1 00:00:00,000 --> 00:00:18,519 *36C3 Preroll music* 2 00:00:18,519 --> 00:00:23,300 Herald: So hello and welcome to a quantum computing talk by the Andreas [Dewes], who 3 00:00:23,300 --> 00:00:27,050 gave a talk exactly five years ago and it's almost exactly five years ago, it's 4 00:00:27,050 --> 00:00:33,580 like one year and two or three hours. And then he gave a talk at 31C3 about quantum 5 00:00:33,580 --> 00:00:37,710 computing titled "Let's Build a Quantum Computer". And I think back then we 6 00:00:37,710 --> 00:00:41,080 basically had just found out that Google was planning to partner with the 7 00:00:41,080 --> 00:00:46,459 University of California at Santa Barbara to try to build a quantum computer. Of 8 00:00:46,459 --> 00:00:50,140 course, now we're five years later, we've had a lot of developments, I think, in the 9 00:00:50,140 --> 00:00:54,510 field. We've had some big announcements by Google and other groups. And Andreas has 10 00:00:54,510 --> 00:00:58,470 now come back to give us an update. So please welcome him to the stage. 11 00:00:58,470 --> 00:01:05,310 *Applause* 12 00:01:05,310 --> 00:01:10,010 Andreas: Okay. Hi, everyone, so I'm very happy to be here again. After five years 13 00:01:10,010 --> 00:01:16,199 of giving the first version of this talk. My motivation for given this talk is quite 14 00:01:16,199 --> 00:01:22,880 simple. I was often so I did my PHD on exponential quantum computing from 2009 to 15 00:01:22,880 --> 00:01:28,400 2012. I left that field afterwards to work in industry, but always people would come 16 00:01:28,400 --> 00:01:33,120 to me and ask, hey, Andreas, did you see like this new experiment. Did you see, you 17 00:01:33,120 --> 00:01:37,850 can like use quantum computers on Amazon's cloud now? Did you see, like Google has 18 00:01:37,850 --> 00:01:42,320 this new quantum thing? This is really working? Can we use quantum computers yet? 19 00:01:42,320 --> 00:01:47,950 Why are you not working on this? And I couldn't I couldn't really answer the 20 00:01:47,950 --> 00:01:53,960 question. So I said, OK, I want to go back to this and find out what happened in the 21 00:01:53,960 --> 00:01:57,900 last five years since I finished my PhD. What kind of progress was made in the 22 00:01:57,900 --> 00:02:02,110 field? And do we actually have quantum computers today that are working already 23 00:02:02,110 --> 00:02:11,999 or are we not yet quite just there? So we want to do it like this. I want to first 24 00:02:11,999 --> 00:02:16,020 give you a short introduction to quantum computing. So just that we have a common 25 00:02:16,020 --> 00:02:19,590 understanding of how that works and why it's interesting. Then I will show you a 26 00:02:19,590 --> 00:02:24,290 small example of experimental quantum speed up. Notably the work I did with my 27 00:02:24,290 --> 00:02:29,290 colleagues in Saclay during my PhD thesis. Then we discuss some of the 28 00:02:29,290 --> 00:02:33,569 challenges and problems, why we were not able to build a real quantum computer back 29 00:02:33,569 --> 00:02:38,269 then. And I will discuss some approaches that have come up since then. That would 30 00:02:38,269 --> 00:02:42,750 basically allow us to do that eventually. And then we'll, of course, discuss 31 00:02:42,750 --> 00:02:47,489 Google's recent experiment in collaboration with the University of Santa 32 00:02:47,489 --> 00:02:52,540 Barbara, where they showed basically a very impressive quantum computing system 33 00:02:52,540 --> 00:02:57,849 with 53 Qubits. We will look exactly to try to understand what they did there and 34 00:02:57,849 --> 00:03:02,090 see if that's really like a quantum computer in the in the real sense already 35 00:03:02,090 --> 00:03:05,349 or if there's still something missing. And in the end, of course, I will try to give 36 00:03:05,349 --> 00:03:13,349 you another small outlook to see what we can expect in the coming years. So in 37 00:03:13,349 --> 00:03:17,059 order to talk about quantum computing, we need to first talk about classical 38 00:03:17,059 --> 00:03:21,599 computing just a little bit. You might know that classical computers, they work 39 00:03:21,599 --> 00:03:27,099 with bits, so zeros and ones. They store them in so-called registers. This here for 40 00:03:27,099 --> 00:03:33,549 example of like a bit register. Of course, the bits themselves are not very 41 00:03:33,549 --> 00:03:38,989 interesting. But we have to do stuff with them so we can compute functions over 42 00:03:38,989 --> 00:03:43,959 those bit registers. That's what like modern CPU is doing in a simplified way, 43 00:03:43,959 --> 00:03:48,590 of course. So we take some input, but register values, we compute some function 44 00:03:48,590 --> 00:03:57,869 over then and then we get an output value. So a very simple example would be a search 45 00:03:57,869 --> 00:04:02,269 problem. I would discuss this because later we will also see in the experiment 46 00:04:02,269 --> 00:04:05,889 how we can use a quantum computer to solve this. So I just want to motivate why this 47 00:04:05,889 --> 00:04:10,619 kind of problem can be interesting. And it's a very silly search function. So it 48 00:04:10,619 --> 00:04:16,060 takes two bits as inputs and it returns one bit as an output, indicating, whether 49 00:04:16,060 --> 00:04:20,799 the input bits are the solution to our search problem or not. And you could 50 00:04:20,799 --> 00:04:24,280 imagine that we have a very, very complicated function here. So, for 51 00:04:24,280 --> 00:04:29,120 example, a function that calculates the answer to life, the universe and 52 00:04:29,120 --> 00:04:33,710 everything, while not a complete answer, but only the first two bits. So really 53 00:04:33,710 --> 00:04:39,600 complicated to implement and very costly to execute. So we might think that it 54 00:04:39,600 --> 00:04:43,760 might take like millions of years to run this function once on our inputs. And so 55 00:04:43,760 --> 00:04:47,740 we want to find the right solution to that function with as few function calls as 56 00:04:47,740 --> 00:04:56,800 possible, of course. Overall, there are four possibilities, so for input states, 00 57 00:04:56,800 --> 00:05:02,320 01 10 and 11 that we can apply our function to and only for one of these 58 00:05:02,320 --> 00:05:09,140 states. The 01 state, because the answer is 42. So that's 0 times 1 plus to plan 2 59 00:05:09,140 --> 00:05:13,780 plus some other stuff. So the first two bits are 0 1 for this for a value, the 60 00:05:13,780 --> 00:05:19,570 function returns a 1 for all of the other values, the function returns 0. Now let's 61 00:05:19,570 --> 00:05:25,280 think about how we can implement a central search function and in principle, if we 62 00:05:25,280 --> 00:05:28,880 don't know anything about the function. So we can imagine it's so complicated that we 63 00:05:28,880 --> 00:05:34,590 can't do any optimizations. We don't know where to look. So we have to really try 64 00:05:34,590 --> 00:05:39,130 each of these values in sequence. And for this we can have a simple algorithm so we 65 00:05:39,130 --> 00:05:45,520 can start initializing out our a bit register with 00 value. Then we can call 66 00:05:45,520 --> 00:05:50,860 the function on that register. We can see what the result is. In this case, the 67 00:05:50,860 --> 00:05:54,900 result would be zero. If the result would be 1, then we know, okay, we have found 68 00:05:54,900 --> 00:05:59,480 our solution so we can stop our algorithm. But in this case, the result is zero. So 69 00:05:59,480 --> 00:06:05,340 we can just go back to the left value and to the left step and increase the register 70 00:06:05,340 --> 00:06:11,360 value, go to 0 1 and try again. And in the worst case, depending if you're optimistic 71 00:06:11,360 --> 00:06:15,460 or not, we have to do this three or four times. So if you want to really be sure 72 00:06:15,460 --> 00:06:19,570 that we find the right answers, we have to do it four times in the worst case. And 73 00:06:19,570 --> 00:06:23,210 this is sort of say the time complexity or the computational complexity of the 74 00:06:23,210 --> 00:06:27,150 search. You know, if you imagine that in our algorithm, the most expensive 75 00:06:27,150 --> 00:06:32,550 operation is really calling this function F, then the calling time of the complexity 76 00:06:32,550 --> 00:06:37,380 of calling this function will be what dominates the complexity of our algorithm. 77 00:06:37,380 --> 00:06:42,120 And in this case, the complexity is very similar, simple here because it's linear 78 00:06:42,120 --> 00:06:46,540 in the number of the search space. So if you have n states, for example, in our 79 00:06:46,540 --> 00:06:50,480 examples, we have four different input spaced states. We also need to evaluate 80 00:06:50,480 --> 00:06:54,450 the function four times. So and please keep this graph in mind because we're 81 00:06:54,450 --> 00:06:58,780 gonna revisit that later a bit to see,if we can do better with a different paradigm 82 00:06:58,780 --> 00:07:03,300 of computing. And so classically. This is really the best we can do for the search 83 00:07:03,300 --> 00:07:06,580 problem here because we don't know anything else about the function that 84 00:07:06,580 --> 00:07:12,560 would allow us to optimize that further. But now the interesting thing is that we 85 00:07:12,560 --> 00:07:17,580 might imagine that we don't use classical computing for solving our problem. And in 86 00:07:17,580 --> 00:07:23,270 fact, the discipline that we call quantum computing was kind of like inspired by 87 00:07:23,270 --> 00:07:29,510 lecturer or like a seminar of Richard Feynman, who thought about, how it would be 88 00:07:29,510 --> 00:07:34,820 possible to similar and or if it would be possible to simulate quantum systems on a 89 00:07:34,820 --> 00:07:39,380 classical computer. A Turing machine, if you want. And he found that because 90 00:07:39,380 --> 00:07:43,300 quantum mechanics is so complicated for classical computers that it is not 91 00:07:43,300 --> 00:07:47,040 possible to do that efficiently, but that if you would use the laws of quantum 92 00:07:47,040 --> 00:07:51,620 mechanics themselves to make a computer like quantum computer, then it would be 93 00:07:51,620 --> 00:07:55,850 possible to simulate this quantum systems and just kind of like sparked this whole 94 00:07:55,850 --> 00:08:00,020 idea of using quantum mechanics to do computation. And in the following years, 95 00:08:00,020 --> 00:08:04,570 they were not only as solutions found for simulating quantum systems, which such a 96 00:08:04,570 --> 00:08:08,940 quantum computer, but also for solving other not related problems to quantum 97 00:08:08,940 --> 00:08:18,610 computing. So like search problems or factorization problems, for example. And 98 00:08:18,610 --> 00:08:22,540 quantum computers can do, can do computation faster than classical 99 00:08:22,540 --> 00:08:27,010 computers, because they have several differences in how they work. So one of 100 00:08:27,010 --> 00:08:32,130 the key differences here is superposition, which means that if you use a quantum 101 00:08:32,130 --> 00:08:37,070 computer, instead of a classical computer, we cannot only load a single register 102 00:08:37,070 --> 00:08:42,620 value into our bit register. So for example, the first value of only zeros. 