0:00:00.000,0:00:18.519 36C3 Preroll music 0:00:18.519,0:00:23.300 Herald: So hello and welcome to a quantum[br]computing talk by the Andreas [Dewes], who 0:00:23.300,0:00:27.050 gave a talk exactly five years ago and[br]it's almost exactly five years ago, it's 0:00:27.050,0:00:33.580 like one year and two or three hours. And[br]then he gave a talk at 31C3 about quantum 0:00:33.580,0:00:37.710 computing titled "Let's Build a Quantum[br]Computer". And I think back then we 0:00:37.710,0:00:41.080 basically had just found out that Google[br]was planning to partner with the 0:00:41.080,0:00:46.459 University of California at Santa Barbara[br]to try to build a quantum computer. Of 0:00:46.459,0:00:50.140 course, now we're five years later, we've[br]had a lot of developments, I think, in the 0:00:50.140,0:00:54.510 field. We've had some big announcements by[br]Google and other groups. And Andreas has 0:00:54.510,0:00:58.470 now come back to give us an update. So[br]please welcome him to the stage. 0:00:58.470,0:01:05.310 Applause 0:01:05.310,0:01:10.010 Andreas: Okay. Hi, everyone, so I'm very[br]happy to be here again. After five years 0:01:10.010,0:01:16.199 of giving the first version of this talk.[br]My motivation for given this talk is quite 0:01:16.199,0:01:22.880 simple. I was often so I did my PHD on[br]exponential quantum computing from 2009 to 0:01:22.880,0:01:28.400 2012. I left that field afterwards to work[br]in industry, but always people would come 0:01:28.400,0:01:33.120 to me and ask, hey, Andreas, did you see[br]like this new experiment. Did you see, you 0:01:33.120,0:01:37.850 can like use quantum computers on Amazon's[br]cloud now? Did you see, like Google has 0:01:37.850,0:01:42.320 this new quantum thing? This is really[br]working? Can we use quantum computers yet? 0:01:42.320,0:01:47.950 Why are you not working on this? And I[br]couldn't I couldn't really answer the 0:01:47.950,0:01:53.960 question. So I said, OK, I want to go back[br]to this and find out what happened in the 0:01:53.960,0:01:57.900 last five years since I finished my PhD.[br]What kind of progress was made in the 0:01:57.900,0:02:02.110 field? And do we actually have quantum[br]computers today that are working already 0:02:02.110,0:02:11.999 or are we not yet quite just there? So we[br]want to do it like this. I want to first 0:02:11.999,0:02:16.020 give you a short introduction to quantum[br]computing. So just that we have a common 0:02:16.020,0:02:19.590 understanding of how that works and why[br]it's interesting. Then I will show you a 0:02:19.590,0:02:24.290 small example of experimental quantum[br]speed up. Notably the work I did with my 0:02:24.290,0:02:29.290 colleagues in Saclay during my PhD[br]thesis. Then we discuss some of the 0:02:29.290,0:02:33.569 challenges and problems, why we were not[br]able to build a real quantum computer back 0:02:33.569,0:02:38.269 then. And I will discuss some approaches[br]that have come up since then. That would 0:02:38.269,0:02:42.750 basically allow us to do that eventually.[br]And then we'll, of course, discuss 0:02:42.750,0:02:47.489 Google's recent experiment in[br]collaboration with the University of Santa 0:02:47.489,0:02:52.540 Barbara, where they showed basically a[br]very impressive quantum computing system 0:02:52.540,0:02:57.849 with 53 Qubits. We will look exactly to[br]try to understand what they did there and 0:02:57.849,0:03:02.090 see if that's really like a quantum[br]computer in the in the real sense already 0:03:02.090,0:03:05.349 or if there's still something missing. And[br]in the end, of course, I will try to give 0:03:05.349,0:03:13.349 you another small outlook to see what we[br]can expect in the coming years. So in 0:03:13.349,0:03:17.059 order to talk about quantum computing, we[br]need to first talk about classical 0:03:17.059,0:03:21.599 computing just a little bit. You might[br]know that classical computers, they work 0:03:21.599,0:03:27.099 with bits, so zeros and ones. They store[br]them in so-called registers. This here for 0:03:27.099,0:03:33.549 example of like a bit register. Of course,[br]the bits themselves are not very 0:03:33.549,0:03:38.989 interesting. But we have to do stuff with[br]them so we can compute functions over 0:03:38.989,0:03:43.959 those bit registers. That's what like[br]modern CPU is doing in a simplified way, 0:03:43.959,0:03:48.590 of course. So we take some input, but[br]register values, we compute some function 0:03:48.590,0:03:57.869 over then and then we get an output value.[br]So a very simple example would be a search 0:03:57.869,0:04:02.269 problem. I would discuss this because[br]later we will also see in the experiment 0:04:02.269,0:04:05.889 how we can use a quantum computer to solve[br]this. So I just want to motivate why this 0:04:05.889,0:04:10.619 kind of problem can be interesting. And[br]it's a very silly search function. So it 0:04:10.619,0:04:16.060 takes two bits as inputs and it returns[br]one bit as an output, indicating, whether 0:04:16.060,0:04:20.799 the input bits are the solution to our[br]search problem or not. And you could 0:04:20.799,0:04:24.280 imagine that we have a very, very[br]complicated function here. So, for 0:04:24.280,0:04:29.120 example, a function that calculates the[br]answer to life, the universe and 0:04:29.120,0:04:33.710 everything, while not a complete answer,[br]but only the first two bits. So really 0:04:33.710,0:04:39.600 complicated to implement and very costly[br]to execute. So we might think that it 0:04:39.600,0:04:43.760 might take like millions of years to run[br]this function once on our inputs. And so 0:04:43.760,0:04:47.740 we want to find the right solution to