,075 --> 00:29:07,994
some of these types of really complex quantum

777
00:29:07,994 --> 00:29:08,490
questions

778
00:29:08,970 --> 00:29:11,690
without a quantum calculator. We only have classical

779
00:29:11,690 --> 00:29:14,670
calculators, and we're trying to calculate these extremely

780
00:29:14,730 --> 00:29:15,230
complex

781
00:29:16,009 --> 00:29:17,309
things about the universe.

782
00:29:17,769 --> 00:29:19,929
So what we did here is a really,

783
00:29:19,929 --> 00:29:21,069
you know, like, toy

784
00:29:21,904 --> 00:29:23,585
study of those types of things, but we

785
00:29:23,585 --> 00:29:25,585
were able to, you know, study this complex

786
00:29:25,585 --> 00:29:26,085
scrambling.

787
00:29:26,625 --> 00:29:28,964
In particular, we entangle everything on hypercubes.

788
00:29:29,505 --> 00:29:31,365
One of the things that's special about hypercubes

789
00:29:31,424 --> 00:29:33,585
is they're very, very connected and that scrambles

790
00:29:33,585 --> 00:29:35,744
information very rapidly, very similar to a black

791
00:29:35,744 --> 00:29:36,244
hole.

792
00:29:36,779 --> 00:29:39,500
And, we didn't learn anything here about, you

793
00:29:39,500 --> 00:29:40,960
know, emergence of gravity

794
00:29:41,340 --> 00:29:44,059
from complex entangled systems, but it does start

795
00:29:44,059 --> 00:29:45,259
to give us a bit of a hint

796
00:29:45,259 --> 00:29:47,680
in terms of, you know, how we can,

797
00:29:47,980 --> 00:29:50,299
you know, what types of systems can we

798
00:29:50,299 --> 00:29:52,005
simulate with these logical qubits.

799
00:29:52,964 --> 00:29:53,285
And,

800
00:29:55,684 --> 00:29:57,125
if I can just nerd out for a

801
00:29:57,125 --> 00:29:59,045
second, one of the things that we did,

802
00:29:59,045 --> 00:30:01,144
which was really, you know,

803
00:30:02,085 --> 00:30:05,069
very tailored to this logical cubit processor here,

804
00:30:05,549 --> 00:30:07,549
is that once we're so we've we've had,

805
00:30:07,549 --> 00:30:09,329
you know, several decades of exploring,

806
00:30:09,869 --> 00:30:11,730
you know, processing with physical qubits.

807
00:30:12,589 --> 00:30:13,089
And

808
00:30:13,630 --> 00:30:16,190
physical qubits have very particular rules that we're

809
00:30:16,190 --> 00:30:17,089
used to following.

810
00:30:17,714 --> 00:30:19,394
And one of them is that, you know,

811
00:30:19,394 --> 00:30:21,954
it's very easy to do arbitrary rotations of

812
00:30:21,954 --> 00:30:23,794
cubits like we do in NMR, and it's

813
00:30:23,794 --> 00:30:25,015
very hard to do entanglement.

814
00:30:25,714 --> 00:30:27,794
In these, you know, first error corrected algorithms

815
00:30:27,794 --> 00:30:29,759
that we were doing with these logical cubits,

816
00:30:30,240 --> 00:30:31,599
one of the things that we saw is

817
00:30:31,599 --> 00:30:33,700
it's very hard to do arbitrary rotation,

818
00:30:34,400 --> 00:30:36,480
but it's very easy to create entanglement. So

819
00:30:36,480 --> 00:30:38,259
it actually very well suited to something

820
00:30:38,720 --> 00:30:40,579
like this black hole scrambling.

821
00:30:41,119 --> 00:30:43,119
And this was just one example of us

822
00:30:43,119 --> 00:30:45,214
doing quantum simulation with these air corrected cubits,

823
00:30:45,214 --> 00:30:47,714
but it starts to open a new scientific

824
00:30:47,775 --> 00:30:48,275
frontier

825
00:30:48,654 --> 00:30:51,375
of exploring how to do quantum simulations and

826
00:30:51,375 --> 00:30:52,195
quantum computations,

827
00:30:53,055 --> 00:30:55,455
in ways that are highly tailored to these,

828
00:30:55,455 --> 00:30:58,255
you know, new set of weird rules that

829
00:30:58,255 --> 00:31:00,180
we have to work with with logical cubits.