103 00:08:42,620 --> 00:08:49,041 But instead we can kind of load all of the possible state values and it at once or in 104 00:08:49,041 --> 00:08:54,949 parallel. And this so-called quantum state or quantum superposition state where each 105 00:08:54,949 --> 00:08:59,699 of these values here has an amplitude which is shown on the left, that is 106 00:08:59,699 --> 00:09:04,559 basically a complex number that relates them to the other Qubit, to other states 107 00:09:04,559 --> 00:09:11,360 and ups. If you have like for example, n-Qubits, then the total number of Qubits 108 00:09:11,360 --> 00:09:15,509 states can be very large 2 to the power of N. So we can imagine that if you 109 00:09:15,509 --> 00:09:19,730 have a large Qubit quantum quantum bit register, then your number of quantum 110 00:09:19,730 --> 00:09:26,290 states can be really, really large and this can be very powerful for computation. 111 00:09:26,290 --> 00:09:31,550 So in the rest of the talk, we gonna just indicate this by like showing the register 112 00:09:31,550 --> 00:09:37,129 as like a small rectangle to indicate that it's not only a single value in there, but 113 00:09:37,129 --> 00:09:41,309 that we have a superposition values of all the possible input values to our function, 114 00:09:41,309 --> 00:09:46,579 for example. And there is a condition and so called normalization condition that 115 00:09:46,579 --> 00:09:51,110 puts some constraints on these amplitude. Because the sum of the squares of the 116 00:09:51,110 --> 00:09:55,230 absolute values of these amplitude needs to sum to one, which basically means that 117 00:09:55,230 --> 00:09:59,029 the entire the probability of each of these of all of these states together 118 00:09:59,029 --> 00:10:07,439 needs to be 100 percent. So. And this is the first ingredient that makes quantum 119 00:10:07,439 --> 00:10:12,300 computers interesting for computation because we can basically implement any 120 00:10:12,300 --> 00:10:16,259 classical function that we can also run on a classical computer, on a quantum 121 00:10:16,259 --> 00:10:21,620 computer. The difference is that we cannot only run it for one value at a time, but 122 00:10:21,620 --> 00:10:25,260 we can call it can run it down on a superposition of all possible input 123 00:10:25,260 --> 00:10:29,140 values. So if you want, you have like this massive paralellyzation where you run you 124 00:10:29,140 --> 00:10:33,910 off computation on all possible inputs at once and also calculate and all of the 125 00:10:33,910 --> 00:10:39,260 possible output values. And that sounds, of course, very cool and very useful. 126 00:10:39,260 --> 00:10:44,040 There's a catch that we will discuss later. So it's not as easy as that. But 127 00:10:44,040 --> 00:10:50,110 this is one step off like the power that makes quantum computing interesting. The 128 00:10:50,110 --> 00:10:54,360 next thing that is different is that we can on a quantum computer, not only run 129 00:10:54,360 --> 00:10:58,619 classical gates or classical functions, but we can also run so-called quantum 130 00:10:58,619 --> 00:11:04,170 gates. And the quantum gates, they're different in respect to the classical 131 00:11:04,170 --> 00:11:09,490 functions because they cannot only like classical operations like and or or on 132 00:11:09,490 --> 00:11:15,000 like two Qubits in a predictable way. But they can kind of like act on the whole 133 00:11:15,000 --> 00:11:20,350 Qubit state at once and also create so- called entangled states which are really 134 00:11:20,350 --> 00:11:25,369 weird quantum states where we can't really separate the state of one Qubit from the 135 00:11:25,369 --> 00:11:29,100 state of or other Qubits. So it's kind of like if we want to try to make a small 136 00:11:29,100 --> 00:11:32,860 change to one of two Qubits in our system, we also changing other Qubits 137 00:11:32,860 --> 00:11:38,110 there. So we can never like separate the bits, the Qubits out like we can with a 138 00:11:38,110 --> 00:11:42,269 classical computer. And this is another resource that we can use in quantum 139 00:11:42,269 --> 00:11:49,970 computing to solve certain problems faster than we could with a classical computer. 140 00:11:49,970 --> 00:11:54,939 Now, the catch, as I said, is that we, of course, do not, we do not want to only 141 00:11:54,939 --> 00:12:00,670 make computation with our Qubits, Qubits register, but we also want to read out the 142 00:12:00,670 --> 00:12:05,580 result of our computation. And if we try that. So we make like computation. And 143 00:12:05,580 --> 00:12:10,339 when we want to measure the state of our quantum register, we have a small problem 144 00:12:10,339 --> 00:12:15,240 because, well, the measurement process is actually quite complicated. But in a very 145 00:12:15,240 --> 00:12:19,699 simplified way, you can just imagine, that God is trying some dice here. And then if 146 00:12:19,699 --> 00:12:23,709 we have a quantum vector, a quantum state vector that has like this amplitude on the 147 00:12:23,709 --> 00:12:28,730 left. So a one to a n. And then we will pick. He or she would pick a state 148 00:12:28,730 --> 00:12:34,170 randomly from the possible states. And the probability of getting a given state as a 149 00:12:34,170 --> 00:12:38,860 result is proportional, as is that before to the square of the absolute value of the 150 00:12:38,860 --> 00:12:44,099 amplitude. So that means we can perform computation on all of the possible input 151 00:12:44,099 --> 00:12:48,379 states of our function. But when we read out the result, we will only get one of 152 00:12:48,379 --> 00:12:53,990 the possible results. So it's kind of like destroys at the first glimpse the utility 153 00:12:53,990 --> 00:12:57,870 of quantum computing because we can do like computation on all states in 154 00:12:57,870 --> 00:13:02,029 parallel, but we cannot read out the result. So not a very interesting computer 155 00:13:02,029 --> 00:13:08,380 because we can't learn about the output. So to say or not easily at least. But it 156 00:13:08,380 --> 00:13:14,759 turns out that there's actually a way of still using quantum computing to be faster 157 00:13:14,759 --> 00:13:19,420 than a classical computer. And the first kind of practical algorithm for a search 158 00:13:19,420 --> 00:13:24,090 problem, notably the search problem that we discussed before, was given by Love 159 00:13:24,090 --> 00:13:30,490 Grover, who was a researcher at the Bell Labs, and who found the Grover algorithm 160 00:13:30,490 --> 00:13:36,619 that is named after him. That's basically a search algorithm which can prove it can, 161 00:13:36,619 --> 00:13:40,980 as we will see, solved the search problem that we have in a much more efficient way 162 00:13:40,980 --> 00:13:46,879 than any classical computer could. And in my opinion, it's still one of the most 163 00:13:46,879 --> 00:13:52,529 beautiful quantum algorithms because it's very simple and it's very powerful and 164 00:13:52,529 --> 00:13:56,140 does also prove, unlike for other algorithms like the factorization 165 00:13:56,140 --> 00:14:01,489 algorithms from Shor that the Grover algorithm can be will be faster always 166 00:14:01,489 --> 00:14:06,350 than any classical computer classical algorithm. So in my opinion, it's a very 167 00:14:06,350 --> 00:14:11,889 nice example of really a quantum algorithm that is more powerful than a classical 168 00:14:11,889 --> 00:14:20,609 one. Let's see how it works. So they're three steps again and the algorithm. First 169 00:14:20,609 --> 00:14:26,769 we initialize our Qubit register, our state vector to a superposition of the 170 00:14:26,769 --> 00:14:33,730 four possible output values, so 00 01 10 and 10, again, all with equal 171 00:14:33,730 --> 00:14:40,440 probability in this case, zero amplitude. Then we evaluate the function on this 172 00:14:40,440 --> 00:14:44,619 input state here and what the function then does. So we made some special 173 00:14:44,619 --> 00:14:50,490 encoding here that basically marks the solution of our problem by changing the 174 00:14:50,490 --> 00:14:54,980 sign of the amplitude of the corresponding state. We can see that in the output state 175 00:14:54,980 --> 00:15:01,449 here, the 01 state has a sign which is negative, which means that it's the 176 00:15:01,449 --> 00:15:06,300 solution of the problem that we search. Still, if we were to the read out now 177 00:15:06,300 --> 00:15:10,410 directly, we wouldn't be able to learn anything about the solution, because as 178 00:15:10,410 --> 00:15:14,899 you can see, the amplitude is still equal for all of the four states. So if you 179 00:15:14,899 --> 00:15:20,050 would make a read out now, we would only get like one of the four possible states 180 00:15:20,050 --> 00:15:24,490 at random so we wouldn't learn anything with a hundred percent probability about 181 00:15:24,490 --> 00:15:29,350 the solution of our problem. In order to do that, we need to apply another step to 182 00:15:29,350 --> 00:15:35,249 so-called Grover or Diffusion, diffusion operator, which now takes this phase 183 00:15:35,249 --> 00:15:39,360 difference or the