WEBVTT

1
00:00:57.750 --> 00:01:01.170
Silence.

2
00:01:02.369 --> 00:01:07.560
Okay, good afternoon class. This is.

3
00:01:07.560 --> 00:01:19.260
Quantum computing class, 26, Thursday, December 3rd, 2020 and this will be the 1st day of student presentations. Now.

4
00:01:20.790 --> 00:01:24.659
Still trying to get my video will.

5
00:01:25.739 --> 00:01:33.299
But, okay, wow. Things finally start working sort of.

6
00:01:33.299 --> 00:01:38.760
So, a chance for.

7
00:01:38.760 --> 00:01:46.950
Need to shut up and you, you have to talk a find more updated list here. Let me start.

8
00:01:46.950 --> 00:01:50.250
Tearing things.

9
00:01:50.250 --> 00:01:54.120
And.

10
00:01:55.650 --> 00:01:58.829
Cool.

11
00:01:58.829 --> 00:02:08.550
Okay, well, 1 thing 1st, before you start talking to people have tried to send me or give me links to.

12
00:02:08.550 --> 00:02:15.900
Videos to download and in both cases I've downloaded the videos and then when I tried to play them.

13
00:02:15.900 --> 00:02:23.129
It said, I was not permitted to play that. Now this is something I've not seen before. So.

14
00:02:23.129 --> 00:02:31.139
It's so the solution is, you put your own machine into full screen motor if you.

15
00:02:31.139 --> 00:02:43.469
You know, I don't know how you're generating your videos, but there's some sort of security on them, which is new to me. I can play my own videos so I can pay videos. I download from.

16
00:02:43.469 --> 00:02:47.879
Webex and so on so I don't know what it is. It's something.

17
00:02:47.879 --> 00:02:54.090
Google, for example, who knows what's happening in any case.

18
00:02:54.090 --> 00:03:04.770
What are the ones that I was and so today we've got some poor talks some to group talk the YouTube 1. I can play. So, Justin, I can play your video.

19
00:03:04.770 --> 00:03:09.210
You speak with sharing stock part of you all, but that doesn't matter.

20
00:03:09.210 --> 00:03:12.750
Question is.

21
00:03:12.750 --> 00:03:17.550
David, can you go.

22
00:03:17.550 --> 00:03:24.599
And well, okay, we'll tell you later today, David then see if you can upload yours to YouTube and then you send me.

23
00:03:24.599 --> 00:03:28.530
Um, you know.

24
00:03:28.530 --> 00:03:32.129
I'll see what I can do. It's tricky. Do I.

25
00:03:32.129 --> 00:03:39.509
To install it, it'll take a while for me to get a URL from. However, you send it to me. I don't know if you put it in chat or something.

26
00:03:39.509 --> 00:03:45.150
Okay, but 1st Connor Conner and Sid.

27
00:03:45.150 --> 00:03:54.689
Would you like to go 1st are you there? Yeah, I'm not sitting with her. I thought I was doing my own thing.

28
00:03:54.689 --> 00:04:04.319
Oh, this is news to me yes. Yeah. You got the name, right? It's color. Em, that. Okay, but you're giving 2 separate talks.

29
00:04:04.319 --> 00:04:08.460
No, no, we're working together just you got the you got the last names wrong.

30
00:04:08.460 --> 00:04:12.419
On his own and Connor Evans are together.

31
00:04:12.419 --> 00:04:19.889
Oh, so you're corner? M. yeah. Okay. Let me make a note of this. Um, okay.

32
00:04:21.300 --> 00:04:28.649
Got M. G. okay. So Monday the 7th that will be Connor g.

33
00:04:29.668 --> 00:04:32.728
If you're there Yep. Yep.

34
00:04:33.809 --> 00:04:36.959
Thank you sorry about that. Okay.

35
00:04:36.959 --> 00:04:42.959
Corner and said floor is open for you if you'd like to.

36
00:04:42.959 --> 00:04:47.278
Share your screen, I'll stop and Don caught me.

37
00:04:47.278 --> 00:04:50.309
And you're on.

38
00:04:51.658 --> 00:04:57.418
All right can you see my screen? You have your full screen I'll mute myself. I see you full screen now.

39
00:04:57.418 --> 00:05:11.189
Perfect all right so today we're going to be talking about, uh, the content for transform and some error mitigation, uh, techniques that are available to use, uh, on real quantum computers.

40
00:05:11.189 --> 00:05:16.139
So, we're going over what the is and how it's used.

41
00:05:16.139 --> 00:05:22.259
Um, an overview of the tests we've run, uh, which I've been on both IBM and Amazon bracket.

42
00:05:22.259 --> 00:05:29.098
Very good to have our implementation for each of these, uh, systems, and then go over the test results.

43
00:05:29.098 --> 00:05:33.928
Um, from each of them, and then some further experiments and work that could happen.

44
00:05:33.928 --> 00:05:40.528
Uh, to, uh, as an analysis of, uh, the error suites observed and how they can be.

45
00:05:40.528 --> 00:05:44.459
Improved so, 1st, we'll be going over the content for your transform.

46
00:05:44.459 --> 00:05:50.608
So, declining for our transform is a quantum version. Obviously the discrete for your transform.

47
00:05:50.608 --> 00:05:55.829
So, people may be familiar religious from signals or a digital signal processing class.

48
00:05:55.829 --> 00:06:02.278
But because it Transformers a discreet signal from the time domain to the frequency domain. But these are both a.

49
00:06:02.278 --> 00:06:07.348
Discreet signals, there's no continuous nature. Like, did these discreet time for it transformed.

50
00:06:07.348 --> 00:06:15.509
Um, so naturally it can be expressed as a matrix multiplication and the end of being complex rotations. We can see that in the matrix in the bottom, right?

51
00:06:15.509 --> 00:06:23.189
And when you think of complex rotations, you think of, if you think of Cubans, um, that's basically having to keep it right to.

52
00:06:23.189 --> 00:06:26.968
State is getting rotated, um, by the different Gates so.

53
00:06:26.968 --> 00:06:31.288
It could be natural to think that a quantum computers could be.

54
00:06:31.288 --> 00:06:34.379
Used to implement this transformation.

55
00:06:34.379 --> 00:06:47.548
And this is called the clinical transform, so we can see on the top, right? The equation for the corner for transform and it has pretty much the exact same structure as the discreet for transform but instead uses.

56
00:06:47.548 --> 00:06:53.369
Cubics, instead of just data, like the discrete for transform.

57
00:06:53.369 --> 00:06:56.639
So, let me write it as it is from the bottom.

58
00:06:56.639 --> 00:07:00.509
As a, uh, tenser product irritation to.

59
00:07:00.509 --> 00:07:03.598
Uh, we can see that.

60
00:07:03.598 --> 00:07:12.449
Could be naturally in clinic clinic computer, because we're just taking each of the States, uh, and based on the other Cuba's performing a rotation.

61
00:07:12.449 --> 00:07:19.288
So, this results, the output is a superposition of analytics of frequencies based on the input signal.

62
00:07:19.288 --> 00:07:23.879
Um, so why is this useful? Why do we need a quantum version of this transformation?

63
00:07:23.879 --> 00:07:30.149
Oh, actually sorry, I'm bringing it up myself. So I want to instead of going through the math and trying to derive it.

64
00:07:30.149 --> 00:07:34.559
Wanted to get instead give a kind of intuitive sense of what's happening in this transformation.

65
00:07:34.559 --> 00:07:40.649
To the top road, this is an example for you that's just animation is provided from get documentation.

66
00:07:40.649 --> 00:07:47.038
Will the top row shows to inputs? You can set up your Cubics and either a 0T input or a 1 input and.

67
00:07:47.038 --> 00:07:53.968
We can see the Cubans there is a significant bit Cubics, threes and most significant a bit and it's just binary counterclaims.

68
00:07:53.968 --> 00:07:57.088
From 0T to 6 0T to 15.

69
00:07:57.088 --> 00:08:04.019
I'm on the bottom, we have the state of the Cuba's in the 4 year basis for each of these inputs.

70
00:08:04.019 --> 00:08:08.218
The 4 year basis is denoted with the little tell day above the number, because you can see.

71
00:08:08.218 --> 00:08:12.358
Um, and basically, so what the 4.

72
00:08:12.358 --> 00:08:16.379
1, before you're transformed, does it just transforms the input from the top.

73
00:08:16.379 --> 00:08:19.588
To the, for your business on the bottom and we can see.

74
00:08:19.588 --> 00:08:24.538
Basically that the outputs kind of behave counters they're just counting how.

75
00:08:24.538 --> 00:08:28.918
Uh, what's the input is changing? We can see the cube 3 on the right side.

76
00:08:28.918 --> 00:08:32.818
Changes the quickest every time the input increases by 1.

77
00:08:32.818 --> 00:08:37.019
Cube it rotates by 180 degrees around the access.

78
00:08:37.019 --> 00:08:42.839
And then each next cube bit rotates by half of that. So, keep it to is rotating by.

79
00:08:42.839 --> 00:08:47.308
5 or 2 or 90 degrees, keep it 3, but power for.

80
00:08:47.308 --> 00:08:57.418
Keep it 0T by rate so this is the transformation that's happening and we'll be using these diagrams to show what our expected input and output is in our, uh, test in the future.

81
00:08:58.918 --> 00:09:03.028
So this circuit is implemented.

82
00:09:03.028 --> 00:09:06.778
In the following a block diagram, this is again provided by kit.

83
00:09:06.778 --> 00:09:12.028
And essentially, what we're doing is, we're taking you to, to keep it on the left.

84
00:09:12.028 --> 00:09:19.859
And we're getting them to that a 4 year basis we saw on the previous slide. So, for each, hopefully can see my cursor.

85
00:09:19.859 --> 00:09:25.528
For each kid, we're performing the hand of our gate, which brings it from the, uh.

86
00:09:25.528 --> 00:09:29.339
0T on 1 down to the, uh.

87
00:09:29.339 --> 00:09:32.788
X Y, plane on those, uh, those fears.

88
00:09:32.788 --> 00:09:40.828
And then performing a rotation from each based on each other cube that, uh, based on that tenser product we saw a few slides previously.

89
00:09:40.828 --> 00:09:46.168
And we do that again for each other cube it. Um, and then at the end, we see, there are some sockets.

90
00:09:46.168 --> 00:09:50.308
Uh, this is implementation dependence. Some do some don't because.

91
00:09:50.308 --> 00:09:54.749
When you work out the math, basically, the orders are reversed so.

92
00:09:54.749 --> 00:10:01.048
You can either just re, label your outputs. So this you could say is output. And this is output uh, 1.

93
00:10:01.048 --> 00:10:05.519
Or some implementations will actually performance soap so all the labels are correct.