830
00:31:00,500 --> 00:31:02,820
And so we're not yet doing any practical

831
00:31:02,820 --> 00:31:06,100
calculations that are, you know, curing cancer or

832
00:31:06,100 --> 00:31:08,180
completely changing things like that. But what it

833
00:31:08,180 --> 00:31:09,940
is very clear is that in the near

834
00:31:09,940 --> 00:31:10,440
term

835
00:31:10,820 --> 00:31:12,660
and in all already what we've done and

836
00:31:12,660 --> 00:31:13,605
also in the near term,

837
00:31:14,404 --> 00:31:16,164
there will be a real value in just

838
00:31:16,164 --> 00:31:18,884
learning what these quantum computers can do from

839
00:31:18,884 --> 00:31:20,964
these types of experiments. So it's practical in

840
00:31:20,964 --> 00:31:22,744
that sense, but, yeah.

841
00:31:23,444 --> 00:31:24,884
Oh, that's great. I mean, I have to

842
00:31:24,884 --> 00:31:26,964
say, I wasn't expecting that you'd say black

843
00:31:26,964 --> 00:31:27,464
holes.

844
00:31:28,589 --> 00:31:30,990
So that's that's great. You learn learn something

845
00:31:30,990 --> 00:31:33,390
new every day. So what's what's next,

846
00:31:33,869 --> 00:31:34,849
for you guys?

847
00:31:35,390 --> 00:31:37,309
Mikhail, what what what do you have planned

848
00:31:37,309 --> 00:31:39,470
for the future? Are you are you going

849
00:31:39,470 --> 00:31:40,769
to try to create

850
00:31:41,150 --> 00:31:41,650
more

851
00:31:42,194 --> 00:31:44,755
logical cubits in your system? Or is there

852
00:31:44,755 --> 00:31:47,414
some other avenue that you can pursue to,

853
00:31:47,714 --> 00:31:49,815
sort of, to to gain your understanding

854
00:31:50,115 --> 00:31:52,694
of, of this quantum computer system?

855
00:31:53,794 --> 00:31:54,534
Yes. Certainly

856
00:31:55,714 --> 00:31:56,615
scaling up

857
00:31:57,059 --> 00:31:57,559
this,

858
00:31:58,420 --> 00:32:01,620
you know, quantum, you know, computation is definitely

859
00:32:01,620 --> 00:32:03,000
very much in our agenda.

860
00:32:03,539 --> 00:32:04,039
And,

861
00:32:05,059 --> 00:32:07,380
I would say that, you know, definitely one

862
00:32:07,380 --> 00:32:09,620
would like to have more logical qubits. You

863
00:32:09,620 --> 00:32:12,764
know? But most more importantly or equally important,

864
00:32:12,904 --> 00:32:15,944
one would like to actually improve these logical

865
00:32:15,944 --> 00:32:17,865
qubits. Right? Because, you know, even if we

866
00:32:17,865 --> 00:32:18,764
encode information,

867
00:32:19,944 --> 00:32:21,704
you know, at least up to now, what

868
00:32:21,704 --> 00:32:24,184
we and others have done is encoding offers

869
00:32:24,184 --> 00:32:26,609
some protection, but it's not, you know, I

870
00:32:26,609 --> 00:32:28,130
mean, it's not perfect. So we would like

871
00:32:28,130 --> 00:32:28,789
to actually,

872
00:32:29,250 --> 00:32:31,670
you know, make this logical qubit better,

873
00:32:32,369 --> 00:32:34,769
and then reduce error rates. And the key

874
00:32:34,769 --> 00:32:37,029
goal here is really to start,

875
00:32:37,730 --> 00:32:39,345
doing computation, which have

876
00:32:39,904 --> 00:32:42,085
computations, which have deeper circuits.