sign difference between these individual quantum states and 184 00:15:39,360 --> 00:15:45,310 applies a quantum operator to that, that basically transfers the amplitude from all 185 00:15:45,310 --> 00:15:49,129 of the states that are not a solution to a problem to the states that is the 186 00:15:49,129 --> 00:15:54,730 solution. And for on this case, with two Qubits here and with four possible values, 187 00:15:54,730 --> 00:15:58,959 there's only one step we need. And after executing that, you can see that now the 188 00:15:58,959 --> 00:16:04,119 amplitude of our solution state is one versus very. But the amplitude of the 189 00:16:04,119 --> 00:16:09,579 other states is all zero. So that's great, because now we can just do a Qubit 190 00:16:09,579 --> 00:16:14,389 measurement and then we will have a hundred percent probability find a 191 00:16:14,389 --> 00:16:18,839 solution to our search problem. And that's where kind of like the magic of quantum 192 00:16:18,839 --> 00:16:24,339 mechanics shows, because you can evaluate its function only once. So remember that 193 00:16:24,339 --> 00:16:27,920 in the first step we only call the search function once of all of the values in 194 00:16:27,920 --> 00:16:33,549 parallel. So from the computational complexity, we are much lower than the 195 00:16:33,549 --> 00:16:38,170 classical algorithm, but still we are able to find 100 percent position in this case 196 00:16:38,170 --> 00:16:45,209 to see which state is the solution to our search problem. So and that's working not 197 00:16:45,209 --> 00:16:49,799 only for the case of two Qubits, but also with larger Qubit registers. So for 198 00:16:49,799 --> 00:16:54,529 example, if you would take 10 Qubits, you would need to execute a few more of these 199 00:16:54,529 --> 00:16:59,549 steps, two and three. So instead of one iteration, you would need 25 iterations, 200 00:16:59,549 --> 00:17:04,760 for example, here, which is still much better than the 1024 iterations that you 201 00:17:04,760 --> 00:17:09,010 would need if you would really look into every possible solution of the function in 202 00:17:09,010 --> 00:17:15,870 the classical algorithm. So the speed up here is very good for, so to say, all of 203 00:17:15,870 --> 00:17:21,630 the like. It's quadratical for the solution space. And if you like, look at 204 00:17:21,630 --> 00:17:28,550 the complexity plot again, we can now compare our classical algorithm with the 205 00:17:28,550 --> 00:17:34,130 quantum algorithm on the Grover search. And as you can see, the time complexity 206 00:17:34,130 --> 00:17:39,690 or the number of variations of F that we need is only a square root of N, where N 207 00:17:39,690 --> 00:17:44,550 is the size of the search space, which shows that that we have really a 208 00:17:44,550 --> 00:17:48,420 speed advantage hier of the quantum computer versus the classical computer. 209 00:17:48,420 --> 00:17:54,180 And nice thing is the larger our search space becomes, the more dramatic our speed 210 00:17:54,180 --> 00:17:58,550 up will be, because for example, for a search space with one million 211 00:17:58,550 --> 00:18:02,510 elements. We will only have to evaluate the search function 1000 times instead of 212 00:18:02,510 --> 00:18:14,480 one million times. So that's quite so to say a speed up in that sense. Now, how can 213 00:18:14,480 --> 00:18:20,260 we build a system that realizes this quantum algorithm? Here, I show on the 214 00:18:20,260 --> 00:18:25,440 quantum processor that I built with my colleagues at the Saclay during my PhD. So 215 00:18:25,440 --> 00:18:28,730 if you want more information about this, you should check out my last talk. I just 216 00:18:28,730 --> 00:18:33,210 want to go briefly over the different aspects here. So we use a so called 217 00:18:33,210 --> 00:18:39,600 superconducting Qubits, transmit Qubits for realizing our quantum computer. A 218 00:18:39,600 --> 00:18:44,980 quantum processor. You can see the chip here on the top. It's about one centimeter 219 00:18:44,980 --> 00:18:49,870 across. You can see the two Qubits in the middle. The other, like snake like 220 00:18:49,870 --> 00:18:53,920 structures are coupling a wave guides where we can manipulate the Qubits using 221 00:18:53,920 --> 00:18:58,740 microwaves. So we use frequencies that are similar to the ones that are used by 222 00:18:58,740 --> 00:19:03,870 mobile phones to manipulate and read out our Qubits. And if you look in the 223 00:19:03,870 --> 00:19:09,240 middle, you can see the red area, which contains the Qubit, each Qubit itself. And 224 00:19:09,240 --> 00:19:13,000 then there's another zoom in here, which contains the actual qubit structure, which 225 00:19:13,000 --> 00:19:18,410 is just some two layers of aluminum that have been placed on top of each other and 226 00:19:18,410 --> 00:19:23,040 which create, when they are cooled, to a very low temperature, a so-called 227 00:19:23,040 --> 00:19:27,660 superconducting state, where we can use the superconducting face between these two 228 00:19:27,660 --> 00:19:34,400 values, layers to indicate to to realize our Qubits. There's also coupler in the 229 00:19:34,400 --> 00:19:39,120 middle. So this green element that you see, which allows us to run quantum gate 230 00:19:39,120 --> 00:19:49,110 operations between the two Qubits. To use that in practice, we need to put this in a 231 00:19:49,110 --> 00:19:53,021 delusion crisis that which is really like just a very fancy refrigerator, you could 232 00:19:53,021 --> 00:19:59,380 say, you cool it down to about 10 milli K. So very low temperature just above the 233 00:19:59,380 --> 00:20:04,300 absolute zero temperature. You can see the sample holder here on the left side with 234 00:20:04,300 --> 00:20:08,270 the chip mounted to it. So this whole thing is put in the delusion fridge and 235 00:20:08,270 --> 00:20:13,000 it's cool down to the temperature. And then we can, as I said, manipulated by 236 00:20:13,000 --> 00:20:18,950 using his microwave transmission lines. And what we did is we implemented the 237 00:20:18,950 --> 00:20:24,000 Grover search for the two Qubits. So we ran this algorithm that I discussed 238 00:20:24,000 --> 00:20:29,430 before. I don't want to go to too much into the details. The results are obtained 239 00:20:29,430 --> 00:20:35,090 by running this algorithm many times. And as you can see, we have achieved not a 240 00:20:35,090 --> 00:20:39,080 hundred percent success probability, but over 50 percent for the most cases, which 241 00:20:39,080 --> 00:20:44,180 is like, yeah, not perfect, of course, but it's good enough to, in our case, show 242 00:20:44,180 --> 00:20:49,350 that there was really a quantum speedup possible. And if you ask why, okay, why is 243 00:20:49,350 --> 00:20:53,560 not 100 percent probability possible or why can't we build larger systems with 244 00:20:53,560 --> 00:20:57,400 data, what kept us from, for example, building a 100 or 1000 qubit quantum 245 00:20:57,400 --> 00:21:03,230 processor? Well, there's several things on this, of course, that we have like we make 246 00:21:03,230 --> 00:21:07,540 errors when we manipulate the Qubits. So the microwave signals are not perfect, for 247 00:21:07,540 --> 00:21:11,520 example. So we introduce small errors when like making two Qubit and single Qubit 248 00:21:11,520 --> 00:21:16,550 interactions. We also need a really high degree of connectivity if we want to build 249 00:21:16,550 --> 00:21:20,170 a large scale quantum computer. So if every Qubit is connected to every other 250 00:21:20,170 --> 00:21:24,490 Qubit, for example, that would make one million connections for 1000 Qubit code 251 00:21:24,490 --> 00:21:28,290 processors, which processor which is just on the engineering side, very hard to 252 00:21:28,290 --> 00:21:33,790 realize. And then also our Qubits has errors because they can the environment 253 00:21:33,790 --> 00:21:39,170 that the Qubits are in, like the chip and the vicinity there also introduces noise 254 00:21:39,170 --> 00:21:43,241 that will destroy our quantum state and that limits how many operations we can 255 00:21:43,241 --> 00:21:50,160 perform on a single Qubit. So is possible solution, which is the surface code 256 00:21:50,160 --> 00:21:55,390 architecture which was introduced in 2009 already actually by David DiVincenzo from 257 00:21:55,390 --> 00:21:58,770 the Jülich Research Center. And the idea here is that we do not have a quantum 258 00:21:58,770 --> 00:22:03,400 process of a full connectivity. So we do not connect every Qubit to every other 259 00:22:03,400 --> 00:22:08,780 Qubit. Instead, we only connect a Qubit to its four neighbors via so-called tunable 260 00:22:08,780 --> 00:22:12,060 coupler. And this is, of course, much easier because you don't need so many 261 00:22:12,060 --> 00:22:15,830 connections on a chip. But it turns out that you can still run most of the quantum 262 00:22:15,830 --> 00:22:19,840 algorithms that you could also run with a fully connected processor. You just have 263 00:22:19,840 --> 00:22:25,000 to pay like a penalty for the limited connectivity. And the nice thing is also 264 00:22:25,000 --> 00:22:30,410 that you can encode a single logical Qubit. So Qubit that we want to do 265 00:22:30,410 --> 00:22:35,820 calculations with as for example, five physical Qubits. And so all of these 266 00:22:35,820 --> 00:22:40,700 Qubits here that are on the chip would together form one logical Qubit, which 267 00:22:40,700 --> 00:22:43,910 would then allow us to do error corrections so we can, if there had been 268 00:22:43,910 --> 00:22:47,750 some error of one of the Qubits, for example, of relaxation or a defacing 269 00:22:47,750 --> 00:22:52,350 error, then we can use the other Qubits that we prepared in exactly the same of 270 00:22:52,350 --> 00:22:56,520 same way to correct this error and continue doing the calculations. And this 271 00:22:56,520 --> 00:23:00,460 is quite important because in these superconducting Qubit systems, there are 272 00:23:00,460 --> 00:23:04,600 always error present errors present, and we will not probably be able to eliminate 273 00:23:04,600 --> 00:23:09,480 all of them. So we need to find a way to correct the errors, while we perform the 