94
00:10:06.899 --> 00:10:11.578
So, why is this, uh, transformation useful? What, what's the purpose of implementing it?

95
00:10:11.578 --> 00:10:17.908
So 1st of all, what does a measurement of the give you? We look at the topic and we see this.

96
00:10:17.908 --> 00:10:27.688
The same background from before if we take a measurement to the immediate queue that we can see, they're equally space between the 0T and the 1 state. So when you take a measurement.

97
00:10:27.688 --> 00:10:34.229
Those kids, 0, 1 with 50% probability and well, that doesn't seem very useful. Right? You don't get any information out of this transformation.

98
00:10:34.229 --> 00:10:37.499
So this is.

99
00:10:37.499 --> 00:10:43.589
Taking a measurement of something a signal where you've taken the forth quantify transform isn't useful.

100
00:10:43.589 --> 00:10:49.408
Instead you have to perform a clinic for a transform to get your system into this.

101
00:10:49.408 --> 00:10:55.889
State, you can do some work in a 4 year basis, similar, like, doing work in the frequency basis and just.

102
00:10:55.889 --> 00:10:59.188
In standard single processing.

103
00:10:59.188 --> 00:11:03.418
And then use the inverse clinic for a transform to get back to your competition basis where you can actually.

104
00:11:03.418 --> 00:11:10.139
Perform useful measurements, so we've talked previously in this class about shores algorithm and phase measurements.

105
00:11:10.139 --> 00:11:16.229
Or phase estimation, and this is 1 of the uses where you can easily compute the period of.

106
00:11:16.229 --> 00:11:21.298
Basically, not end to easily factor in it makes use of this corner for a transform.

107
00:11:21.298 --> 00:11:26.908
And then phase estimation using the counting property, you can list, you compare phases of different Cubics.

108
00:11:28.198 --> 00:11:35.578
Um, and then finally some implementation issues that come up along when you actually implement the clinic for a transform.

109
00:11:35.578 --> 00:11:49.109
So, we saw that you have to perform a lot of rotation around the access to actually get your human to correct state. So, but maybe 1 or 2. Q this is not too bad. You were hurting by, like, pie or pie over 2.

110
00:11:49.109 --> 00:11:54.448
And there are some specialized gates for that, but then when you need to do some arbitrary rotation round, does he access.

111
00:11:54.448 --> 00:12:01.589
That can get very hard to do it. Precisely. So there are some quantum computers that support this. You can give them an angle and.

112
00:12:01.589 --> 00:12:06.688
Did the best wrote into that, but they'll always be error. So that's going to show up as error in your output.

113
00:12:06.688 --> 00:12:14.339
An alternative is to use multiple applications of H. R. Gates and T gates with rotate around disease by prior before.

114
00:12:14.339 --> 00:12:17.639
And he's have a special special property.

115
00:12:17.639 --> 00:12:23.729
Since rotation or irrational, multiple of pie can get you to any arbitrary angles. If you just rotate enough times.

116
00:12:23.729 --> 00:12:28.019
They can get you to some degree of precision and new angle you want.

117
00:12:28.019 --> 00:12:33.058
So, that can let you implement these precise rotations at the cost of a lot of Gates.

118
00:12:33.058 --> 00:12:45.958
And obviously, as in any final algorithm Dolby noise and accuracy some rotations, and just due to external energy areas in Q4 and transform. So we performed this on different computers and we'll be looking at how different computers compare.

119
00:12:47.278 --> 00:12:55.379
So, our test process, we didn't want to so, as mentioned previously, you can't run for a transform and then perform a measurement and get anything useful.

120
00:12:55.379 --> 00:12:59.759
So, instead we've implemented inverse for our transform on different computers.

121
00:12:59.759 --> 00:13:07.649
We'll initialize our cubes into 4 year basis and remember to inverse 1 for your transform to see if we get back to the expected computational state.

122
00:13:07.649 --> 00:13:13.048
So, in Amazon bracket, or we ran our test on that, we're getting estimate machine.

123
00:13:13.048 --> 00:13:17.849
And in kiss, get, we run our tests on the IBM quantum computers and, uh, you.

124
00:13:17.849 --> 00:13:24.239
Agnes, as well as using the igneous error medication scheme, which helps produce, uh, errors of the measurement.

125
00:13:25.619 --> 00:13:32.458
The 1st, Amazon bracket, they have several back ends, including traditional quantum computers provided by and.

126
00:13:32.458 --> 00:13:39.688
And klonopin Needless provided by D wave declining. The dealers are suited to linear programming problems, but not more traditional.

127
00:13:39.688 --> 00:13:43.078
Problems we were working with here, so we could not use the systems.

128
00:13:43.078 --> 00:13:48.719
Uh, the eye on Q, systems unfortunately do not support the, uh, arbitrary rotations.

129
00:13:48.719 --> 00:13:55.438
That are required for this, and as as we mentioned previously and implement this with multiple.

130
00:13:55.438 --> 00:14:03.479
H N. T. Gates, that's not directly supported by the compiler and we can I can find the settings to make the value that automatically.

131
00:14:03.479 --> 00:14:08.188
So, I've been looking into, uh, doing manually, but we haven't run test for this.

132
00:14:08.188 --> 00:14:13.649
Presentation, so this is the circuit that was implemented using 3. Q bits.

133
00:14:13.649 --> 00:14:20.609
Unfortunately bracket doesn't give you a nice picture output like this get does, but you can still see the, um.

134
00:14:22.019 --> 00:14:28.318
Uh, the state or the set of the circuit, we're performing each of the head of our Gates different rotations and then.

135
00:14:28.318 --> 00:14:33.298
Uh, the bracket information performs a slot beforehand, but then.

136
00:14:33.298 --> 00:14:36.989
So these right here are setting up our initial.

137
00:14:36.989 --> 00:14:40.708
State in the 4 year basis and then.

138
00:14:40.708 --> 00:14:44.548
We perform a rotation based on each other cubic to.

139
00:14:44.548 --> 00:14:48.958
Or from the inverse for your transform, and finally we measure the output of the States.

140
00:14:48.958 --> 00:14:54.418
So, for 3, I'll be ran the following tests where we set up to.

141
00:14:54.418 --> 00:15:00.538
In the state right here and the expected output is basically too, we expect their.

142
00:15:00.538 --> 00:15:05.458
Uh, yeah, when we run the idea to base, like this out per right here.

143
00:15:05.458 --> 00:15:11.548
And when we run that with 3 Q, bits on the estimate machine, this histogram is the results of what we got.

144
00:15:11.548 --> 00:15:16.649
Approximately 66% resulted in the correct solution. We asked the machine.

145
00:15:16.649 --> 00:15:22.589
And this was very good result. It was only 3 events, but since we're rotating by.

146
00:15:22.589 --> 00:15:27.208
On the pie or fiber 2, it can use, um.

147
00:15:27.208 --> 00:15:32.278
Some very basic case from those rotations, which have a lot higher level of precision.

148
00:15:32.278 --> 00:15:35.458
We need to move to higher number of Cubics the, uh.

149
00:15:35.458 --> 00:15:38.458
Accuracy drops or they.

150
00:15:38.458 --> 00:15:44.818
Success rate drops very quickly, but for that, we only had 6.2. correct 5. Q. that's 209.

151
00:15:44.818 --> 00:15:48.178
And thank you, that's basically a.

152
00:15:48.178 --> 00:15:52.168
Sampling of the all of the output States.

153
00:15:52.168 --> 00:15:58.948
Some counts, like we're up to 12, but generally it was like, 1 to 2 counts of all of the up States.

154
00:15:58.948 --> 00:16:05.068
So clearly, some sort of error, mitigation techniques, or higher quality.

155
00:16:05.068 --> 00:16:10.469
Um, quantum computers will be required to make use of just transform at higher.

156
00:16:10.469 --> 00:16:19.469
Cubic counts com is going to be going over the IBM case, get implementation and some error mitigation techniques.

157
00:16:19.469 --> 00:16:26.009
Yeah, so next we're going to look at how the IBM computers performance compares to the we're getting that. We just saw.

158
00:16:26.009 --> 00:16:31.019
Here's the same circuit from before, but implemented and.

159
00:16:31.019 --> 00:16:41.969
Uh, the only real difference here is that you had those 2 swaps at the beginning, which is just because I find the way that Jessica orders the Cubics to be a little bit confusing. So I just fix that right there the beginning.

160
00:16:41.969 --> 00:16:49.859
Since has the prettier circuit drawing feature hopefully it's a bit easier to see how this relates to the forward.

161
00:16:49.859 --> 00:16:56.489
Where at each stage, you have the rotation controlled by each pair of Cubics and then the head of market.

162
00:16:56.489 --> 00:16:59.788
And then finally, at the end, do you have the measurements.

163
00:16:59.788 --> 00:17:05.878
And those gates at the beginning are just setting up the, uh, the input in the, the 48 basis.

164
00:17:05.878 --> 00:17:14.189
We used 5 Cubics for this pretty much just because all of the available quantum computers for IBM or 5 cubic computers.

165
00:17:14.189 --> 00:17:25.348
And I had a little note there at the bottom. I took a look at the quantum assembly after the job ran and it turns out that when all this gets to the actual computer.

166
00:17:25.348 --> 00:17:32.759
Even though, this looks fairly condensed here on the screen. This actually gets decomposed into 117 individual Gates.

167
00:17:32.759 --> 00:17:37.318
So, even though this doesn't look too bad, it's actually a decently complex.

168
00:17:40.769 --> 00:17:48.808
Just like with the, we're getting, we're going to look at the simulated results 1st for these tests. I gave the circuit an input of 2007, because that's my favorite number.

169
00:17:48.808 --> 00:17:52.979
And you can see what that looks like in the 4 year basis at the top.

170
00:17:52.979 --> 00:17:59.999
And then at the bottom, you can also see what that looks like on the blogosphere diagram after it's been.

171
00:17:59.999 --> 00:18:03.509
Put through the inverse for a transform.

172
00:18:03.509 --> 00:18:07.798
You can see from there it should be 11011.

173
00:18:07.798 --> 00:18:15.118
And we run the simulator, it matches that. And 11011 is in fact, 2007 in binary. So.

174
00:18:15.118 --> 00:18:26.759
Based on this good sign that our circuit is correct now here are the results from running the circuit on the IBM queue Athens machine.

175
00:18:26.759 --> 00:18:33.148
You can see that we get much wider spread of states obviously, than we are getting from the simulator.

176
00:18:33.148 --> 00:18:38.489
And unfortunately that we only get the correct output 10.8% of the time.

177
00:18:38.489 --> 00:18:50.038
Now, that's better than the same circuit on there. We're getting computer. So from this, we can say that the IBM computers are a bit better at running these content for transforms and they're getting used based on the data. We've collected.