877
00:32:42,464 --> 00:32:42,964
So,

878
00:32:43,585 --> 00:32:44,484
I would say,

879
00:32:44,944 --> 00:32:46,944
you know, over last year, you know, in

880
00:32:46,944 --> 00:32:48,704
addition to the work that we have done,

881
00:32:48,704 --> 00:32:51,684
there was also some very nice experiments from

882
00:32:51,744 --> 00:32:54,565
across several different platforms. So for example,

883
00:32:55,190 --> 00:32:57,289
the very recent work of by Google,

884
00:32:57,990 --> 00:32:58,490
actually,

885
00:32:59,269 --> 00:33:02,069
demonstrated just one logical qubit, but what they

886
00:33:02,069 --> 00:33:03,769
have done is they basically,

887
00:33:05,109 --> 00:33:08,069
did experiments which involve multiple cycles of error

888
00:33:08,069 --> 00:33:10,744
correction. Alright? And it's kind of it's a

889
00:33:10,744 --> 00:33:12,585
little bit like scaling up if you want

890
00:33:12,585 --> 00:33:14,825
them on different axis as compared to what

891
00:33:14,825 --> 00:33:17,565
we have done. So but, but in reality,

892
00:33:17,785 --> 00:33:19,464
what one needs to do is one needs

893
00:33:19,464 --> 00:33:21,545
to really combine these two things, you know,

894
00:33:21,545 --> 00:33:23,085
to basically start implementing

895
00:33:23,670 --> 00:33:26,390
kind of, you know, deep circuits involving large

896
00:33:26,390 --> 00:33:29,750
number of of logical qubits and eventually try

897
00:33:29,750 --> 00:33:32,309
to, you know, figure out, you know, how

898
00:33:32,309 --> 00:33:33,930
to answer, you know,

899
00:33:34,710 --> 00:33:36,984
your previous question in different ways. You know?

900
00:33:36,984 --> 00:33:38,664
So the question is, what can we do

901
00:33:38,664 --> 00:33:40,524
with this kind of systems, basically?

902
00:33:41,144 --> 00:33:43,724
And, what emerged from our work

903
00:33:44,105 --> 00:33:44,845
is that

904
00:33:45,144 --> 00:33:48,105
answering this question would really real has to

905
00:33:48,105 --> 00:33:50,265
rely on this idea, which we sometimes call

906
00:33:50,265 --> 00:33:52,160
codesign. So if you have if you want

907
00:33:52,160 --> 00:33:54,880
to solve some specific problem. So what you

908
00:33:54,880 --> 00:33:55,700
like to do

909
00:33:56,000 --> 00:33:57,700
or you would need to do, basically,

910
00:33:58,240 --> 00:33:59,859
you need to think about algorithm

911
00:34:00,319 --> 00:34:01,539
to solve this problem,

912
00:34:02,000 --> 00:34:04,019
kind of codesigned with first

913
00:34:04,345 --> 00:34:06,825
error correcting code, which really fits this problem

914
00:34:06,825 --> 00:34:07,485
very well

915
00:34:07,865 --> 00:34:09,405
together with decoder,

916
00:34:09,785 --> 00:34:12,925
with compiler, and eventually with your hardware system.

917
00:34:12,985 --> 00:34:15,385
So this is it's very clear to us

918
00:34:15,385 --> 00:34:17,065
that for the next, you know, 5 years,

919
00:34:17,065 --> 00:34:19,144
maybe in the next decade, this is a

920
00:34:19,144 --> 00:34:20,099
way to make progress.

921
00:34:20,660 --> 00:34:22,739
And we are really excited about kind of

922
00:34:22,739 --> 00:34:25,160
starting to put these things together to really

923
00:34:25,380 --> 00:34:27,720
kind of, you know, you know, build,

924
00:34:28,180 --> 00:34:28,840
you know,

925
00:34:30,260 --> 00:34:31,720
systems and and

926
00:34:32,114 --> 00:34:34,195
come up with more examples where you can

927
00:34:34,195 --> 00:34:36,914
really, you know, enable, like, deep circuit kind

928
00:34:36,914 --> 00:34:37,574
of useful,

929
00:34:38,195 --> 00:34:40,594
quantum computation with, you know, large number of

930
00:34:40,594 --> 00:34:42,534
of qubits. Maybe I'll let Dolesv.

931
00:34:42,914 --> 00:34:44,514
Do you have anything to add to that,

932
00:34:44,514 --> 00:34:45,014
Dolesv?