274 00:23:09,480 --> 00:23:18,270 computation. Now the Google processor follows the surface code approach, here I 275 00:23:18,270 --> 00:23:23,510 show you an image from the Nature article which was released, I think, one months 276 00:23:23,510 --> 00:23:28,420 ago. So it's a very impressive system, I find, it contains 50 trees superconducting 277 00:23:28,420 --> 00:23:34,340 Qubits, 86 couplers, tunable couplers between those Qubits and they achieve 278 00:23:34,340 --> 00:23:40,410 fidelity. So the success probability, if you like, for performing one and two Qubit 279 00:23:40,410 --> 00:23:45,880 gates, which is higher than 99 percent. So this is already pretty, very, very good. 280 00:23:45,880 --> 00:23:52,360 And almost enough fidelity to realize quantum error correction as I discussed 281 00:23:52,360 --> 00:23:57,520 before. And with the system, you can really run quite complex quantum 282 00:23:57,520 --> 00:24:03,850 algorithms, much more complex than the ones that we run in 2012. So the paper, 283 00:24:03,850 --> 00:24:07,730 for example, they run sequences with 10 to 20 individual quantum operations or 284 00:24:07,730 --> 00:24:14,870 Quantum Gates. And just to give you an impression of the crisis study, a 285 00:24:14,870 --> 00:24:21,760 cryogenic engineering and microwave engineering here, this is so to say, the 286 00:24:21,760 --> 00:24:26,620 delusion crisis that where the Qubit ship is mounted and you can see that it's quite 287 00:24:26,620 --> 00:24:31,780 a bit more complex than the system we had in 2012. So it really looks way more like 288 00:24:31,780 --> 00:24:39,590 a professional quantum computer, I would say. If you ask a physicist now, why would 289 00:24:39,590 --> 00:24:45,130 you build such a system? The answer would be, of course. Well, it's awesome. So why 290 00:24:45,130 --> 00:24:51,060 not? But it turns out that if an organization like Google gives like 100 or 291 00:24:51,060 --> 00:24:55,510 200 million US dollars for realizing such research, they also want to see some 292 00:24:55,510 --> 00:25:02,710 results. So that's why the team, of course, under John Martinez tried to use 293 00:25:02,710 --> 00:25:08,940 this quantum process for something, that shows how powerful or dead, so to say, can 294 00:25:08,940 --> 00:25:17,980 outperform a classical computer. And this sounds easy, but actually it's not so not 295 00:25:17,980 --> 00:25:22,840 so easy to find a problem that is both doable on this quantum computer, which has 296 00:25:22,840 --> 00:25:28,440 like 50 Qubits and a bit more than 50 Qubits and like 80 couplers. But it's not 297 00:25:28,440 --> 00:25:33,040 possible to simulate on a classical computer. So we could think, for example, 298 00:25:33,040 --> 00:25:38,830 about the factoring of numbers into prime components, which is, of course, always 299 00:25:38,830 --> 00:25:43,331 like the motivation of certain agencies to push for quantum computing, because that 300 00:25:43,331 --> 00:25:47,620 would allow them to read everyone's email. But unfortunately, in both, the number of 301 00:25:47,620 --> 00:25:53,150 Qubits that he would require for this and the number of operations is much too high 302 00:25:53,150 --> 00:25:58,100 to be able to realize something like this on this processor. The next thing, which 303 00:25:58,100 --> 00:26:01,730 would be very interesting is the simulation of quantum systems. So if you 304 00:26:01,730 --> 00:26:06,130 have like molecules or other quantum systems that have many degrees of freedom, 305 00:26:06,130 --> 00:26:10,610 it's very difficult to simulate those on classical computers. On a quantum computer 306 00:26:10,610 --> 00:26:14,930 you could do it efficiently. But again, since the Google team did not do this, I 307 00:26:14,930 --> 00:26:19,990 assume the quantum computer was just or they didn't have like a feasible problem 308 00:26:19,990 --> 00:26:24,240 where they could actually perform such a simulation that would not be not be 309 00:26:24,240 --> 00:26:28,830 performing well or like calculable on a classical computer. So but in the near- 310 00:26:28,830 --> 00:26:32,620 term, in the future, this might actually be very relevant application of such a 311 00:26:32,620 --> 00:26:38,150 processor. The last possibility would be to run, for example, the search algorithm 312 00:26:38,150 --> 00:26:42,430 that we discussed before. But again, for the number of Qubits that are in the 313 00:26:42,430 --> 00:26:47,690 system and the size of the search space, it's still not possible because the 314 00:26:47,690 --> 00:26:52,190 algorithm requires too many steps. And the limited coherence times of Qubits in this 315 00:26:52,190 --> 00:26:56,690 processor make it impossible to to run this kind of like algorithm there, at 316 00:26:56,690 --> 00:27:04,740 least to my knowledge. So what what they did then, was therefore to perform a 317 00:27:04,740 --> 00:27:09,410 different kind of experiment, one that was doable with the processor, which is so- 318 00:27:09,410 --> 00:27:15,320 called randomized benchmarking. And in this case, what you do is that you instead 319 00:27:15,320 --> 00:27:19,900 of like running an algorithm that does something actually useful, like a search 320 00:27:19,900 --> 00:27:24,200 algorithm, you run just a random sequence of gates. So you have, for example, your 321 00:27:24,200 --> 00:27:28,730 53 Qubits and then you run first like some single Qubit gates. So you changed the 322 00:27:28,730 --> 00:27:33,910 Qubit values individually. Then you run two Qubit gates between random Qubits to 323 00:27:33,910 --> 00:27:37,630 create like a superposition and an entangled state. And in the end, it just 324 00:27:37,630 --> 00:27:43,590 read out the resulting qubit state from your register. And this is also very 325 00:27:43,590 --> 00:27:48,850 complex operation. So you really need a very high degree of like control of your 326 00:27:48,850 --> 00:27:53,840 quantum processor, which the Martinez is, the Google team was able to achieve here. 327 00:27:53,840 --> 00:27:59,270 It's not it's just not solving a really practical problem yet, so to say. But on 328 00:27:59,270 --> 00:28:03,960 the other hand, it's the it's it's the system. It's an algorithm that can be run 329 00:28:03,960 --> 00:28:08,160 on the quantum computer easily, but which is, as we will see, very difficult to 330 00:28:08,160 --> 00:28:13,600 simulate or reproduce on a classical system. And the reason that it's so 331 00:28:13,600 --> 00:28:17,540 difficult to reproduce on a classical system is that, if you want to simulate 332 00:28:17,540 --> 00:28:21,200 the action of these quantum gates that we run on the quantum computer using a 333 00:28:21,200 --> 00:28:26,770 classical machine, a classical computer, then for every Qubit that we add, roughly 334 00:28:26,770 --> 00:28:31,990 the size of our problem, space quadruples. So you can imagine if you have like two 335 00:28:31,990 --> 00:28:36,500 Qubits, then it's very easy to simulate that you can do it on like your iPhone or 336 00:28:36,500 --> 00:28:42,000 like your computer, for example. If you add more and more Qubits store, you can 337 00:28:42,000 --> 00:28:46,760 see that the problem size becomes really really big really fast. So if you have 338 00:28:46,760 --> 00:28:51,280 like 20 Qubits, 30 Qubits, for example, you cannot do it on a personal computer 339 00:28:51,280 --> 00:28:55,700 anymore. You will need like supercomputer. And then if you keep increasing the number 340 00:28:55,700 --> 00:29:00,510 of Qubits, then at some point in this case, 50 Qubits or 53 Qubits, it would be 341 00:29:00,510 --> 00:29:04,870 impossible even for the fastest supercomputers that we have right now. And 342 00:29:04,870 --> 00:29:09,090 that's what is called the so-called quantum supremacy regime here for this 343 00:29:09,090 --> 00:29:15,180 randomized gate sequences, which is basically just the area here on the curve. 344 00:29:15,180 --> 00:29:20,300 That is C, that is still doable for this quantum processor that Google realized. 345 00:29:20,300 --> 00:29:25,760 But it's not simulatorable or verifiable by any classical computer, even like a 346 00:29:25,760 --> 00:29:32,620 supercomputer in a reasonable amount of time. And if we can run something in this 347 00:29:32,620 --> 00:29:37,640 regime here, it proves that we have a quantum system that is able to do 348 00:29:37,640 --> 00:29:41,860 computation, which is not classically reproducible. So it's something that 349 00:29:41,860 --> 00:29:46,220 really can only be done on a quantum computer. And that's why running this kind 350 00:29:46,220 --> 00:29:50,750 of experiment is is interesting, because it really shows us that quantum computers 351 00:29:50,750 --> 00:29:55,050 can do things that classical computers cannot do, even if there are for the 352 00:29:55,050 --> 00:30:01,550 moment not really useful. And the gate sequence that they run looks something 353 00:30:01,550 --> 00:30:06,670 like this. We can see again, like here five, four of the Qubits that the Google 354 00:30:06,670 --> 00:30:11,420 team has. And they run sequences of operations of different lengths, then 355 00:30:11,420 --> 00:30:14,960 perform a measurement and then just sample the output of their measurements. So what 356 00:30:14,960 --> 00:30:20,750 they get as a result is a sequence of long bit strings, so zeros and ones. For each 357 00:30:20,750 --> 00:30:26,710 experiment, they run and to reproduce the, to check that a quantum computer is 358 00:30:26,710 --> 00:30:30,830 actually doing the right thing, you have to compare it to the results of a 359 00:30:30,830 --> 00:30:37,310 classical simulation of this algorithm. And that's, of course, a problem now, 360 00:30:37,310 --> 00:30:42,870 because, we just said that we realized the quantum computer, a quantum processor, 361 00:30:42,870 --> 00:30:48,440 which is able to do this computation on 53 Qubits and that no classical computer can 362 00:30:48,440 --> 00:30:54,670 verify that. So the question is now, how can they prove or show that what the 363 00:30:54,670 --> 00:30:58,240 quantum computer calculates is actually the correct answer or that he does not 364 00:30:58,240 --> 00:31:01,790 just produce some