178
00:18:50.038 --> 00:18:54.269
But that's still not even the most, like the result here on this chart.

179
00:18:54.269 --> 00:18:59.788
So, what's going on the 1st thing that I noticed about this?

180
00:18:59.788 --> 00:19:07.828
Was that that big cheek in the middle is only 1 cube off from the value that we're actually looking for?

181
00:19:07.828 --> 00:19:17.999
And I found that if you add together, all of the states that are most 1 Cupid away from the correct state, we get 42.2%, which is a lot better.

182
00:19:17.999 --> 00:19:21.239
So 1 question that you might ask is.

183
00:19:21.239 --> 00:19:33.749
Is it just the final measurements that are off? You know, maybe the circuit works mostly correctly, but then at the end, when you have to measure those uncertain queue, that's 1 of them just gets measured a little bit wrong. And that's how we end up with that error.

184
00:19:33.749 --> 00:19:39.118
So with that in mind, let's talk about error correction.

185
00:19:40.259 --> 00:19:52.558
As we just saw, and I'm sure you've heard many times this semester noise in the hardware is a huge factor that limits the usefulness of our current quantum computers.

186
00:19:52.558 --> 00:19:58.618
I mean, like, we just tried to do a 5 bit inverse for a transform and the results were barely meaningful.

187
00:19:58.618 --> 00:20:08.969
So, even if we consider the exponential information that gets stored in Cuba, it's 2 to the 5th classical bits is the size of a single imager, almost classical computers. So.

188
00:20:08.969 --> 00:20:14.729
In theory, we're not dealing with that heart of a problem, but the noise is completely ruining our algorithm.

189
00:20:14.729 --> 00:20:23.608
So, we know that a lot of people in the field are working on trying to fix the noise and the hardware in order to solve this problem.

190
00:20:23.608 --> 00:20:29.848
But for right now, and especially at our level, that's not really an option that we have to try to address this.

191
00:20:29.848 --> 00:20:36.628
So I'd like to take a little bit to talk about what we as software engineers can do to try to mitigate some of this noise.

192
00:20:36.628 --> 00:20:45.269
And so to start, let's look at where the noise comes from, right? It can come from either the gates themselves that are carrying out the operations on the cube.

193
00:20:45.269 --> 00:20:53.669
Or it can come from the measurement at the end and based on what we're talking about in the last slide, we're going to 1st, talk about the measurement error.

194
00:20:53.669 --> 00:21:00.148
And what we want to do is we want to try to quantify the error that the measurement Gates create.

195
00:21:00.443 --> 00:21:15.023
And then remove it from the results and as it turns out, while I was looking into the documentation, it has built in functionality to do this. So, I wanted to try it out on this quantum and this inverse quantum for transform circuit and see what that fit for our data.

196
00:21:17.068 --> 00:21:27.868
It can go to the next slide so just going to do a little bit of math here not too bad, but we're gonna say, right that if you have a cubic that's in state and you measure it.

197
00:21:27.868 --> 00:21:34.318
There is going to be some probability that it ends up as each of the Q2, the and possible states right?

198
00:21:34.318 --> 00:21:38.818
So, hopefully, most of the time state is going to measure that state. I.

199
00:21:38.818 --> 00:21:43.858
But it could be measured as any of the other states really depending on the noise.

200
00:21:43.858 --> 00:21:49.739
So, if we say that we have a vector, our of our state probabilities.

201
00:21:49.739 --> 00:21:55.588
And you have, you know, what they're supposed to be before you measure them, we're going to call that are true.

202
00:21:55.588 --> 00:22:06.778
You can construct that matrix there and that's how that's how you end up with the, the noisy data, which is what we actually get from the quantum computer.

203
00:22:06.778 --> 00:22:13.798
So, if you want to get our true, we just have to invert that Matrix, which I'm just going to call them from now, on for.

204
00:22:13.798 --> 00:22:24.628
Brevity and multiply that by our noisy and we should get a slightly better estimate of our results because we've removed that measurement noise.

205
00:22:24.628 --> 00:22:28.348
And so, 1st, we have to calculate that matrix in order to do that.

206
00:22:28.348 --> 00:22:37.138
So before we do our actual job, we do a separate job where we just initialize the system to a certain state. So, let's just say, 00000.

207
00:22:37.138 --> 00:22:41.308
And then we measure that state, so we measure all 5 of the.

208
00:22:41.308 --> 00:22:44.459
And we do that for, say, a 1000 shots.

209
00:22:44.459 --> 00:22:49.108
And then from there, we just measure the probabilities.

210
00:22:49.108 --> 00:22:56.159
That they measure 000T, then it measures 00001 and so on.

211
00:22:56.159 --> 00:23:01.828
And then that becomes the 1st column of that matrix and you just repeat that for all 2 out of the States and you can stop that matrix.

212
00:23:01.828 --> 00:23:06.419
Now, ideally, if there was no noise, that would just be guy, then the identity matrix.

213
00:23:06.419 --> 00:23:15.058
But we all know if that's not going to be the case. So if we go to the next slide, we can see what happens when we do that.

214
00:23:15.058 --> 00:23:18.808
It turns out it's a little bit better.

215
00:23:18.808 --> 00:23:25.138
After we ran that sort of measurement gate calibration on IBM. Q Athens.

216
00:23:25.138 --> 00:23:32.939
And then applied that inverse matrix. It brought up the probability of the correct answer from 10.012.2%.

217
00:23:32.939 --> 00:23:36.509
So, definitely some improvement there.

218
00:23:36.509 --> 00:23:43.048
Now, this is not just random break. This is actually getting improvement from moving that noise.

219
00:23:43.048 --> 00:23:51.598
But it's still not great, right? Like, it's still not the highest. So we can say most of the error here is not due to measurement, but some of it clearly was.

220
00:23:51.598 --> 00:23:54.628
Silence.

221
00:23:54.628 --> 00:24:05.278
Now, because we are presenting today on the 1st day, we're not quite done with all of our experimentation. So I want to talk a little bit about what we plan to do over the next week.

222
00:24:05.278 --> 00:24:09.388
And remember.

223
00:24:09.388 --> 00:24:18.358
A couple slides back I said that we're going to talk gate error as well and we do have a solution for that, and it's called 0T noise. Extrapolation.

224
00:24:18.358 --> 00:24:25.348
So, the idea here is that, since we have to deal with the noise, anyway, we may as well try to extract some data from it.

225
00:24:25.348 --> 00:24:31.679
So, the way we do this is we just alter the circuit that we feed into the quantum computer.

226
00:24:31.679 --> 00:24:35.848
So to do that, we just take each gate we're going to call that queue.

227
00:24:35.848 --> 00:24:39.298
And then we add queue and it's inverse after that.

228
00:24:39.298 --> 00:24:44.278
So, if you look at that, I mean, the queue in the inverse cancel out and you just left with. Q. so it's effectively the same circuit.

229
00:24:44.278 --> 00:24:48.538
But we've tripled the noise and.

230
00:24:48.538 --> 00:25:00.568
Our hope here is that by adding more noise to the circuit, and we can do this in stages and controllable increase the noise that's created in our quantum circuit.

231
00:25:00.568 --> 00:25:10.229
We might be able to find some trends and how the gate noise affects our results. And so they should help to address the noise that's due to the individual Gates.

232
00:25:10.229 --> 00:25:18.659
I put that little line there on the bottom as a note to anyone who might be interested in trying this. The IBM compiler is smart.

233
00:25:18.659 --> 00:25:22.618
And if you try to do this, it will just.

234
00:25:22.618 --> 00:25:34.679
Optimize this back to your original circuit, because it knows that you're just adding more noise. And in most cases that that's not desirable. So you do have to explicitly tell the compiler to.

235
00:25:34.679 --> 00:25:43.078
Not optimize it out. All right and the tool that we are going to use for this.

236
00:25:43.078 --> 00:25:50.128
Is a piece of software called mimic? It's quite new. It was just released in September and it's still.

237
00:25:50.128 --> 00:25:53.398
Kind of buggy, which is why I haven't quite figured out yet.

238
00:25:53.398 --> 00:26:01.739
But it's a cross compiler that for quantum computers that supports this sort of analysis you can see that block diagram on the right there.

239
00:26:01.739 --> 00:26:04.828
You feed your circuit antiemetic.

240
00:26:04.828 --> 00:26:09.868
And it creates those noise scaled circuits that I was talking down the last slide.

241
00:26:09.868 --> 00:26:13.949
Execute it and then you get your results back.

242
00:26:13.949 --> 00:26:18.479
And then, once you have those results, you can create these plots that you see down here.

243
00:26:18.479 --> 00:26:25.138
Where the X axis you have the noise scaling so that's like, by what factor you've increased the noise in your circuit.

244
00:26:25.138 --> 00:26:32.038
And then your Y, axis is for any given state what that the probability of getting that state measurement is.

245
00:26:32.038 --> 00:26:38.459
And you have that data, and then what it does is, it fits.

246
00:26:38.459 --> 00:26:41.699
A bunch of different trend lines to the data.

247
00:26:41.699 --> 00:26:46.348
And tries to figure out if it extrapolates that trend.

248
00:26:46.348 --> 00:26:55.169
What the the results should be if you're noise scaling factor is 0T, which is your ideal quantum computer was 0T noise.

249
00:26:55.169 --> 00:26:59.429
And so you can see there, both of these plots, the.

250
00:26:59.429 --> 00:27:06.719
Ideal value here is 1, this is just a circuit that should be measuring just a 1, because it's just a bunch of rotations and then there in versus.

251
00:27:06.719 --> 00:27:09.808
And so you can see.

252
00:27:09.808 --> 00:27:13.528
Almost all of the trend lines from the.

253
00:27:13.528 --> 00:27:17.578
The extrapolation give you a better results than.

254
00:27:17.578 --> 00:27:21.538
What you started out with with the scaling factor of 1.

255
00:27:21.538 --> 00:27:27.509
So this is something that we are going to hope to.

256
00:27:27.509 --> 00:27:35.338
Collect data from and also be able to report on how this affects the results of the quantum for a transform in a more complicated circuit.

257
00:27:35.338 --> 00:27:41.699
Oh, yeah, I think I missed a point here. I say, supposedly supports circuit because.

258
00:27:41.699 --> 00:27:51.868
It seems like it has some features that aren't supported by it just because it's been worked on a little bit longer. Like there are some Gates and casket specifically.

259
00:27:51.868 --> 00:27:55.078
The controlled rotation gate.

260
00:27:55.078 --> 00:27:58.709
That gets seems to support, but not.

261
00:27:58.709 --> 00:28:04.739
So definitely still some work that needs to be done on the software there. But I think, I think you can work around it.

262
00:28:06.598 --> 00:28:11.608
All right, I think that's all we have to say. So, if anyone like to ask questions, please, do.