933
00:34:45,650 --> 00:34:47,489
Yeah. Absolutely. So, yeah, I would say that

934
00:34:47,489 --> 00:34:49,170
last year, we learned a lot about how

935
00:34:49,170 --> 00:34:51,489
to do error corrected algorithms. Now, you know,

936
00:34:51,489 --> 00:34:52,550
it's a really important

937
00:34:53,010 --> 00:34:55,170
frontier to learn how to do deeper error

938
00:34:55,170 --> 00:34:56,070
corrected algorithms,

939
00:34:56,450 --> 00:34:57,829
and improve the performance.

940
00:34:59,144 --> 00:35:01,704
There is really exciting progress happening across the

941
00:35:01,704 --> 00:35:03,625
field, both, you know, between, you know, neutral

942
00:35:03,625 --> 00:35:05,305
atoms, trapped ions, who are gonna think you,

943
00:35:05,305 --> 00:35:07,144
but people are really starting to experiment with

944
00:35:07,144 --> 00:35:09,065
these systems. And I think one of the

945
00:35:09,065 --> 00:35:09,885
things that

946
00:35:10,210 --> 00:35:11,030
has become

947
00:35:11,410 --> 00:35:12,949
extremely clear in 2024

948
00:35:13,489 --> 00:35:15,269
is that error correction is,

949
00:35:15,730 --> 00:35:17,190
you know, definitely works.

950
00:35:17,969 --> 00:35:21,170
And also, it is currently a real inflection

951
00:35:21,170 --> 00:35:21,670
point

952
00:35:22,130 --> 00:35:24,164
in the sense of you really start to

953
00:35:24,164 --> 00:35:26,485
get below characteristic thresholds in the system. You

954
00:35:26,485 --> 00:35:28,805
really start to come up with creative ways

955
00:35:28,805 --> 00:35:30,344
to do logic operations.

956
00:35:30,724 --> 00:35:32,644
And it really the field is now going

957
00:35:32,644 --> 00:35:36,184
to start transitioning toward doing algorithms and computations

958
00:35:36,325 --> 00:35:38,609
and simulations with error correction. And that is

959
00:35:38,609 --> 00:35:40,130
going to be a, I think, a very

960
00:35:40,130 --> 00:35:41,029
dramatic inflection.

961
00:35:41,730 --> 00:35:42,630
And so

962
00:35:43,250 --> 00:35:45,329
that, in my view, is extremely exciting. I

963
00:35:45,329 --> 00:35:46,609
mean, even just in the past few months

964
00:35:46,609 --> 00:35:48,609
alone across many different systems, there's been really

965
00:35:48,609 --> 00:35:50,230
remarkable error correction progress.

966
00:35:50,609 --> 00:35:53,375
But there is, however, one pretty huge elephant

967
00:35:53,375 --> 00:35:55,234
in the room, which is that

968
00:35:55,695 --> 00:35:57,454
for a lot of the computations that we

969
00:35:57,454 --> 00:35:58,974
have in mind, we need things at the

970
00:35:58,974 --> 00:36:00,994
scale of tens of millions of cubits.

971
00:36:01,534 --> 00:36:03,875
And we are working to reduce that number,

972
00:36:04,494 --> 00:36:06,414
but the main thing is that we are

973
00:36:06,414 --> 00:36:07,795
not yet close to that.

974
00:36:08,469 --> 00:36:10,230
We do have ideas in terms of how

975
00:36:10,230 --> 00:36:10,969
to get there.

976
00:36:11,510 --> 00:36:13,190
But currently we're working with systems that have

977
00:36:13,190 --> 00:36:14,890
hundreds of cubits at the most.

978
00:36:15,750 --> 00:36:16,650
And so,

979
00:36:18,150 --> 00:36:20,550
there will be many challenges in trying to

980
00:36:20,550 --> 00:36:22,090
get to these much larger systems.

981
00:36:22,875 --> 00:36:24,155
But I do think that the field is

982
00:36:24,155 --> 00:36:26,175
going to develop in a very different way

983
00:36:26,235 --> 00:36:27,695
than it has in the past.

984
00:36:28,074 --> 00:36:29,594
And I would say it's for 2 key

985
00:36:29,594 --> 00:36:31,215
reasons that are emerging now.