garbage values? And that's a very interesting question, 365 00:31:01,790 --> 00:31:07,320 actually. And the way they did it here is by extrapolation. So instead of, for 366 00:31:07,320 --> 00:31:11,820 example, solving the full circuits, so that contains all of the connections and 367 00:31:11,820 --> 00:31:17,240 all of the gates of the full algorithm, they created simplified circuits in two 368 00:31:17,240 --> 00:31:21,570 different ways. So, for example, they cut they cut some of the connections between 369 00:31:21,570 --> 00:31:26,330 the Qubits and the algorithms, so that the problem space would become a bit smaller 370 00:31:26,330 --> 00:31:30,350 or in the other case, with the allied circuit, they just changed the operations 371 00:31:30,350 --> 00:31:34,710 in order to allow for some shortcuts in the classical computation of the classical 372 00:31:34,710 --> 00:31:39,820 simulation of the algorithm. So in both cases, they were able to then verify the 373 00:31:39,820 --> 00:31:43,600 result of the quantum computation with this classical simulation performed on a 374 00:31:43,600 --> 00:31:48,850 supercomputer. And then they basically just did this for a larger and larger 375 00:31:48,850 --> 00:31:53,770 number of Qubits. They plotted the resulting curve and they extrapolated that 376 00:31:53,770 --> 00:31:58,960 to the supremacy regime to see that. OK. Based on the error models that they 377 00:31:58,960 --> 00:32:02,760 developed, based on the simulation, they can with a certain confidence, of course, 378 00:32:02,760 --> 00:32:06,930 say that probably the quantum computer is doing the right thing even in the 379 00:32:06,930 --> 00:32:11,920 supremacy regime, even though we can't they cannot verify it using the classical 380 00:32:11,920 --> 00:32:18,720 simulations. And in case the quantum computer did wrong still, they have also 381 00:32:18,720 --> 00:32:22,730 archive to the results. And maybe ten years when we have better supercomputers, 382 00:32:22,730 --> 00:32:28,240 we might be able to just go back to them and then verify them against the 53, 53 383 00:32:28,240 --> 00:32:31,920 Qubits processor here, by which time, of course, they might already have like a 384 00:32:31,920 --> 00:32:39,610 larger quantum processor again. So the key results of this, I would say, are that for 385 00:32:39,610 --> 00:32:43,940 the first time they show that really quantum computer can beat a classical 386 00:32:43,940 --> 00:32:49,220 computer, even though it is at a very artificial and probably not very useful 387 00:32:49,220 --> 00:32:53,280 problem. And what the experiment also shows is that really, I would say an 388 00:32:53,280 --> 00:32:59,060 astounding level of control of such a large scale on medium size quantum 389 00:32:59,060 --> 00:33:05,410 processor, because even five years ago, six years ago, 2012, 2013, the systems 390 00:33:05,410 --> 00:33:10,610 that we worked with mostly consisted of three or four Qubits and we could barely 391 00:33:10,610 --> 00:33:16,000 fabricate the chips and manipulate them to get like algorithms running. And now if I 392 00:33:16,000 --> 00:33:21,090 see like a 50 tree Qubit processor with such a high degree of control and fidelity 393 00:33:21,090 --> 00:33:25,530 there, I can really say that is really an amazing progress in the last five years 394 00:33:25,530 --> 00:33:30,600 that what was achieved, especially by the Google Martinez team here. And I think it 395 00:33:30,600 --> 00:33:34,200 is a very good wild milestone on the way to fully work on quantum computer because 396 00:33:34,200 --> 00:33:38,950 it nicely shows the limitations of the current system and gives a good direction 397 00:33:38,950 --> 00:33:44,590 on new areas of research, for example, an error correction, where we can improve the 398 00:33:44,590 --> 00:33:50,150 different aspects of the quantum processor. The research has also been 399 00:33:50,150 --> 00:33:55,330 criticized from various sides, so I just want to iterate a few of the points 400 00:33:55,330 --> 00:34:00,050 here. One of the criticisms is, of course, that it doesn't do anything useful. So 401 00:34:00,050 --> 00:34:05,840 there's really no applicability of this experiment and why that's true. It's, of 402 00:34:05,840 --> 00:34:11,450 course, very difficult to go from like a basic, very simple quantum process of two 403 00:34:11,450 --> 00:34:15,450 Qubits to a system that can really factorize prime numbers or do anything 404 00:34:15,450 --> 00:34:20,109 useful. So we will always need to find problems that are both hard enough so that 405 00:34:20,109 --> 00:34:24,210 we can solve them in a reasonable timeframe. A couple of years, for example, 406 00:34:24,210 --> 00:34:28,540 that still proved the progress that we make on the road to quantum computing. So 407 00:34:28,540 --> 00:34:33,070 in this sense, while quantum supremacy does not really show anything useful in 408 00:34:33,070 --> 00:34:37,690 terms of computation that is done. I think it is still a very good problem as a 409 00:34:37,690 --> 00:34:41,580 benchmark for any kind of quantum processor, because it requires that you 410 00:34:41,580 --> 00:34:46,110 have very good control over your system and that you can run such a number of 411 00:34:46,110 --> 00:34:50,621 gates at a very high fidelity, which is really currently, I would say, the state 412 00:34:50,621 --> 00:34:56,899 of the art. The research also took, took some shortcuts. For example, they used 413 00:34:56,899 --> 00:35:00,290 like a two Qubits, quantum gates, which are not, as we call them, canonical 414 00:35:00,290 --> 00:35:04,230 gates, which might be problematic because if you want to run a quantum 415 00:35:04,230 --> 00:35:07,900 algorithm on the system, you need to implement certain quantum gates that 416 00:35:07,900 --> 00:35:12,240 you need for that. And since they only have like non canonical gates here, which 417 00:35:12,240 --> 00:35:16,740 are still universal, by the way, they could not do that directly, but with some 418 00:35:16,740 --> 00:35:21,060 modification of the system, that should also be possible. And the last criticism 419 00:35:21,060 --> 00:35:25,640 might be that, okay, here you have a problem that was engineered to match a 420 00:35:25,640 --> 00:35:31,540 solution, which is of course that, okay, we need some solution, as I said, some 421 00:35:31,540 --> 00:35:36,520 problem that we can't realistically solve on a such a system. I think, though, also 422 00:35:36,520 --> 00:35:40,230 like the other points, if you want to build a large scale quantum processor, you 423 00:35:40,230 --> 00:35:45,030 need to define reasonable milestones and having such a benchmark that other groups, 424 00:35:45,030 --> 00:35:49,720 for example, can also run that process against is a very good thing because it 425 00:35:49,720 --> 00:35:54,200 makes the progress visible and also makes it easy to compare how different groups 426 00:35:54,200 --> 00:35:59,820 or are companies or organizations are are at competing on the 427 00:35:59,820 --> 00:36:12,040 number of Qubits under control they have about them. So, if you want to make a more 428 00:36:12,040 --> 00:36:16,770 kind of Moore's Law for quantum computing, there would be several possibilities that 429 00:36:16,770 --> 00:36:22,550 you could do. Here I show you, for example, the number of Qubits that have 430 00:36:22,550 --> 00:36:28,620 been realized for superconducting systems over the years. This is, of course 431 00:36:28,620 --> 00:36:32,340 incomplete because it could like the number of Qubits alone doesn't tell you 432 00:36:32,340 --> 00:36:36,810 much about your system. I mean, we could do a Qubit chip of 1000 or 10000 Qubits 433 00:36:36,810 --> 00:36:41,230 today. But if you don't have the connectivity and don't have to controllability 434 00:36:41,230 --> 00:36:45,119 of individual Qubits, then this chip wouldn't be good. So there are other 435 00:36:45,119 --> 00:36:49,210 things, that we also need to take into account here. As I said, just as like the 436 00:36:49,210 --> 00:36:53,550 coupling between individual Qubits and the coherence time and the fidelity of the 437 00:36:53,550 --> 00:37:00,010 Qubit operations. So this is really just one one very small aspect of this whole 438 00:37:00,010 --> 00:37:03,470 whole problem space. But I think it shows nicely that in the last years there was 439 00:37:03,470 --> 00:37:07,790 really tremendous progress in terms of the power of the superconducting systems, 440 00:37:07,790 --> 00:37:15,350 because the original Qubit, which was developed in at NYC in Japan by 441 00:37:15,350 --> 00:37:20,710 Professor Nakamura, was done in like around 2000. So that very, very bad 442 00:37:20,710 --> 00:37:24,920 coherence time, very bad properties. But still it showed for the first time that he 443 00:37:24,920 --> 00:37:28,970 could coherently control such a system. And then it didn't take long for other 444 00:37:28,970 --> 00:37:32,390 groups, for example, to Quantronics Group and so Saclay, to pick up on this 445 00:37:32,390 --> 00:37:37,480 work and to do to keep improving it. So after a few years, we already had Qubits 446 00:37:37,480 --> 00:37:42,250 of a few hundred or even a microsecond of coherence time, which was like in like 447 00:37:42,250 --> 00:37:46,100 three or orders of magnitude better than what we had before. And there were other 448 00:37:46,100 --> 00:37:51,490 advances then made by groups in the US, for example, in Yale, the ShowCoupLab, 449 00:37:51,490 --> 00:37:56,010 which developed new Qubit architectures that allowed us to couple the Qubits more 450 00:37:56,010 --> 00:37:59,570 efficiently with each other and to again have better control of them manipulating 451 00:37:59,570 --> 00:38:04,910 them. And then there's also groups like the research group at IBM or companies 452 00:38:04,910 --> 00:38:09,410 like WeGetty that took again these ideas and that added engineering and their 453 00:38:09,410 --> 00:38:14,130 own research on top of that in order to make the systems even better. So in 2018, 454 00:38:14,130 --> 00:38:19,610 we already had systems with 17 or 18 Qubits in them. And now with this Google 455 00:38:19,610 --> 00:38:25,440 and UC Santa