263
00:28:11.608 --> 00:28:15.868
Cool. Thank you. Very much. Floor is open.

264
00:28:16.919 --> 00:28:21.749
So you think, um, I'll talk to so IBM does this better.

265
00:28:21.749 --> 00:28:27.239
I think they're getting or whatever. It's certainly what it seems like from the data. Yeah.

266
00:28:28.499 --> 00:28:34.858
Okay, how many pets do you think you could do? Or how many cubes could you usefully.

267
00:28:34.858 --> 00:28:42.088
Do for you transform, like, looking at the results on 5 I'm thinking you could do about 4, usefully.

268
00:28:42.088 --> 00:28:47.398
Or something uh, yeah, so we were able to do a larger on a.

269
00:28:47.398 --> 00:28:52.078
Forgetting and, like, obviously 10, we are up to 30 and that was just outline.

270
00:28:52.078 --> 00:28:56.159
But probably, maybe 4 with more shots, maybe with some.

271
00:28:56.159 --> 00:29:05.489
Or many techniques, you can get some useful data that okay, thank you any other questions. Okay, I'll show justin's video now.

272
00:29:05.489 --> 00:29:13.798
Assuming I can get it to work. Well, I'll get Justin and today I'll give you a chance and.

273
00:29:13.798 --> 00:29:17.638
To make sure things are working and.

274
00:29:17.638 --> 00:29:24.659
Hello, everybody my name is Justin and you just give me a 2nd tier to.

275
00:29:25.709 --> 00:29:31.979
Share everything I think there was 1 question in the chat.

276
00:29:31.979 --> 00:29:39.778
Oh, sorry. Yeah, so to compare the accuracy of the 2 systems, fearing different, using different preparation States.

277
00:29:39.778 --> 00:29:46.469
So, while, yeah, I guess they do start in different like initial States.

278
00:29:46.469 --> 00:29:53.489
But that's more of a like a label. We're just defining this numbers to be these starting States.

279
00:29:53.489 --> 00:29:57.028
So, technically they are off by like, a.

280
00:29:57.028 --> 00:30:01.169
The internal Jupiter off by relative phase um.

281
00:30:01.169 --> 00:30:05.519
So, it could have different noise characteristics, which may be, uh.

282
00:30:05.519 --> 00:30:09.898
A valid point of error so maybe we should try.

283
00:30:09.898 --> 00:30:14.219
Comparing them with the same initial state, it's a.

284
00:30:14.219 --> 00:30:17.818
Good point, but in the end they're all performing the same computation.

285
00:30:17.818 --> 00:30:21.509
So, that'd be a dude errors due to that, I think, would be due to.

286
00:30:21.509 --> 00:30:26.308
Inherent biases inside the Cubics, which could compare between the systems.

287
00:30:28.469 --> 00:30:32.788
Okay, thank you else.

288
00:30:32.788 --> 00:30:37.979
So.

289
00:30:39.989 --> 00:30:46.229
I'm going to be talking about using Matt lab to simulate a quantum computer.

290
00:30:47.308 --> 00:30:51.568
For our, I'm going to start off by discussing why Matlab.

291
00:30:51.568 --> 00:30:57.419
Is a really nice tool for modeling behavior of Cubans and their interactions with quantum Gates.

292
00:30:57.419 --> 00:31:01.108
Then I'm going to give a little bit of a mathematical background.

293
00:31:01.108 --> 00:31:08.788
On the mathematical model of quantum where quantum states and quantum gates can be represented by using matrices.

294
00:31:08.788 --> 00:31:14.669
And once this foundation is established, I'm going to be showing off some of my own Matlab functions that I wrote.

295
00:31:14.669 --> 00:31:17.999
In order to demonstrate how that lab is so intuitive and.

296
00:31:17.999 --> 00:31:21.778
Relatively effortless when it comes to modeling quantum behavior.

297
00:31:21.778 --> 00:31:28.108
And this will hopefully provide some inspiration for you to want to create your own quantum simulate.

298
00:31:28.108 --> 00:31:32.548
Finally, I'm going to be discussing existing quantum simulators.

299
00:31:32.548 --> 00:31:36.269
That use Matlab and I'm going to make some suggestions as.

300
00:31:36.269 --> 00:31:39.509
To how these simulators could potentially be improvement.

301
00:31:39.509 --> 00:31:42.598
So, what makes Matlab such a good tool.

302
00:31:42.598 --> 00:31:48.328
Well, Matlab was designed for computational mathematics and it's a matrix based language.

303
00:31:48.328 --> 00:31:54.148
And, as a matter of fact, seen in the 1st bullet point, here we see that that lab actually stands for.

304
00:31:54.148 --> 00:31:58.858
Matrix laboratory and and specializes in matrix operations.

305
00:31:58.858 --> 00:32:04.769
Also, Matlab has built in matrix functions, such as buying the trace of the inverse of matrix and.

306
00:32:04.769 --> 00:32:07.979
Finding the tenser product between 2 matrices.

307
00:32:07.979 --> 00:32:13.888
In addition to being a tool for matrix operations that lab also recognizes the letter. I.

308
00:32:13.888 --> 00:32:18.628
As an imaginary number, and it can handle operations that involve complex numbers.

309
00:32:18.628 --> 00:32:23.939
So some built in functions for complex numbers, include find the angle.

310
00:32:23.939 --> 00:32:27.088
Of a complex number, or it's complex conjugate or.

311
00:32:27.088 --> 00:32:34.019
Even plotting accomplish, finally, Matlab has excellent plotting functions to visualize data.

312
00:32:34.019 --> 00:32:40.528
So, on the next few slides, we're going to see why this is all such good news for creating a quantum civic.

313
00:32:41.969 --> 00:32:49.709
So, we're going to start off by talking about the States held by the Cubans and how they can be represented by using matrices.

314
00:32:49.709 --> 00:32:53.009
So, we can represent Cuban States as a.

315
00:32:53.009 --> 00:32:58.949
1 column matrix showed right here and the number of rows in the matrix is.

316
00:32:58.949 --> 00:33:03.479
Equal to 2 to the end where and is the number of Cubics involved.

317
00:33:03.479 --> 00:33:07.828
So, in this 1st, example, here we see that the 1st row.

318
00:33:07.828 --> 00:33:14.159
Corresponds to a state of 0T while the 2nd row corresponds to state 1.

319
00:33:14.159 --> 00:33:18.388
And the numbers of these states may contain imaginary numbers.

320
00:33:18.388 --> 00:33:23.068
But the magnitude of these states cannot exceed 1.

321
00:33:23.068 --> 00:33:27.028
So, when normalized as seen in this 3rd bullet point here.

322
00:33:27.028 --> 00:33:31.048
The states must add to a probability of 1.

323
00:33:31.048 --> 00:33:36.209
So, on the next side, and we're going to be discussing the quantum case that could be applied.

324
00:33:36.209 --> 00:33:44.189
To these states quantum, the operations can be represented as unitary matrices.

325
00:33:44.189 --> 00:33:49.078
Which could be applied to the state matrices by using matrix multiplication.

326
00:33:49.078 --> 00:33:55.259
There are a number of different types of gates that could be either real imaginary or a complex.

327
00:33:55.259 --> 00:33:58.439
So, on the next slide, the had our gate.

328
00:33:58.439 --> 00:34:04.288
Is going to be used as an example of applying a quantum gate to a 1 Cuba.

329
00:34:06.118 --> 00:34:10.528
And this example a, had a gate is being applied to a 1 cubic state.

330
00:34:10.528 --> 00:34:16.978
Where are the probability of measuring state? 0T is equal to 1 while the probability of measuring state 1 is equal to 0.

331
00:34:16.978 --> 00:34:22.378
And this is the initial state of the cube after the matrix multiplication completed.

332
00:34:22.378 --> 00:34:30.989
And the gate has been applied, we see that. The new Cuban state is either in state 0T with probably 1 half or and stay 1 with probability.

333
00:34:30.989 --> 00:34:35.039
So this is an operation that could easily be done by using that lab.

334
00:34:35.039 --> 00:34:40.199
So, on the next slide, I'm going to show you my 1 cube, the head of our game function that I created.

335
00:34:41.668 --> 00:34:45.358
The 1, you've had a market function, takes a quantum state.

336
00:34:45.358 --> 00:34:49.918
Such as 1, 0T or 1 over square to 2, 1 over square to you.

337
00:34:49.918 --> 00:34:53.818
As inputs and outputs, the probability of.

338
00:34:53.818 --> 00:35:02.398
Each state after it goes through the head of R. D. the function also plots the probability of each state by using a Barbra a show.

339
00:35:03.418 --> 00:35:09.329
So the function begins by declaring a 2 by 2, unitary matrix and.

340
00:35:09.329 --> 00:35:13.858
Creating a column vector to represent the current state, or the initial.

341
00:35:13.858 --> 00:35:17.338
Then, uh, we see that line for.

342
00:35:17.338 --> 00:35:21.599
Performs that matrix multiplication that was seen on the previous example of.

343
00:35:21.599 --> 00:35:28.498
Maybe a slide and afterwards, the probability for each state is calculated and plotted on the partner.

344
00:35:28.498 --> 00:35:33.478
So, make sure on the right shows the same output as calculated in the previous.

345
00:35:33.478 --> 00:35:37.619
So, on the next slide, I'm going to be showing.

346
00:35:37.619 --> 00:35:40.739
A 1 cubic white function as well.

347
00:35:40.739 --> 00:35:45.898
Now, the wagging function is nearly identical to the head of market function.

348
00:35:45.898 --> 00:35:50.579
With the exception that a different unitary matrix is being applied.

349
00:35:50.579 --> 00:35:59.068
To the state, so this function was written to show that Matlab also could form computations using imaginary numbers or complex numbers.

350
00:35:59.068 --> 00:36:03.599
So the fact that has such a great tool kit for.

351
00:36:03.599 --> 00:36:06.778
Matrix operations imaginary numbers and data.

352
00:36:06.778 --> 00:36:11.278
Makes a very good tool for a beginner to create their own quantum simulate.

353
00:36:11.278 --> 00:36:15.449
And in addition to being able to create your own quantum simulator.

354
00:36:15.449 --> 00:36:20.278
There are also existing quantum computers simulator toolkits that you could download from MetLife.

355
00:36:20.278 --> 00:36:28.139
Which we will see on the next slide 1, existing quantum computer simulator tool kit that I found this called.

356
00:36:28.139 --> 00:36:40.498
Quantum computer simulator written by Martin, this simulator uses matrices to model quantum states and uses matrix multiplication and tenser product to represent quantum gate operations.

357
00:36:40.498 --> 00:36:44.759
The result of running the simulation is a measurement probability distribution.