986
00:36:31,515 --> 00:36:33,215
One is that error correction

987
00:36:33,594 --> 00:36:35,819
clearly works, and I think that is really

988
00:36:35,819 --> 00:36:37,420
starting to be at an inflection point that

989
00:36:37,420 --> 00:36:39,579
it was not nearly at the same level

990
00:36:39,579 --> 00:36:41,739
2 years ago. And 2 is that when

991
00:36:41,739 --> 00:36:43,579
we build error corrected processors, we can build

992
00:36:43,579 --> 00:36:45,420
them differently than we're used to building physical

993
00:36:45,420 --> 00:36:46,239
CUDA processors.

994
00:36:46,699 --> 00:36:48,779
And so we are not yet close to

995
00:36:48,779 --> 00:36:49,920
our end goal.

996
00:36:50,514 --> 00:36:52,375
Although, of course, the goal will always evolve.

997
00:36:52,835 --> 00:36:54,434
Although we can, you know, start to explore

998
00:36:54,434 --> 00:36:56,994
interesting science in the meantime. But I also

999
00:36:56,994 --> 00:36:59,014
think that things are going to start developing

1000
00:36:59,394 --> 00:37:01,894
across all these very various different systems

1001
00:37:02,355 --> 00:37:04,675
more rapidly than is being expected because of

1002
00:37:04,675 --> 00:37:06,719
these two key changes that I think will

1003
00:37:06,719 --> 00:37:08,019
both be key inflections.

1004
00:37:09,199 --> 00:37:11,940
So, yeah, many challenges, but also very exciting.

1005
00:37:12,800 --> 00:37:15,519
Oh, well, that's great. Well, thanks. Thanks, for

1006
00:37:15,519 --> 00:37:16,659
coming on the podcast.

1007
00:37:22,894 --> 00:37:25,375
This is one of 2 podcasts with our

1008
00:37:25,375 --> 00:37:27,394
breakthrough of the year winners.

1009
00:37:28,095 --> 00:37:31,474
The other features Google's Hartmut Kniven,

1010
00:37:31,775 --> 00:37:33,695
and you can find it on the Physics

1011
00:37:33,695 --> 00:37:34,595
World website

1012
00:37:35,030 --> 00:37:37,929
or at your favorite podcast provider.

1013
00:37:38,630 --> 00:37:41,349
You can also read more about our top

1014
00:37:41,349 --> 00:37:43,289
10 breakthroughs of 2024

1015
00:37:44,230 --> 00:37:45,769
on the Physics World website.

1016
00:37:46,630 --> 00:37:49,449
This served as the shortlist for our breakthrough

1017
00:37:49,670 --> 00:37:51,994
of the year, and it covers a range

1018
00:37:51,994 --> 00:37:54,575
of fantastic research in physics.

1019
00:37:55,114 --> 00:37:56,655
So do check it out.

1020
00:37:58,155 --> 00:37:59,994
I'm afraid that's all the time we have

1021
00:37:59,994 --> 00:38:01,215
for this week's podcast.

1022
00:38:01,595 --> 00:38:04,894
Thanks to Mikhail Lukin and Dolev Blufstein

1023
00:38:05,500 --> 00:38:06,800
for joining me today,

1024
00:38:07,099 --> 00:38:09,739
and a special thanks to our producer Fred

1025
00:38:09,739 --> 00:38:10,239
Iles.

1026
00:38:11,019 --> 00:38:12,320
Physics World's coverage

1027
00:38:12,700 --> 00:38:15,280
of the breakthrough of the year is supported

1028
00:38:15,340 --> 00:38:17,985
by Reports on Progress in Physics,

1029
00:38:18,785 --> 00:38:19,525
which offers

1030
00:38:19,825 --> 00:38:20,325
unparalleled

1031
00:38:20,945 --> 00:38:21,445
visibility

1032
00:38:21,905 --> 00:38:23,045
for your groundbreaking

1033
00:38:23,505 --> 00:38:24,005
research.

1034
00:38:24,545 --> 00:38:26,885
You can find the journal at iopscience.i0p.org.

1035
00:38:30,545 --> 00:38:32,485
We'll be back again next week.