Barbara work, we have the first systems with more than 50 Qubits 456 00:38:25,440 --> 00:38:32,510 after not even 20 years, which I think is quite some progress in this area. And of 457 00:38:32,510 --> 00:38:38,520 course, if you ask me how close we are to and actually working quantum computer, 458 00:38:38,520 --> 00:38:44,860 it's still very difficult to say, I find, because we've proven the group prove the 459 00:38:44,860 --> 00:38:49,840 quantum supremacy for its randomized algorithm. But in order to do something 460 00:38:49,840 --> 00:38:56,370 applicable or useful with such a quantum system, I think we need like at least 461 00:38:56,370 --> 00:39:03,440 again, 50 maybe 200 additional Qubits and a larger number of Qubit operations. But 462 00:39:03,440 --> 00:39:07,299 it's really hard to say. That's why I also say don't believe in this chart because 463 00:39:07,299 --> 00:39:11,890 there's also, of course, a lot of work in the theory of quantum algorithms, because 464 00:39:11,890 --> 00:39:16,360 up to now we are still discovering new approaches of doing quantum simulations 465 00:39:16,360 --> 00:39:19,800 for examples. And right now, there are a lot of research groups that are looking 466 00:39:19,800 --> 00:39:24,120 for ways to make these medium scale quantum computers. So quantum computers 467 00:39:24,120 --> 00:39:29,940 with 50 or 100 Qubits already useful for using quantum simulations. So it's really 468 00:39:29,940 --> 00:39:35,420 an interplay between what the theory can give us in terms of quantum algorithm and 469 00:39:35,420 --> 00:39:40,070 what in terms of experimental realization we can build as a quantum processor. So in 470 00:39:40,070 --> 00:39:44,390 my opinion, quantum simulation will definitely be something that where we will 471 00:39:44,390 --> 00:39:49,390 see the first applications in the next. I would say three to five years. Other 472 00:39:49,390 --> 00:39:54,900 things, optimizations. I have to admit I am less an expert and I think they're a 473 00:39:54,900 --> 00:39:58,810 bit more complex. So we will probably see the first applications in those areas a 474 00:39:58,810 --> 00:40:04,011 bit later. And the big motivation for like the three letter agencies always is, 475 00:40:04,011 --> 00:40:10,350 of course, the factoring out the breaking of cryptosystems, which is the most 476 00:40:10,350 --> 00:40:14,619 challenging one, though, because in order to do that, you would both need very large 477 00:40:14,619 --> 00:40:19,850 numbers of Qubits. So at least 8000 Qubits for an 8000 bits RSA key, for 478 00:40:19,850 --> 00:40:23,831 example. And you would also need a very large amount of Qubit operations because 479 00:40:23,831 --> 00:40:29,520 you need to run the sure operation. And that involves a lot of steps for the 480 00:40:29,520 --> 00:40:34,350 quantum processor. And so to say the most, I would say from my perspective 481 00:40:34,350 --> 00:40:39,550 unrealistic application of superconducting quantum processes in the next year. But I 482 00:40:39,550 --> 00:40:43,320 think, if somebody would build a quantum computer, maybe we would also not just 483 00:40:43,320 --> 00:40:52,749 know about it. So who knows? So to summarize, quantum computers, quantum 484 00:40:52,749 --> 00:40:57,020 processors are getting really, seriously complex and very impressive. So we have 485 00:40:57,020 --> 00:41:02,260 seen tremendous progress in the last five years. I still think that we are like five 486 00:41:02,260 --> 00:41:06,840 years away from building really practical quantum computers and there are still some 487 00:41:06,840 --> 00:41:11,510 challenges. For example, an error correction in the Quantum Gatefidelity and 488 00:41:11,510 --> 00:41:15,500 indeed, again, general architecture of these systems that we need to overcome. 489 00:41:15,500 --> 00:41:18,690 And they might also be some challenges which we haven't even identified yet with 490 00:41:18,690 --> 00:41:22,750 which we might only encounter at a later stage when trying to build really large 491 00:41:22,750 --> 00:41:28,390 scale quantum processors. And as a last point, I just want to stress again, that 492 00:41:28,390 --> 00:41:33,700 quantum computing research is not only done by Google or by IBM, there a lot 493 00:41:33,700 --> 00:41:37,540 of groups in the world involved in this kind of research, both in theory and an 494 00:41:37,540 --> 00:41:43,030 experiment. And as I said before, a lot of the breakthroughs that we use today for 495 00:41:43,030 --> 00:41:47,410 building quantum processes were done in very different places like Japan, Europe, 496 00:41:47,410 --> 00:41:52,690 USA. So it's really, I would say, a global effort. And you should also, when you 497 00:41:52,690 --> 00:41:58,281 look, when you see this marketing PR that companies like Google and IBM do, maybe 498 00:41:58,281 --> 00:42:03,869 not believe all of the hype they're creating and keep on down to earth views, 499 00:42:03,869 --> 00:42:10,950 so to say, of the limits and the potential of quantum computing. So that's it. And I 500 00:42:10,950 --> 00:42:13,940 would be happy to take on your questions now. And if you have any 501 00:42:13,940 --> 00:42:19,160 feedback, there's also my Twitter handle and my email address. And I think we also 502 00:42:19,160 --> 00:42:23,540 have some time for questions here right now. Thank you. 503 00:42:23,540 --> 00:42:32,430 *Applause* 504 00:42:32,430 --> 00:42:37,050 Herald: Thank you, Andreas. We have almost 20 minutes for Q and A. If you're leaving 505 00:42:37,050 --> 00:42:42,560 now, please do so very quietly and if you can avoid it, just don't do it. Thank you. 506 00:42:42,560 --> 00:42:47,390 Okay. Q and A. You know the game. There's eight microphones in this room, so just 507 00:42:47,390 --> 00:42:53,580 queue behind them and we will do our best to get everyone sorted out sequentially. 508 00:42:53,580 --> 00:42:58,490 We will start with a question from the Internet. 509 00:42:58,490 --> 00:43:01,859 Signal-Angel: Thank you. Do you have information about the energy consumption 510 00:43:01,859 --> 00:43:07,840 of a quantum computer over the calculation power? 511 00:43:07,840 --> 00:43:11,729 Andreas: Yeah, that's an interesting point. I mean, for superconducting quantum 512 00:43:11,729 --> 00:43:16,490 computers, there are like several costs associated. I think right now the biggest 513 00:43:16,490 --> 00:43:20,970 cost is probably of keeping the system cooled down. So as that you need very low 514 00:43:20,970 --> 00:43:25,570 temperatures, 20 or 10 millikelvin. In order to achieve that, you need the so- 515 00:43:25,570 --> 00:43:29,490 called delusion crisis that and these systems that consume a lot of energy and 516 00:43:29,490 --> 00:43:36,300 also materials like helium mixtures, which are expensive and like maybe not so well, 517 00:43:36,300 --> 00:43:40,510 kind of like a real material right now. I think that would be the biggest 518 00:43:40,510 --> 00:43:46,290 consumption in terms of energy use. I honestly don't have so much of an idea. I 519 00:43:46,290 --> 00:43:50,060 mean, the manipulation of the Qubit system is done via microwaves and the power that 520 00:43:50,060 --> 00:43:54,070 goes into the system is very small compared to any of the power that we use 521 00:43:54,070 --> 00:43:58,330 for cooling the system. So I would say for the foreseeable future, the power 522 00:43:58,330 --> 00:44:02,370 consumption should be dominated by like the cooling and the setup costs and the 523 00:44:02,370 --> 00:44:06,180 cost of the electronics as well. So the classical electronics that that controls 524 00:44:06,180 --> 00:44:10,440 the Qubit, which can also be quite extensive for large system. So the Qubit 525 00:44:10,440 --> 00:44:14,690 chip itself should be very should be really negligible in terms of energy 526 00:44:14,690 --> 00:44:18,200 consumption. Herald: Thank you. Microphone number one 527 00:44:18,200 --> 00:44:22,560 please. Mic 1: Hello. I have a question in regards 528 00:44:22,560 --> 00:44:28,320 to quantum simulation. So I would have thought that with 53 Qubits, 529 00:44:28,320 --> 00:44:34,900 there would already be something interesting to do, since I think their 530 00:44:34,900 --> 00:44:41,680 border the limit for more or less exact quantum chemistry calculations on 531 00:44:41,680 --> 00:44:46,930 classical computers is that there are 10 to 20 particles. So is there a more 532 00:44:46,930 --> 00:44:53,720 complicated relation from particles to Qubits that's missing here or what's the 533 00:44:53,720 --> 00:44:57,570 problem? Andreas: Yeah. So in the paper I couldn't 534 00:44:57,570 --> 00:45:02,920 find an exact reason why they choose this problem. I think there are probably two 535 00:45:02,920 --> 00:45:09,960 aspects. One is that you don't have in the system the like arbitrary Qubit control. 536 00:45:09,960 --> 00:45:14,940 So to say you cannot run like any Hamiltonian or quantum algorithm that you 537 00:45:14,940 --> 00:45:18,660 want. You are like limited in terms of connectivity. So it's possible that they 538 00:45:18,660 --> 00:45:25,100 were not able to run any quantum algorithm for simulation, which was not easy to run 539 00:45:25,100 --> 00:45:28,810 also on a classical system, you know, so. But I'm really not not sure why they 540 00:45:28,810 --> 00:45:33,000 didn't. I think just if they would have a have had this chance to do a quantum 541 00:45:33,000 --> 00:45:36,630 simulation, they would probably have done that instead, because that's, of course, 542 00:45:36,630 --> 00:45:41,530 more impressive than randomization or randomized algorithms. So because they 543 00:45:41,530 --> 00:45:45,869 didn't do it, I think it was just probably too complicated or not possible to realize 544 00:45:45,869 --> 00:45:49,949 on the system. Yeah. Okay. So it's this. But again, I don't know for sure yet. 545 00:45:49,949 --> 00:45:52,789 Thank you. Herald: Yes, and also speaking as a sometimes 546 00:45:52,789 --> 00:45:58,270 quantum chemist, you can't directly map Qubits to to atoms. They're not two level 547 00:45:58,270 --> 00:46:02,880 systems. And you don't I mean, you usually also simulate electrons and not just 548 00:46:02,880 --> 00:46:07,310 atoms, but I'm not a speaker. We can discuss later. Maybe microphone number two 549 00:46:07,310 --> 00:46:10,869 please. Mic 2: Thanks. Can you compare this 550 00:46:10,869 --> 00:46:16,530 classic or general quantum computer to the one by D-wave? That's one of the quantum 551 00:46:16,530 --> 00:46:22,480 computers by a AWS offered. They have two thousand Qubits or something. 