358
00:36:44.759 --> 00:36:52.768
Which comes from normalizing the state that the simulation is characterized by its main function called quantum computer.

359
00:36:52.768 --> 00:36:56.458
Which is responsible for running quantum algorithms.

360
00:36:58.498 --> 00:37:02.668
The 1st, input to the quantum computer function is called Q algorithm.

361
00:37:02.668 --> 00:37:05.818
And this is the quantum circuit that's going to be simulated.

362
00:37:05.818 --> 00:37:12.268
So, this works with any number of events, and it uses all of the well known quantum Gates.

363
00:37:12.268 --> 00:37:17.878
As shown at the examples below, we see that putting a space between each gate.

364
00:37:17.878 --> 00:37:23.369
Well, cause it gets to be placed on the same cubic row while putting a semi colon between them.

365
00:37:23.369 --> 00:37:27.539
Will cause the gates to be placed on separate cubic rooms?

366
00:37:28.858 --> 00:37:32.458
The next input to the algorithm is measured Cubans.

367
00:37:32.458 --> 00:37:36.358
And this input, you put the Cubics that you wish to put through the quantum circuit.

368
00:37:36.358 --> 00:37:41.398
And finally, the last input is the initial state.

369
00:37:41.398 --> 00:37:44.849
And this was to be entered as a 1 column matrix with.

370
00:37:44.849 --> 00:37:48.809
Um, to to the end rows, where, and is the number of measured Cubans.

371
00:37:48.809 --> 00:37:54.748
Unfortunately, the simulator is not able to check if these initial States are able to be normalized.

372
00:37:54.748 --> 00:37:57.898
So the user should watch out for that before and putting it.

373
00:37:57.898 --> 00:38:05.309
As the initial state, so shown below are some examples for inputs for the initial state.

374
00:38:05.309 --> 00:38:09.989
So, as you can see here, 1 over square to to 1 over square to normalize 2, 1.

375
00:38:09.989 --> 00:38:15.449
And so does 1 app on that 1 and now it's time to run the function.

376
00:38:16.889 --> 00:38:22.079
So Here's an example of a 3 cube circuit being run by using the.

377
00:38:22.079 --> 00:38:30.090
Quantum computer function it has an initial state of 1 and the 1st state and 0T and all the other 7 States is shown right here.

378
00:38:31.110 --> 00:38:36.659
And the function returns for outputs and, uh, these outputs include.

379
00:38:37.710 --> 00:38:40.710
A graph of the probability distribution for each state.

380
00:38:40.710 --> 00:38:51.239
The vector of each output state, the probability distribution of the output States, and the overall matrix operation that was applied to the.

381
00:38:51.239 --> 00:38:56.219
The initial Cuban state, this output is known as out.

382
00:38:56.219 --> 00:39:02.099
And it's not included in this slide, but in our case is equal to.

383
00:39:02.099 --> 00:39:05.280
The head of our matrices multiplied by.

384
00:39:05.280 --> 00:39:15.929
The Spock Gates matrices, the quantum computer function has a limit. So it suggested that you only run small quantum circuits on it.

385
00:39:15.929 --> 00:39:22.440
I try to working with a different number of Cubics, and I ended up finding that fortune cube was the cut off point.

386
00:39:22.440 --> 00:39:26.039
At which the algorithm would start to take around 10 minutes to run.

387
00:39:26.039 --> 00:39:29.940
And this is probably because of the print statements as well as updating the graph.

388
00:39:29.940 --> 00:39:33.900
But, overall, if he keeps the algorithm, as it is, it will take around.

389
00:39:33.900 --> 00:39:39.119
10 minutes to run at 14 Cubans. So if we compare this to IBM circuit composer.

390
00:39:39.119 --> 00:39:45.539
Uh, on their browser, the output states that are calculated when you create the circuit are competed almost.

391
00:39:45.539 --> 00:39:53.880
And however, the measurement probabilities, uh, if you were to click this drop down tab and click on measuring probabilities.

392
00:39:53.880 --> 00:39:57.539
Not able to.

393
00:39:57.539 --> 00:40:03.119
Display these probabilities when it comes to circuits that are over 9 Cubans.

394
00:40:03.119 --> 00:40:06.119
So, I would recommend using IBM simulator.

395
00:40:06.119 --> 00:40:09.570
If all you need is the state back there, but.

396
00:40:09.570 --> 00:40:14.400
If you would like to calculate the measurement probabilities, uh, you may have to actually wait 10.

397
00:40:14.400 --> 00:40:17.460
10 minutes and use the quantum computer function.

398
00:40:18.900 --> 00:40:22.500
Now, that we're finished talking about the quantum computer function.

399
00:40:22.500 --> 00:40:25.889
I want to give and honorable mention to another quantum.

400
00:40:25.889 --> 00:40:31.019
A computer simulator called, which was created by Professor road.

401
00:40:31.019 --> 00:40:34.170
At the University of Queensland and Australia.

402
00:40:34.170 --> 00:40:41.730
The simulator contains functions. That would be a great addition to our current simulator of interest such as the.

403
00:40:41.730 --> 00:40:44.849
Circuit printing function and a measure function.

404
00:40:44.849 --> 00:40:48.989
Unfortunately, crack was discontinued in 2014.

405
00:40:48.989 --> 00:40:52.260
And the software is nowhere to be found on the Internet.

406
00:40:52.260 --> 00:40:57.030
However, still influenced me to make these suggestions for.

407
00:40:57.030 --> 00:41:00.449
The simulator that we are currently looking at, and I can.

408
00:41:00.449 --> 00:41:04.920
Correspond here the Z measure function um.

409
00:41:04.920 --> 00:41:08.070
Would be a great measurement function to add to the current simulator.

410
00:41:08.070 --> 00:41:11.489
As well, as this, um, print the quantum circuit function.

411
00:41:11.489 --> 00:41:14.940
1 way you can maybe do the.

412
00:41:14.940 --> 00:41:20.489
See measure function is use random number generator and correspondence to the probabilities that we're working with.

413
00:41:20.489 --> 00:41:30.780
To create a random measure and finally the last suggestion that was to check for a normalized state back there to ensure that the user's not putting in.

414
00:41:30.780 --> 00:41:34.829
In a legal state for the Cubans.

415
00:41:36.900 --> 00:41:40.800
Finally, here's an example of sample code running from.

416
00:41:40.800 --> 00:41:52.199
The quantum simulator, um, although the program, no longer exists there's still some code out there that could inspire to you to recreate the simulator or and improve an existing 1.

417
00:41:55.469 --> 00:42:01.199
Thank you for listening to this presentation here are my references and are there any questions.

418
00:42:01.199 --> 00:42:05.369
Silence.

419
00:42:06.389 --> 00:42:11.909
Questions or anyone.

420
00:42:11.909 --> 00:42:20.159
So, 1 few, so that using Matlab, is it using parallel facilities of bad lab? Because that sort of thing? Matrices looks like it would paralyze enormously.

421
00:42:21.599 --> 00:42:27.869
Even using maybe if you're on, are you online? Just sorry. Sorry?

422
00:42:27.869 --> 00:42:38.550
Uh, I was questioning Matlab does things in parallel on multi core computers that also can call out to a cheap and attach and.

423
00:42:38.550 --> 00:42:42.570
I'm thinking that might make the simulator a lot faster because it's all matrices.

424
00:42:42.570 --> 00:42:49.019
Oh, yeah, that would definitely speed it up. I think the biggest problem was a lot of print treatments involved.

425
00:42:49.019 --> 00:42:55.139
So, if you were to actually optimize the function, and if you only want to find a probability.

426
00:42:55.139 --> 00:43:00.090
You might just want to get rid of creating the graph because if you're making a large cube, it.

427
00:43:00.090 --> 00:43:04.590
Graphic kind of makes a bunch of nonsense as a bunch of lines. You can't.

428
00:43:04.590 --> 00:43:08.159
Uh, characterize like 16000 States.

429
00:43:08.159 --> 00:43:13.199
So, getting rid of all of that to speed up, the computations would definitely help to do.

430
00:43:13.199 --> 00:43:17.219
Okay, thanks any 1 in the class have a question.

431
00:43:17.219 --> 00:43:20.699
I know.

432
00:43:22.050 --> 00:43:28.800
Okay, 1 final thing, maybe if you wrote the guy and down in Australia and ask.

433
00:43:28.800 --> 00:43:38.969
You might have it available I don't know. Actually, I might try that for the report. Actually possibly took it off line because he trying to commercialize it also.

434
00:43:38.969 --> 00:43:42.750
Right. That was the idea. Okay. Well, thanks.

435
00:43:42.750 --> 00:43:51.150
Let me try to run David's YouTube video. Now. We'll see what happens.

436
00:43:52.170 --> 00:43:55.949
Silence.

437
00:43:59.070 --> 00:44:03.389
Just a 2nd, here.

438
00:44:04.500 --> 00:44:16.619
Silence.

439
00:44:17.699 --> 00:44:21.570
Hello today I have.

440
00:44:21.570 --> 00:44:25.650
It's a little clumsy.

441
00:44:30.300 --> 00:44:34.349
There may be better ways to do this, but this is.

442
00:44:34.349 --> 00:44:38.489
A way that actually works, so.

443
00:44:41.610 --> 00:44:49.469
To be talking about the quantum supremacy.

444
00:44:50.909 --> 00:45:01.650
So, 1st of all, what is context? Well, it is the point at which quantum computers can perform calculations that a traditional computer cannot practically perform.

445
00:45:01.650 --> 00:45:05.130
As traditional computers can pretty much.

446
00:45:05.130 --> 00:45:08.489
Calculate any calculations.

447
00:45:08.489 --> 00:45:15.389
Given enough resources and time what we mean by practically performed.

448
00:45:15.389 --> 00:45:20.190
Is that if it is able to do that within a reasonable timeframe?

449
00:45:20.190 --> 00:45:26.730
For human use, the term was coined by John press hill as seen in the picture.

450
00:45:26.730 --> 00:45:31.230
Was a theoretical for the browser any coin determined back in 2012.

451
00:45:31.230 --> 00:45:35.579
For the term he chose the supremacy over other words, such as an advantage.

452
00:45:35.579 --> 00:45:39.420
To emphasize quantum computing speed, relative to traditional computing.

453
00:45:40.679 --> 00:45:46.199
It does not want people to think that corner configure is only slightly better.

454
00:45:46.199 --> 00:45:51.000
And slightly faster than traditional competing you want it to make sure that people understood.

455
00:45:51.000 --> 00:45:54.059
That it was leaps and bounds better and faster.

456
00:45:54.059 --> 00:45:58.469
So is 1 of supremacy important.

457
00:45:58.469 --> 00:46:03.210
Well, yes, and no, so let's give him to why, why not.