552 00:46:22,480 --> 00:46:26,050 Andreas: Yeah, that's a very interesting question. So D-wave system is this so- 553 00:46:26,050 --> 00:46:32,020 called adiabatic quantum computer, to my knowledge. So this in this case the 554 00:46:32,020 --> 00:46:36,750 computation works a bit differently. It's the normal with this quantum computer that 555 00:46:36,750 --> 00:46:40,560 Google produced. You have a gate sequence that you run on your input Qubits and then 556 00:46:40,560 --> 00:46:44,290 you get a result that you read out. With the D-wave system it's more that you like 557 00:46:44,290 --> 00:46:48,560 engineer like in Hamiltonian, which is also which also consists of local 558 00:46:48,560 --> 00:46:53,160 interactions between different Qubits. And then you slowly changed this Hamiltonian 559 00:46:53,160 --> 00:46:58,320 in order to like change to the ground state of the system to a solution of a 560 00:46:58,320 --> 00:47:02,680 problem that you're looking for. So. So it's a different approach to quantum 561 00:47:02,680 --> 00:47:09,520 computation. They also claimed that they can can achieve what I did, achieve a 562 00:47:09,520 --> 00:47:14,270 quantum supremacy, I think in a different way for like an optimization problem. But 563 00:47:14,270 --> 00:47:19,859 to my knowledge, the proof they have is less rigid probably than, what the Google 564 00:47:19,859 --> 00:47:23,610 Group produced here. So but again, I'm not like an expert on that, a bit of quantum 565 00:47:23,610 --> 00:47:29,800 computing. So I'm more like a gate based person. So, yeah, I think though, the 566 00:47:29,800 --> 00:47:34,570 proof that here the Google show is more convincing in terms of like reproduce 567 00:47:34,570 --> 00:47:40,300 reproducibility and really make the proof that you are actually doing something that 568 00:47:40,300 --> 00:47:47,490 cannot be done on a classical computer. D-Wave will see the different view though. 569 00:47:47,490 --> 00:47:53,850 Herald: All right. Let's go to the back. Number seven, please. Hello. 7. You just 570 00:47:53,850 --> 00:47:58,630 waved to me. Mic 7: Hey, uh, hello. Uh, I was reading 571 00:47:58,630 --> 00:48:06,369 that earlier this year IBM released the first commercial Q one system or whatever 572 00:48:06,369 --> 00:48:11,760 the name is. And you were mentioning before to keep our expectations down to 573 00:48:11,760 --> 00:48:18,520 Earth. So my question is, what kind of commercial expectations is IBM actually 574 00:48:18,520 --> 00:48:22,921 creating? Andreas: Mm hmm. So I spoke to some 575 00:48:22,921 --> 00:48:30,369 companies here in Germany that are collaborating with IBM or D-Wave or Google 576 00:48:30,369 --> 00:48:35,290 as well. And to ask what they're actually doing with the quantum computers. They are 577 00:48:35,290 --> 00:48:41,090 the the companies offer. And I think the answer is that right now, a lot of 578 00:48:41,090 --> 00:48:45,500 commercially, a lot of companies are investigating this as something that could 579 00:48:45,500 --> 00:48:50,670 potentially be very useful or very relevant in five to 10 years. So they want 580 00:48:50,670 --> 00:48:54,751 to get some experience and they want to start collaborating. I don't think, at 581 00:48:54,751 --> 00:49:00,040 least I don't know any reproduction use of these systems where the quantum computer 582 00:49:00,040 --> 00:49:05,130 would do some calculations, that would not be doable on a classical system. But 583 00:49:05,130 --> 00:49:08,560 again, I don't have a full overview of that. I think now it's mostly for 584 00:49:08,560 --> 00:49:12,890 experimentation and forgetting to notice systems. I think the companies or most of 585 00:49:12,890 --> 00:49:17,260 the customers there probably expect that in five years or 10 years, the system will 586 00:49:17,260 --> 00:49:21,030 systems will really be powerful enough to do some useful computations with them as 587 00:49:21,030 --> 00:49:24,520 well. Herald: Thanks. All right. The Internet, 588 00:49:24,520 --> 00:49:27,270 please. Signal-Angel: With a quantum computer, you 589 00:49:27,270 --> 00:49:32,260 can calculate things in parallel. But there is this usability requirement. So 590 00:49:32,260 --> 00:49:36,820 how much faster is a quantum computer at the end of the day? 591 00:49:36,820 --> 00:49:42,440 Andreas: Mm hmm. Yeah, it's true, so that if you want to and, if you want to realize 592 00:49:42,440 --> 00:49:46,700 classical algorithm, you have to do it in a reversible way. But to my knowledge, you 593 00:49:46,700 --> 00:49:51,920 can from an efficiency perspective, implement any classical non reversible 594 00:49:51,920 --> 00:49:59,210 algorithm as a reversible algorithm without loss in complexity. So you can 595 00:49:59,210 --> 00:50:02,510 have also like for a reversible computation, you have universal gaits like 596 00:50:02,510 --> 00:50:07,291 the control not gate that you can use to express any logic function that you 597 00:50:07,291 --> 00:50:12,580 require. You might need some additional Qubits in compared to the amount of the 598 00:50:12,580 --> 00:50:16,220 classical bits that you need for the computation. But in principle, there is 599 00:50:16,220 --> 00:50:20,009 nothing that keeps you from implementing any classical function on a quantum 600 00:50:20,009 --> 00:50:24,670 computer. In terms of actual runtime, of course it depends on how fast you can run 601 00:50:24,670 --> 00:50:28,760 individual operations. Right now, a single Qubits operation, for example, on this 602 00:50:28,760 --> 00:50:34,820 Google machine takes about I think 20 to 40 nanoseconds. So in that sense, the 603 00:50:34,820 --> 00:50:39,010 quantum computers are probably much slower than classical computers. But the idea is 604 00:50:39,010 --> 00:50:43,109 anyway that you do only really the necessary computations that you can't do 605 00:50:43,109 --> 00:50:46,450 on a classical machine, on a quantum computer and anything else you can do on a 606 00:50:46,450 --> 00:50:52,410 normal classical system. So the quantum process in this sense is only like a like 607 00:50:52,410 --> 00:50:56,990 inside a core processor, like a GPU, in that sense, I would say. 608 00:50:56,990 --> 00:50:59,850 Herald: All right. Microphone number four, please. 609 00:50:59,850 --> 00:51:05,270 Mic 4: On the slide that shows Richard Feynman, you said that quantum computers 610 00:51:05,270 --> 00:51:14,020 were invented to simulate quantum systems. And can you please elaborate on that? 611 00:51:14,020 --> 00:51:17,760 Herald: You went past, huh? Andreas: Yeah. So I don't have to link to 612 00:51:17,760 --> 00:51:21,830 the lecture here. Unfortunately, the link is broken, but you can find that online. 613 00:51:21,830 --> 00:51:27,130 It's a 1982 lecture from Feynman, where he discusses like how you would actually go 614 00:51:27,130 --> 00:51:32,780 about simulating a quantum system, because as we have shown like the if you want to 615 00:51:32,780 --> 00:51:36,849 simulate a full quantum system, you need to simulate the density matrix of the system 616 00:51:36,849 --> 00:51:42,414 and that takes about that take, it takes an exponential amount of memory and 617 00:51:42,414 --> 00:51:46,849 computation in terms of like the number of Qubits or quantum degrees of freedom that 618 00:51:46,849 --> 00:51:51,590 you want to simulate. And with a classical Turing machine, you couldn't do that in an 619 00:51:51,590 --> 00:51:55,980 efficient way because every time you add a single Qubit, you basically quadruple your 620 00:51:55,980 --> 00:52:00,320 computational requirement. And that's really where the idea came from. I think 621 00:52:00,320 --> 00:52:04,860 from Feynman to think about a computing system that would use quantum mechanics in 622 00:52:04,860 --> 00:52:09,109 order to be able to do these kind of simulations, because he saw probably that 623 00:52:09,109 --> 00:52:13,260 for large quantum systems it would never be possible to run, at least with our 624 00:52:13,260 --> 00:52:16,190 current understanding of classical computing. It would never be possible to 625 00:52:16,190 --> 00:52:20,390 run a quantum simulation of a quantum system on a classical computer in an 626 00:52:20,390 --> 00:52:23,550 efficient way. Does that answer the question? 627 00:52:23,550 --> 00:52:25,380 Mic 4: Yeah. Andreas: Okay. 628 00:52:25,380 --> 00:52:28,290 Herald: All right. Microphone eight, please. 629 00:52:28,290 --> 00:52:35,820 Mic 8: As a physicist who's now doing analog circuit design. I'm kind of 630 00:52:35,820 --> 00:52:41,620 wondering why all the presentations about quantum computers always use stage zero 631 00:52:41,620 --> 00:52:45,880 and 1 and not multiple states. Is that a fundamental limitation or is that just 632 00:52:45,880 --> 00:52:49,530 just a simplification for the sake of the presentation? 633 00:52:49,530 --> 00:52:52,730 Andreas: So you mean why you don't use like higher Qubit states or like... 634 00:52:52,730 --> 00:52:57,540 Mic 8: Multi valued logic or even continuous states? 635 00:52:57,540 --> 00:53:01,330 Andreas: So in principle, the quantum bits that we're using, they don't they're not 636 00:53:01,330 --> 00:53:05,090 really two level systems. So there is not only level zero and one, but also level 637 00:53:05,090 --> 00:53:10,520 two tree and so on. You could use them, of course, but the computational power of the 638 00:53:10,520 --> 00:53:15,609 system is given as the number of states, or like m for example, race to the power 639 00:53:15,609 --> 00:53:20,240 of the number of Qubits. So M to the power of N. So in that sense, if you add like 640 00:53:20,240 --> 00:53:26,750 another state, you only change like. Like a small affected and adding another Qubit. 