458
00:46:03.210 --> 00:46:08.159
So 1st of all, why it is not important. It is just an artificial achievement.

459
00:46:08.159 --> 00:46:12.630
It is just a threshold that people just created.

460
00:46:12.630 --> 00:46:16.260
In order to send sort of some sort of goal.

461
00:46:16.260 --> 00:46:22.800
And there's no real scientific backing to what it means for.

462
00:46:22.800 --> 00:46:26.039
There's no clear definition has to once.

463
00:46:26.039 --> 00:46:30.269
Threshold exactly crossed also.

464
00:46:30.269 --> 00:46:37.440
Whether computer has a chief cons pharmacy it does not mean that the computation that it solves.

465
00:46:37.440 --> 00:46:41.190
Has to be useful, so.

466
00:46:41.190 --> 00:46:45.840
Quantum computer is that significantly passive in the traditional computer?

467
00:46:45.840 --> 00:46:50.579
It might only be doing useless tasks and this isn't really that important.

468
00:46:50.579 --> 00:46:54.690
However, why it is considered.

469
00:46:54.690 --> 00:46:58.739
Important is, because well, even though it is an artificial achievement.

470
00:46:58.739 --> 00:47:05.789
It is a breakthrough for computing, developing a whole new platform for computation.

471
00:47:05.789 --> 00:47:09.329
That performs significantly faster and better than.

472
00:47:10.409 --> 00:47:16.380
Methods we have used for decades is quite an accomplishment and quite something to be proud of.

473
00:47:16.380 --> 00:47:19.769
And also proved that real world on the computers.

474
00:47:19.769 --> 00:47:24.420
Are better at certain applications and.

475
00:47:24.420 --> 00:47:29.789
They can be used for applications in the real world.

476
00:47:29.789 --> 00:47:33.329
With computers that exists it also.

477
00:47:33.329 --> 00:47:39.420
Opens the door for more useful applications to run better on the quantum hardware in the future.

478
00:47:40.590 --> 00:47:48.599
So, while if even if Congress forces teeth with not very useful tasks, it could become useful.

479
00:47:48.599 --> 00:47:53.460
And for tasks could be applied to quantum computer as time goes on.

480
00:47:54.809 --> 00:48:01.829
And also, superficially sort of say, sounds cool and that helps market investment in content technology.

481
00:48:01.829 --> 00:48:06.570
As it is not exactly in a.

482
00:48:06.570 --> 00:48:09.719
Self profitable space yet.

483
00:48:09.719 --> 00:48:18.000
It helps people invest in both companies who achieve as well as the content industry as a whole.

484
00:48:18.000 --> 00:48:21.030
So, is it important.

485
00:48:22.409 --> 00:48:28.380
Yes, for the most part so how we achieve.

486
00:48:28.380 --> 00:48:32.460
Supremacy yet most say, yes, but some do disagree.

487
00:48:32.460 --> 00:48:37.170
Back in October of 2019, Google claimed supremacy.

488
00:48:37.170 --> 00:48:41.699
They, their researchers estimated that the task they performed.

489
00:48:41.699 --> 00:48:47.670
On the computer would take 10000 years for a modern supercomputer to compute.

490
00:48:47.670 --> 00:48:52.349
That be with traditional hardware however.

491
00:48:52.349 --> 00:48:56.579
Ibm disagrees with google's claim, because.

492
00:48:56.579 --> 00:49:00.449
They suggested a method that could reduce the computation.

493
00:49:00.449 --> 00:49:04.469
Time for the task on a supercomputer.

494
00:49:04.469 --> 00:49:09.389
2, 2 and a half days, however, considering it only took.

495
00:49:09.389 --> 00:49:13.019
200 seconds on the content on Google, quantum computer.

496
00:49:13.019 --> 00:49:19.409
It's still ran significantly faster than the best traditional computer theoretically could if claim.

497
00:49:19.409 --> 00:49:26.039
Is correct, and that is still a very significant improvement and speed.

498
00:49:26.039 --> 00:49:30.960
4 computers, so.

499
00:49:30.960 --> 00:49:34.170
How Dave Google reach content supremacy.

500
00:49:34.170 --> 00:49:38.519
Well, they develop the sycamore quantum processor.

501
00:49:38.519 --> 00:49:43.469
Which is their own house processor that uses 54.

502
00:49:44.849 --> 00:49:49.530
Being created a benchmark that runs random complicated quantum circuits.

503
00:49:49.530 --> 00:49:53.519
Until a classical computer could no longer simulate.

504
00:49:53.519 --> 00:49:56.699
Those circuits in a practical amount of time.

505
00:49:58.559 --> 00:50:02.070
Their content computer was able to perform the benchmark.

506
00:50:02.070 --> 00:50:09.989
That they set up within 200 seconds while their researchers estimate that the world's best supercomputer.

507
00:50:09.989 --> 00:50:13.769
Would take 10000 years to get the seminar as long as I said before.

508
00:50:16.409 --> 00:50:21.989
So, what is the 2nd, more processor in that? Google used to claim.

509
00:50:21.989 --> 00:50:25.860
Well, as I said, it's a 54 cubic quantum processor.

510
00:50:25.860 --> 00:50:30.059
It is fully programmed, meaning any.

511
00:50:30.059 --> 00:50:34.320
Classical quantum gates can be used on this platform.

512
00:50:34.320 --> 00:50:37.440
It can be programmed to run on this. Jeff.

513
00:50:38.670 --> 00:50:44.280
But it is laid out in a 2 dimensional grid as seen to the right which I will.

514
00:50:44.280 --> 00:50:50.099
Good for the detail at the end and then each cube it connects to.

515
00:50:50.099 --> 00:50:55.530
Each of the 4 surrounding few bits and this architecture allows for.

516
00:50:55.530 --> 00:50:59.820
All if it's too quickly interact, not with just their.

517
00:50:59.820 --> 00:51:04.139
Directly connected neighbors, both with other connected.

518
00:51:04.139 --> 00:51:10.230
Events throughout the whole system and to the.

519
00:51:10.230 --> 00:51:13.230
The reason why this is advantageous to claim.

520
00:51:13.230 --> 00:51:16.769
Content frustrated because if traditional computer cannot emulate.

521
00:51:16.769 --> 00:51:21.539
This effect efficiently. Google also.

522
00:51:21.539 --> 00:51:27.510
Enhanced operation of 2 ubicated for this processor versus other quantum processors.

523
00:51:27.510 --> 00:51:34.710
As well, as they allowed for better parallelization of different tasks.

524
00:51:34.710 --> 00:51:39.300
Running at the same time, which this is done.

525
00:51:39.300 --> 00:51:42.900
By controlling the interactions between the Cubans.

526
00:51:42.900 --> 00:51:46.590
That are connected to each other and foster reducing errors.

527
00:51:46.590 --> 00:51:50.579
By being influenced into the system.

528
00:51:51.900 --> 00:51:55.320
They also have reduced the cross talk between the Cubans.

529
00:51:55.320 --> 00:51:59.639
As well, as a better calibrated system to account for.

530
00:51:59.639 --> 00:52:05.429
To bed, imperfections and the feedback. So that would also introduce external errors.

531
00:52:05.429 --> 00:52:08.940
Into the system and also.

532
00:52:08.940 --> 00:52:12.960
What is a very beneficial part for the future.

533
00:52:12.960 --> 00:52:18.090
With this processor is that because is compatible with future applications of content error correcting.

534
00:52:18.090 --> 00:52:22.710
This means that this processor can continue to.

535
00:52:22.710 --> 00:52:26.550
And used and will get better with better.

536
00:52:26.550 --> 00:52:31.019
Quantum error correct and so to explain the diagram.

537
00:52:31.019 --> 00:52:38.489
In the top, right? Each of the great axis are Cubans with the wife acts being the inoperable Cuban.

538
00:52:38.489 --> 00:52:44.670
And the blue boxes between each of the cube, it's part of the adjustable couplers.

539
00:52:44.670 --> 00:52:48.840
Which can be controlled to reduce errors between.

540
00:52:52.199 --> 00:52:57.449
So, what is the benchmark? Exactly that Google used to claim content supremacy.

541
00:52:57.449 --> 00:53:04.530
Well, if they use a random circuit that is made up of basic 1 and 2.

542
00:53:04.530 --> 00:53:07.559
The Gates, so.

543
00:53:07.559 --> 00:53:11.880
Any random quantum kit that they could.

544
00:53:11.880 --> 00:53:15.090
Use they combine them in.

545
00:53:15.090 --> 00:53:20.010
Between different cabinets and the multiple times in order to.

546
00:53:20.010 --> 00:53:23.070
Create just a random quantum circuit.

547
00:53:24.420 --> 00:53:29.579
And it's being random ensures that there is not any structure to the system.

548
00:53:29.579 --> 00:53:32.940
Which is very advantageous to claim content supremacy.

549
00:53:32.940 --> 00:53:37.710
Because that means that a traditional algorithm cannot solve.

550
00:53:37.710 --> 00:53:41.730
The.

551
00:53:41.730 --> 00:53:45.150
Circuit efficiently in the bus making it.

552
00:53:45.150 --> 00:53:49.019
Parker and taking much longer time to.

553
00:53:49.019 --> 00:53:58.349
Compute with traditional hardware, then they ran the random circuit on the quantum computer and.

554
00:53:58.349 --> 00:54:02.340
The circuit gives an output that is a strain of events.

555
00:54:02.340 --> 00:54:07.440
They do this multiple times and because of the randomness of the quantum computer.

556
00:54:07.440 --> 00:54:11.070
There isn't always the same output, so.

557
00:54:12.150 --> 00:54:15.480
They look at the probability of distribution.

558
00:54:15.480 --> 00:54:19.980
All of the outputs that are formed after running many.

559
00:54:19.980 --> 00:54:25.320
Multiple paths through the processor.

560
00:54:25.320 --> 00:54:29.760
So, they 1st started with simpler circuits.

561
00:54:29.760 --> 00:54:33.750
You're only 12 of the cube in their processor to determine.

562
00:54:33.750 --> 00:54:36.750
In the 1st place that the method works, they.

563
00:54:36.750 --> 00:54:40.889
Of course, this is necessary just to prove.

564
00:54:40.889 --> 00:54:44.070
Their concept because if.

565
00:54:44.070 --> 00:54:47.460
Their process didn't work and then, of course.

566
00:54:48.869 --> 00:54:52.980
They couldn't claim promised pharmacy because it would not be comfortable.

567
00:54:52.980 --> 00:54:56.550
Or they will not have a comparable comparison between the quantum system.

568
00:54:56.550 --> 00:55:03.150
And the traditional system, so once they were able to establish that.

569
00:55:03.150 --> 00:55:06.449
It worked and that.

570
00:55:06.449 --> 00:55:13.139
With lower levels, it could be simulated on a supercomputer traditional, super computer.