641 00:53:26,750 --> 00:53:30,550 So it's usually not very interesting to add more states. What he would do instead, 642 00:53:30,550 --> 00:53:35,380 is just add more Qubits to your system. And for continuous variable quantum 643 00:53:35,380 --> 00:53:39,660 mechanic quantum computation. I think there is some use cases where this might 644 00:53:39,660 --> 00:53:43,870 outperform like the digital quantum computers, especially if you can engineer 645 00:53:43,870 --> 00:53:48,980 your system to like mimic the Hamiltonian of the system that you want to simulate. 646 00:53:48,980 --> 00:53:54,050 So I think in this sense, in these cases, it makes a lot of sense. For other cases 647 00:53:54,050 --> 00:53:57,720 where you say, OK, you want to run a general quantum computation, then like 648 00:53:57,720 --> 00:54:01,179 such a digital quantum computer is probably the best solution. And you could 649 00:54:01,179 --> 00:54:07,820 also just add that run like a continuous simulation of a quantum system on such a 650 00:54:07,820 --> 00:54:13,460 gate based quantum system, just like the linearly in the same order of complexity, 651 00:54:13,460 --> 00:54:17,850 I would say. Does that answer the question? 652 00:54:17,850 --> 00:54:23,470 Mic 8: I think I delude myself to have understood that the non diagonal elements 653 00:54:23,470 --> 00:54:28,450 in the density matrix grow much faster than the number of states in any and any 654 00:54:28,450 --> 00:54:32,170 diagonal matrix element. Andreas: I guess you could say like that. 655 00:54:32,170 --> 00:54:37,530 Yeah, I have to think about. Herald: All right. Number three, please. 656 00:54:37,530 --> 00:54:43,320 Mic 3: What do you have to say about the scepticism of people like Nikolai that 657 00:54:43,320 --> 00:54:51,100 claim that inherent nice will be a fundamental problem in scaling this 658 00:54:51,100 --> 00:54:54,820 quantum computers? Andreas: I mean, it's a valid concern, I 659 00:54:54,820 --> 00:55:01,410 think. As of today, we don't have even for a single Qubit shown error correction. 660 00:55:01,410 --> 00:55:04,840 There are some first experiments, for example, by the ShowCoup Lab in Yale that 661 00:55:04,840 --> 00:55:08,970 showed some of the elements of error correction for a single Qubit system, but 662 00:55:08,970 --> 00:55:15,230 we haven't even managed today to keep a single Qubit alive indefinitely. So that's 663 00:55:15,230 --> 00:55:19,220 why I would say it's an open question. It's a valid criticism. I think the next 664 00:55:19,220 --> 00:55:23,160 five years will show if we are actually able to run this quantum errors and if our 665 00:55:23,160 --> 00:55:26,310 error models themselves are correct because they only correct for certain 666 00:55:26,310 --> 00:55:30,809 errors or if there's anything else that keeps us from like building a large scale 667 00:55:30,809 --> 00:55:35,400 system. So I think it's a totally valid point. 668 00:55:35,400 --> 00:55:40,990 Herald: Microphone five, please. Mic 5: There has been a study on 669 00:55:40,990 --> 00:55:48,830 factarising on adiabatic machines, which requires a lock squared N Qubits while 670 00:55:48,830 --> 00:55:58,260 Shor requires Log N. But as the adiabatic systems have much higher Qubit numbers, 671 00:55:58,260 --> 00:56:05,140 they were able to factorize on these machines, much larger numbers than on the 672 00:56:05,140 --> 00:56:11,870 normal devices. And that's something that never shows up in the discussion. Do you 673 00:56:11,870 --> 00:56:17,320 want to comment on that? Have you read the study? What do you think? Are adiabatic 674 00:56:17,320 --> 00:56:23,200 machines, bogus? Or, is that worth while resolved? 675 00:56:23,200 --> 00:56:26,130 Andreas: I'm not. Yeah, as I said, like an expert at adiabatic quantum 676 00:56:26,130 --> 00:56:31,980 computing. I know that there were some like studies or investigations of the 677 00:56:31,980 --> 00:56:37,690 D-wave system. Like I haven't read this particular study about factorization. I 678 00:56:37,690 --> 00:56:40,940 think adiabatic quantum computing is a valid approach as well to quantum 679 00:56:40,940 --> 00:56:48,520 computing. I just I'm not just just not sure if currently like the results were 680 00:56:48,520 --> 00:56:54,849 like shown with the same amount of like rigidity or like rigid proves like for the 681 00:56:54,849 --> 00:56:58,380 gate based quantum computer. But yeah, I'm I really would have to look at the study 682 00:56:58,380 --> 00:57:02,730 to to see that. Herald: Can you maybe quickly say the 683 00:57:02,730 --> 00:57:09,270 authors. So it's on the record. Yeah. If your mike is still on number five. 684 00:57:09,270 --> 00:57:14,140 Mic 5: Sorry, I don't. Herald: Okay, no problem. Thank you. All 685 00:57:14,140 --> 00:57:15,790 right. Andreas: But yeah, I don't think adiabatic 686 00:57:15,790 --> 00:57:19,660 quantum computing is like and I think adiabatic quantum computing is a 687 00:57:19,660 --> 00:57:24,190 valid choice or valid approach for doing quantum computation as well. 688 00:57:24,190 --> 00:57:29,339 Mic 5: So I can give you that. I can search for the authors later and give it 689 00:57:29,339 --> 00:57:30,759 to you. Andreas: Okay. Okay. It would be great. 690 00:57:30,759 --> 00:57:33,029 Thank you. Herald: Thank you. Microphone four, 691 00:57:33,029 --> 00:57:36,199 please. Mic 4: What do you say about IBM's claim 692 00:57:36,199 --> 00:57:41,100 that Google's supremacy claim is invalid because the problem was not really hard? 693 00:57:41,100 --> 00:57:45,840 Andreas: Yeah. So basically IBM, I think said, okay, if you do some optimizations 694 00:57:45,840 --> 00:57:49,810 on the way you simulate the systems, then you can reduce this computation time from 695 00:57:49,810 --> 00:57:55,290 10000 years to like maybe a few hours or so. I think it's, of course, valid. It 696 00:57:55,290 --> 00:57:59,910 might be a valid claim. I don't know if it really invalidates the result because as I 697 00:57:59,910 --> 00:58:04,760 said, like the computational power of like the classical systems, they will also will 698 00:58:04,760 --> 00:58:09,700 also increase in the coming years. Right now, you could say that maybe if we 699 00:58:09,700 --> 00:58:14,839 haven't achieved quantum supremacy in regards to elect 2019 hardware, then maybe 700 00:58:14,839 --> 00:58:19,220 we should just like look at the 2015 hardware and then we can say, okay, there, 701 00:58:19,220 --> 00:58:24,369 probably we achieved that. In any case, I think the most what's most impressive 702 00:58:24,369 --> 00:58:28,930 about this result for me is not like, if we are really in the supremacy regime or 703 00:58:28,930 --> 00:58:34,200 maybe not. That's really the amount of.., the degree of controlability of the 704 00:58:34,200 --> 00:58:37,290 Qubits system that this group has achieved. I think that's really the 705 00:58:37,290 --> 00:58:40,920 important point here, regardless of whether they actually achieved the 706 00:58:40,920 --> 00:58:46,190 supremacy or not. Because it shows that these kind of systems seem to be a good 707 00:58:46,190 --> 00:58:50,250 architecture choice for building large scale quantum processes. And this alone is 708 00:58:50,250 --> 00:58:54,680 very valuable, I think, as a guide to future research direction, regardless of 709 00:58:54,680 --> 00:58:59,750 whether this is actually, you know, they achieved this or not. Yeah, but yeah, I 710 00:58:59,750 --> 00:59:05,760 can understand, of course, the criticism. Mic 4: OK. One thing. The article is 711 00:59:05,760 --> 00:59:11,099 called Quantum Annealing for Prime Factorization appeared in Nature in 712 00:59:11,099 --> 00:59:18,739 December 18. Authors Jiang, A. Britt, Alex J. McCaskey, S. Humble and Kais. 713 00:59:18,739 --> 00:59:22,190 Andreas: Okay, great. I think we'll have a look at that again. Thanks. 714 00:59:22,190 --> 00:59:25,359 Herald: All right. Microphone 6, do you have a short question? 715 00:59:25,359 --> 00:59:33,990 Mic 6: Yeah, hopefully. It is known that it is not very easy to understand how 716 00:59:33,990 --> 00:59:41,000 large quantum superposition goes into a macroscopic state. So in the macroscopic 717 00:59:41,000 --> 00:59:47,340 physical description. So apparently there are a couple of things not understood. So 718 00:59:47,340 --> 00:59:51,970 is there anything you know about when you go two thousand, ten thousand, million 719 00:59:51,970 --> 01:00:00,310 Qubits, could you expect the quantum behavior to break down? Are there any 720 01:00:00,310 --> 01:00:07,150 fundamental argument that this will not happen or is this not a problem considered 721 01:00:07,150 --> 01:00:09,760 recently? Andreas: Huh, Okay. I'm not sure if I 722 01:00:09,760 --> 01:00:13,089 fully understand the question. It's mostly about like if you say like quantum 723 01:00:13,089 --> 01:00:17,820 mechanics or some like scale variance so that if you go to a certain scale and some 724 01:00:17,820 --> 01:00:22,160 time, at some point you have like a irreversibility or like a something like 725 01:00:22,160 --> 01:00:27,010 that. Yeah. I mean, I think that a large quantum systems that occur naturally, I 726 01:00:27,010 --> 01:00:30,210 don't know. I like Bose Einstein condensate, for example, has a lot of 727 01:00:30,210 --> 01:00:33,640 degrees of freedom that are not controlled, of course, but that also 728 01:00:33,640 --> 01:00:39,420 quantum mechanical and there it seems to work. So personally, I would think that 729 01:00:39,420 --> 01:00:43,780 there is no such limit. But I mean, who knows? It's like that's why we do like 730 01:00:43,780 --> 01:00:47,960 experimental physics. So we will see as if we reached it. But from like the theory of 731 01:00:47,960 --> 01:00:52,099 quantum mechanics right now, there is no indication that this should be such a 732 01:00:52,099 --> 01:00:55,670 limit to my knowledge. Herald: All right, so maybe we will see 733 01:00:55,670 --> 01:00:57,870 you again in five years. Andreas: Yeah. 734 01:00:57,870 --> 01:01:00,490 Herald. So please thank Andreas, until I ask once again. Thanks. 735 01:01:00,490 --> 01:01:02,184 *Applause* 736 01:01:02,184 --> 01:01:05,654 *36c3 postroll music* 737 01:01:05,654 --> 01:01:28,000 Subtitles created by c3subtitles.de in the year 2020. Join, and help us!