571
00:55:13.139 --> 00:55:16.139
They then used complex circuits.

572
00:55:16.139 --> 00:55:20.880
Using 53 and increasing the number.

573
00:55:20.880 --> 00:55:24.719
Of cycles until the traditional computation.

574
00:55:24.719 --> 00:55:29.280
Became infeasible thus, meaning that they had.

575
00:55:29.280 --> 00:55:32.309
Done a hotel, they have claimed quantum supremacy.

576
00:55:34.469 --> 00:55:38.489
So, what's next to now that quantum pharmacy has been to, you?

577
00:55:38.489 --> 00:55:44.969
Well, the biggest part is finding useful applications to run on these quantum computers.

578
00:55:44.969 --> 00:55:48.030
A random task that they ran.

579
00:55:48.030 --> 00:55:52.199
For the benchmark is not very useful and so.

580
00:55:52.199 --> 00:55:56.130
Of course, applying useful tasks with content and beers.

581
00:55:56.130 --> 00:55:59.429
That can run them significantly after then.

582
00:55:59.429 --> 00:56:02.880
Traditional hardware is a big step forward.

583
00:56:02.880 --> 00:56:09.780
For quantum computers, and then in terms of Google they want to.

584
00:56:09.780 --> 00:56:12.960
Make their process available to others.

585
00:56:12.960 --> 00:56:17.070
For further research, and for the algorithm development for content computers.

586
00:56:18.539 --> 00:56:22.679
They are also investing in developing fault, tolerant, quantum computers.

587
00:56:22.679 --> 00:56:25.679
And which could help with.

588
00:56:27.539 --> 00:56:32.639
Stability of the system and reducing errors and.

589
00:56:32.639 --> 00:56:37.230
Generally next for the content is that they have now.

590
00:56:37.230 --> 00:56:40.619
Clear the path for practical quantum computing.

591
00:56:40.619 --> 00:56:44.130
And now more work just needs to be done and.

592
00:56:44.130 --> 00:56:47.789
Quantum computers can become.

593
00:56:47.789 --> 00:56:51.119
Very practical and very useful in a real world.

594
00:56:51.119 --> 00:56:54.900
Sometimes you, thank you for watching my presentation.

595
00:56:54.900 --> 00:56:59.070
Silence.

596
00:56:59.070 --> 00:57:04.019
Okay, thank you. Dave are you there? Yes, sir. Yes.

597
00:57:04.019 --> 00:57:08.280
Any questions O Connor has a.

598
00:57:08.280 --> 00:57:13.019
Question I got home it's in the chat group.

599
00:57:15.659 --> 00:57:19.889
Can read it and.

600
00:57:19.889 --> 00:57:24.659
Now, how many shots that has to do before they accept.

601
00:57:24.659 --> 00:57:30.360
If they have the correct probability distribution, I imagine it's pretty significant when it comes to measuring performance relative.

602
00:57:30.360 --> 00:57:34.650
The classical computer yeah, so there is.

603
00:57:34.650 --> 00:57:44.340
Not necessarily a, yes, so the correct quantum or the probability distribution that the is.

604
00:57:44.340 --> 00:57:47.489
Yes, is there is a completely random circuit. They just.

605
00:57:47.489 --> 00:57:52.409
Run it yeah, they just run it over and over again.

606
00:57:52.409 --> 00:57:59.610
I don't remember exactly. I don't think I, they specified exactly how many times they did it.

607
00:57:59.610 --> 00:58:04.829
Well, yeah, it was like, it was quite a few times in order to get a.

608
00:58:04.829 --> 00:58:09.059
Very accurate probability distribution.

609
00:58:16.440 --> 00:58:20.250
Anything else okay, thank you.

610
00:58:20.250 --> 00:58:27.119
at one point for the rescue they're presenting next week so we have a reasonable way to do .

611
00:58:27.119 --> 00:58:35.010
Videos, if you, if you're on a slow link, upload your video to YouTube, and then I can play it.

612
00:58:35.010 --> 00:58:40.380
Final presentation for the day is Joseph.

613
00:58:40.380 --> 00:58:45.059
If you're online and I guess if you can share your screen.

614
00:58:46.829 --> 00:58:51.329
Hello can you hear me? I can hear you. I can see you.

615
00:58:51.329 --> 00:58:55.679
All right, let me get the screen share ready just a minute.

616
00:58:59.940 --> 00:59:04.170
Should look good there. All right.

617
00:59:06.000 --> 00:59:12.090
Share screen, and there we go.

618
00:59:12.090 --> 00:59:15.179
Is that visible? And no.

619
00:59:16.380 --> 00:59:25.469
Yes. Okay. All right. So my presentation is on hot bits and what they'll mean for the future quantum Barbara.

620
00:59:26.579 --> 00:59:34.409
So before we get into hot to bits and what they'll mean for the future, we'll.

621
00:59:34.409 --> 00:59:37.440
Let's talk a little bit about the background of.

622
00:59:37.675 --> 00:59:41.724
Some of the mechanics that go into it so the nature of quantum mechanics,

623
00:59:41.755 --> 00:59:47.155
it's 1 of the main drags on progress and developed effective,

624
00:59:47.304 --> 00:59:54.744
quantum computer and hardware in order for Cuba to be reliable and maintain their superposition States without error.

625
00:59:55.050 --> 00:59:59.070
External factors are to be avoided at all costs.

626
00:59:59.070 --> 01:00:02.789
This includes any in all forms of.

627
01:00:02.789 --> 01:00:09.210
Particles potentially interacting with the cube bit, which means that errors in quantum states are.

628
01:00:09.210 --> 01:00:13.500
Basically inevitable if not per preserve that near perfection.

629
01:00:13.500 --> 01:00:20.369
And as you can see here, the result of that is essentially a collapse of the state.

630
01:00:20.369 --> 01:00:29.670
So the primary way to minimize these errors, and these quantum state collapses is to.

631
01:00:29.670 --> 01:00:33.119
Drop the temperature in which the cube it resides.

632
01:00:33.119 --> 01:00:42.090
To avoid to avoid interference, however, the temperature must be dropped to as close to absolute 0T as physically possible.

633
01:00:42.090 --> 01:00:52.949
Which is about minus 273.15 kelvins or? No, no minus 2. 273.15 degrees Celsius. 0T kelvins.

634
01:00:52.949 --> 01:01:00.119
And absolute 0T is sort of like the speed of light where it's.

635
01:01:00.119 --> 01:01:04.139
More so a limit that nature can approach.

636
01:01:04.139 --> 01:01:08.909
Rather than a target for engineers to reach. Exactly.

637
01:01:08.909 --> 01:01:15.239
And as a result of that, essentially our initial objective in trying to.

638
01:01:15.239 --> 01:01:21.780
Save these Cubans from having their quantum states collapsed is to get us close to this temperature.

639
01:01:21.780 --> 01:01:26.909
As possible, and 1 of the ways is to use some.

640
01:01:26.909 --> 01:01:32.010
Refrigeration technology.

641
01:01:32.010 --> 01:01:38.760
However, as it turns out decreasing the temperatures to fractions of a degree.

642
01:01:38.760 --> 01:01:43.739
Above absolute 0T, which is essentially what you would need.

643
01:01:43.739 --> 01:01:50.670
It requires an extraordinary amount of power dedicated to this refrigeration.

644
01:01:50.670 --> 01:01:55.079
And what you can see is a little bit of that going on there and the images.

645
01:01:55.079 --> 01:02:03.599
It's then that refrigeration then needs to be directed at a precise path to cool Cube without interfering with the quantum states.

646
01:02:03.599 --> 01:02:12.360
And this refrigeration comes at a massive cost, which, as you get closer and closer.

647
01:02:12.360 --> 01:02:19.800
To that absolutely. 0T each increment. The costs will grow exponentially because it will get exponentially more difficult.

648
01:02:19.800 --> 01:02:25.980
And this has a variety of consequences on the field of quantum computing as a whole, this cost.

649
01:02:26.635 --> 01:02:38.815
Companies are less likely to want to invest in resources, necessary for quantum computing development and those that do invest have their tests and research limited by hardware and money.

650
01:02:39.594 --> 01:02:43.405
And as a result, a lot of the quantum computing that's done today.

651
01:02:43.710 --> 01:02:48.840
Is done via simulations, or for the cloud this presents a problem in the field.

652
01:02:48.840 --> 01:02:53.340
And in order for quantum computers, to be practical, we need a solution.

653
01:02:54.539 --> 01:02:58.110
So, what could we do.

654
01:02:58.110 --> 01:03:05.519
1 approach to solving the issue is to attempt to lower the cost with some better refrigeration technology.

655
01:03:05.519 --> 01:03:13.409
More efficient right a proposed idea is an event invention called the Nano fridge.

656
01:03:13.409 --> 01:03:21.690
This refrigerator will utilize conductivity to regulate the flow of electrons along. Certain paths of varying resistance is.

657
01:03:21.690 --> 01:03:25.500
And this has its own results to cool hardware.

658
01:03:25.500 --> 01:03:30.329
However, it's several years out 10 to 15.

659
01:03:30.329 --> 01:03:40.440
And the outcome of how effective it would be is still relatively uncertain. And even so the drop and cost is also up in the air.

660
01:03:40.440 --> 01:03:47.340
As it stands with the Nano fridge, the objective temperature, however, remains the same. We're not changing the.

661
01:03:47.340 --> 01:03:51.420
The the temperature, but what if we could raise it?

662
01:03:51.420 --> 01:03:56.460
This is work thought Cubics come in.

663
01:03:56.460 --> 01:04:02.820
Earlier this year a landmark breakthrough was achieved in April 2020.

664
01:04:02.820 --> 01:04:10.559
A team at the University of New South Wales and Disney demonstrated a different approach to the composition of quantum computing ships.

665
01:04:10.559 --> 01:04:17.099
The tech giants currently researching and developing quantum technology, such as IBM and Google.

666
01:04:17.099 --> 01:04:22.710
Are developing superconducting processors such as the 2nd, more quantum processor.

667
01:04:22.710 --> 01:04:27.480
So, to take advantage of the Super conductors, critical temperature.

668
01:04:27.480 --> 01:04:31.199
And to allow for the stability of the quantum states.

669
01:04:31.199 --> 01:04:41.579
The temperature is still stuck at that temperature, close to the usual range of point. 1 kelvins to point 1. 5 kelvins. That's usually the.

670
01:04:41.579 --> 01:04:44.639
The benchmark at the very least.

671
01:04:44.639 --> 01:04:51.150
The university team took a different approach. Their ship was made using silicon.

672
01:04:51.150 --> 01:04:55.860
Which is a semi conducting substance, rather than a superconducting substance.

673
01:04:55.860 --> 01:05:05.610
And this gives more control and their design, it acts essentially as really, really, really small transistors all the way down to the electron level.

674
01:05:05.610 --> 01:05:08.849
Forcing movements of individual electrons.

675
01:05:08.849 --> 01:05:14.340
To spin in different states and moving the upper bound of operable temperatures.

676
01:05:14.340 --> 01:05:19.170
Nearly all the way up to 1.5 kelvins.

677
01:05:19.170 --> 01:05:23.760
And you might be thinking.

678
01:05:23.760 --> 01:05:28.619
1.1 kelvins to 1.5. calvin's. Is that really? It.

679
01:05:28.619 --> 01:05:33.150
Because, you know, that might seem like a really small difference.

680
01:05:33.150 --> 01:05:36.570
And this is true in numerical value, only.

681
01:05:36.570 --> 01:05:45.989
Remember the increase in cost as you go closer and closer to absolute 0T is exponential.

682
01:05:45.989 --> 01:05:51.449
The change from on point 1 kelvins to around 1.5 kelvins.

683
01:05:51.449 --> 01:05:56.010
Lowers costs by several orders of magnitude.

684
01:05:56.010 --> 01:06:01.829
It can change things from millions of dollars to thousands of dollars.

685
01:06:01.829 --> 01:06:06.690
And this has its own set of of results in the future.

686
01:06:06.690 --> 01:06:16.619
For quantum computer development, since feasibility, of course, is the roadblock that prevents a lot of companies from investing in the 1st place.

687
01:06:17.789 --> 01:06:21.420
This will likely mean some very good things looking into the future.

688
01:06:21.420 --> 01:06:26.880
And the breakthroughs on its own.

689
01:06:26.880 --> 01:06:37.050
Have its own positive effects Silicon carbide's, as it turns out, has its own abilities as a result that were discovered as a result.

690
01:06:37.050 --> 01:06:40.829
Of its use in Cuba.

691
01:06:40.829 --> 01:06:44.070
Such as being used and things like.

692
01:06:44.070 --> 01:06:50.159
Matt magnesium meters might be magnet. Now it's not Magnus. It's probably meters.

693
01:06:50.159 --> 01:06:57.690
And other different types of gadgets, there's biosensors, there's Internet tech that's been used.

694
01:06:57.690 --> 01:07:01.380
By quantum computing, you can see some examples here.

695
01:07:01.380 --> 01:07:10.530
And the chemical compound itself has found its way into other fields as a result of this own breakthrough.

696
01:07:10.530 --> 01:07:15.750
On top of what it's already done for quantum computing on its own.

697
01:07:15.750 --> 01:07:20.340
And looking into the future.

698
01:07:20.340 --> 01:07:25.469
This was only this year that this was discovered, and it's already had.

699
01:07:25.469 --> 01:07:31.139
It's already made a massive splash in the field of quantum computing.

700
01:07:31.139 --> 01:07:37.769
And adjacent fields and new discoveries are essentially inevitable.

701
01:07:37.769 --> 01:07:42.420
There will be more discoveries and hot Cubics that will build on this.

702
01:07:42.420 --> 01:07:48.059
Potentially raising the temperature even more, which will, of course, lower the cost even more.

703
01:07:48.059 --> 01:07:53.670
And with lower costs, more companies are likely to throw their hats into that quantum ring.

704
01:07:53.670 --> 01:07:58.829
Which will mean more discoveries on top of the other discoveries.

705
01:07:58.829 --> 01:08:03.030
And continued development into the field, the quantum computer.

706
01:08:03.030 --> 01:08:14.070
And that's all. Oh, thank you. Very much. Anyone have any questions.

707
01:08:16.409 --> 01:08:22.500
You know, anything more, but what's happening down new South Wales I mean, they're doing this nice thing and doing other cool stuff.

708
01:08:22.500 --> 01:08:32.069
So, the research team that was that that ran these experiments and came along with.

709
01:08:32.069 --> 01:08:35.250
With the discovery of hot Cuba.

710
01:08:35.250 --> 01:08:38.460
They're led by a professor.

711
01:08:38.460 --> 01:08:44.100
Andrew Zach.

712
01:08:44.100 --> 01:08:48.899
And.

713
01:08:48.899 --> 01:08:55.439
The research team, that's more or less. A lot of what I was able to.

714
01:08:55.439 --> 01:09:04.289
Figure out in terms of what's actually happening in the new, the new South Wales area, but it is really nice to see how.

715
01:09:04.289 --> 01:09:10.560
There are some research teams that are all around the world essentially, and they aren't limited to.

716
01:09:10.560 --> 01:09:16.859
Just a few stations, I suppose, in in the United States.

717
01:09:16.859 --> 01:09:21.750
To have more backgrounds and more backgrounds and more people.

718
01:09:21.750 --> 01:09:28.890
Around the world who are working on quantum computing and making these discoveries like this, this wasn't a discovery that was made.

719
01:09:28.890 --> 01:09:36.060
By someone you would expect, you know, it's not made by a large company, or a massive University at.

720
01:09:36.060 --> 01:09:39.779
Like, the heart of a big city in the U. S. this was.

721
01:09:39.779 --> 01:09:45.239
At a university in New South Wales and what that shows to me is that.

722
01:09:45.239 --> 01:09:52.890
Some of the big quantum computing discoveries of the future might come from places that we at least expect.

723
01:09:52.890 --> 01:09:56.250
It looks like there is a question in the chat and John.

724
01:09:56.250 --> 01:10:03.270
Yeah, How's the cube realized in Silicon carbide's? Is it a trans bond in the superconducting Cuba?

725
01:10:03.270 --> 01:10:06.960
Or is it something different? So the what if that works?

726
01:10:06.960 --> 01:10:14.010
Is it's sort of like, um, if there's another type, which is very similar, it's called silicone spin.

727
01:10:14.010 --> 01:10:20.399
And they're different in superconducting Cubans because.

728
01:10:20.399 --> 01:10:26.130
Uh, so in superconducting Cuba, you work sort of.

729
01:10:26.130 --> 01:10:32.880
More with essentially an LLC oscillator circuit if you're if you've taken.

730
01:10:32.880 --> 01:10:37.140
Circuits then you might be familiar with the way that was also.

731
01:10:37.140 --> 01:10:45.420
Uh, and you're trying to use low energy levels in order to make the 0T and the 1 of the cube. Now, of course, this is just.

732
01:10:45.420 --> 01:10:50.520
Emulating the way that it works, it's not actual, you know, a little inductors and capacitors.

733
01:10:50.520 --> 01:10:54.689
And that's sort of what the superconducting circuit would would do.

734
01:10:54.689 --> 01:10:59.159
Whereas with the silk on carbide's.

735
01:10:59.159 --> 01:11:05.609
Cuba, you're doing something different, you're essentially taking the flow of electrons and.

736
01:11:05.609 --> 01:11:09.060
Mimicking mimicking components.

737
01:11:09.060 --> 01:11:15.510
That are not linear and so the result of that is being able to.

738
01:11:15.510 --> 01:11:20.880
Have more control over the flow and just the general movement of electrons.

739
01:11:20.880 --> 01:11:24.449
And you're essentially able to.

740
01:11:24.449 --> 01:11:31.590
Get the electrons to move in place to spin up or to spin down.

741
01:11:31.590 --> 01:11:37.920
And this I'm still sort of looking a little bit into it, but.

742
01:11:37.920 --> 01:11:42.899
This I believe is 1 of the things that contributes to allowing the temperature to go up by a bit.

743
01:11:45.119 --> 01:11:57.420
So, did the U. S. W team specifically set out to make hot bits or is it something they came across while doing something else? I think they were.

744
01:11:57.420 --> 01:12:03.449
Going after trying to because this is an entirely new chip design.

745
01:12:03.449 --> 01:12:06.510
So, they were likely trying to.

746
01:12:06.510 --> 01:12:10.470
I experiment with different types of chips and.

747
01:12:10.470 --> 01:12:16.409
They likely came, they came across the discovery of hot.

748
01:12:16.409 --> 01:12:22.079
As a result of successfully making a quantum chip.

749
01:12:22.079 --> 01:12:26.460
From different hardware and trying new things.

750
01:12:26.460 --> 01:12:32.159
And this is, it's sort of.

751
01:12:32.159 --> 01:12:40.470
It was sort of an experiment that came from trying to use different types of hardware to make quantum chips.

752
01:12:40.470 --> 01:12:43.859
And 1 of the byproducts of that was.

753
01:12:43.859 --> 01:12:49.109
We can store these cube at a slightly higher temperature.

754
01:12:49.109 --> 01:12:54.090
And reduced reduce costs massively.

755
01:12:54.090 --> 01:12:57.779
Oh, thank you. Anyone else have any questions.

756
01:13:00.119 --> 01:13:06.060
Okay, well I have a good weekend and.

757
01:13:06.060 --> 01:13:11.850
See, you Monday? Oh, 1st, 1 little engineering in the news story.

758
01:13:13.079 --> 01:13:24.029
Yeah, you all know about DARPA. Okay. And connected computers with phones autonomy and so on autonomous vehicles.

759
01:13:24.029 --> 01:13:28.949
Darpa, apparently funded Madonna, the company that did the vaccine.

760
01:13:28.949 --> 01:13:33.390
So, they're doing all sorts of crazy stuff.

761
01:13:33.390 --> 01:13:41.789
Okay, I have a good weekend. I'll stay around in for a few minutes in case. Anyone has questions other than that.

762
01:13:41.789 --> 01:13:47.789
See, you Monday and again, if anyone would like to have me play your video, I'd be glad to.

763
01:13:47.789 --> 01:13:54.569
The tested method is uploaded to YouTube post us on piaza.

764
01:13:54.569 --> 01:14:05.699
Don't don't for some crazy reason. I have to spend some time learning sending me the video file doesn't is not guaranteed to work, but to you to.

765
01:14:05.699 --> 01:14:08.819
Silence.

766
01:14:13.949 --> 01:14:17.850
Silence.

767
01:14:21.689 --> 01:14:28.920
Should we email you our presentation slides? Like we did for Homer 3? That would be nice. Thank you. Yes.

768
01:14:28.920 --> 01:14:37.560
Okay, I have a record of them, or you could just keep them for your final project and put them as part of your, when you hand to write up in just.

769
01:14:37.560 --> 01:14:47.939
Thing at once. Yeah. It's neat. Or if everything's at 1, pick your favorite format. So not naturally 1 files and zip them up or tar them up or something.

770
01:14:47.939 --> 01:14:54.329
Okay, I might do that just so I can have 1 more slide that has all my references.

771
01:14:54.329 --> 01:14:58.289
Okay, cool. Yeah, thank you. Professor.

772
01:14:58.289 --> 01:15:00.600
Yeah, thank you.