that[br]function with as few function calls as 0:04:47.740,0:04:56.800 possible, of course. Overall, there are[br]four possibilities, so for input states, 00 0:04:56.800,0:05:02.320 01 10 and 11 that we can apply our[br]function to and only for one of these 0:05:02.320,0:05:09.140 states. The 01 state, because the answer[br]is 42. So that's 0 times 1 plus to plan 2 0:05:09.140,0:05:13.780 plus some other stuff. So the first two[br]bits are 0 1 for this for a value, the 0:05:13.780,0:05:19.570 function returns a 1 for all of the other[br]values, the function returns 0. Now let's 0:05:19.570,0:05:25.280 think about how we can implement a central[br]search function and in principle, if we 0:05:25.280,0:05:28.880 don't know anything about the function. So[br]we can imagine it's so complicated that we 0:05:28.880,0:05:34.590 can't do any optimizations. We don't know[br]where to look. So we have to really try 0:05:34.590,0:05:39.130 each of these values in sequence. And for[br]this we can have a simple algorithm so we 0:05:39.130,0:05:45.520 can start initializing out our a bit[br]register with 00 value. Then we can call 0:05:45.520,0:05:50.860 the function on that register. We can see[br]what the result is. In this case, the 0:05:50.860,0:05:54.900 result would be zero. If the result would[br]be 1, then we know, okay, we have found 0:05:54.900,0:05:59.480 our solution so we can stop our algorithm.[br]But in this case, the result is zero. So 0:05:59.480,0:06:05.340 we can just go back to the left value and[br]to the left step and increase the register 0:06:05.340,0:06:11.360 value, go to 0 1 and try again. And in the[br]worst case, depending if you're optimistic 0:06:11.360,0:06:15.460 or not, we have to do this three or four[br]times. So if you want to really be sure 0:06:15.460,0:06:19.570 that we find the right answers, we have to[br]do it four times in the worst case. And 0:06:19.570,0:06:23.210 this is sort of say the time complexity or[br]the computational complexity of the 0:06:23.210,0:06:27.150 search. You know, if you imagine that in[br]our algorithm, the most expensive 0:06:27.150,0:06:32.550 operation is really calling this function[br]F, then the calling time of the complexity 0:06:32.550,0:06:37.380 of calling this function will be what[br]dominates the complexity of our algorithm. 0:06:37.380,0:06:42.120 And in this case, the complexity is very[br]similar, simple here because it's linear 0:06:42.120,0:06:46.540 in the number of the search space. So if[br]you have n states, for example, in our 0:06:46.540,0:06:50.480 examples, we have four different input[br]spaced states. We also need to evaluate 0:06:50.480,0:06:54.450 the function four times. So and please[br]keep this graph in mind because we're 0:06:54.450,0:06:58.780 gonna revisit that later a bit to see,if[br]we can do better with a different paradigm 0:06:58.780,0:07:03.300 of computing. And so classically. This is[br]really the best we can do for the search 0:07:03.300,0:07:06.580 problem here because we don't know[br]anything else about the function that 0:07:06.580,0:07:12.560 would allow us to optimize that further.[br]But now the interesting thing is that we 0:07:12.560,0:07:17.580 might imagine that we don't use classical[br]computing for solving our problem. And in 0:07:17.580,0:07:23.270 fact, the discipline that we call quantum[br]computing was kind of like inspired by 0:07:23.270,0:07:29.510 lecturer or like a seminar of Richard[br]Feynman, who thought about, how it would be 0:07:29.510,0:07:34.820 possible to similar and or if it would be[br]possible to simulate quantum systems on a 0:07:34.820,0:07:39.380 classical computer. A Turing machine, if[br]you want. And he found that because 0:07:39.380,0:07:43.300 quantum mechanics is so complicated for[br]classical computers that it is not 0:07:43.300,0:07:47.040 possible to do that efficiently, but that[br]if you would use the laws of quantum 0:07:47.040,0:07:51.620 mechanics themselves to make a computer[br]like quantum computer, then it would be 0:07:51.620,0:07:55.850 possible to simulate this quantum systems[br]and just kind of like sparked this whole 0:07:55.850,0:08:00.020 idea of using quantum mechanics to do[br]computation. And in the following years, 0:08:00.020,0:08:04.570 they were not only as solutions found for[br]simulating quantum systems, which such a 0:08:04.570,0:08:08.940 quantum computer, but also for solving[br]other not related problems to quantum 0:08:08.940,0:08:18.610 computing. So like search problems or[br]factorization problems, for example. And 0:08:18.610,0:08:22.540 quantum computers can do, can do[br]computation faster than classical 0:08:22.540,0:08:27.010 computers, because they have several[br]differences in how they work. So one of 0:08:27.010,0:08:32.130 the key differences here is superposition,[br]which means that if you use a quantum 0:08:32.130,0:08:37.070 computer, instead of a classical computer,[br]we cannot only load a single register 0:08:37.070,0:08:42.620 value into our bit register. So for[br]example, the first value of only zeros. 0:08:42.620,0:08:49.041 But instead we can kind of load all of the[br]possible state values and it at once or in 0:08:49.041,0:08:54.949 parallel. And this so-called quantum state[br]or quantum superposition state where each 0:08:54.949,0:08:59.699 of these values here has an amplitude[br]which is shown on the left, that is 0:08:59.699,0:09:04.559 basically a complex number that relates[br]them to the other Qubit, to other states 0:09:04.559,0:09:11.360 and ups. If you have like for example,[br]n-Qubits, then the total number of Qubits 0:09:11.360,0:09:15.509 states can be very large 2 to the[br]power of N. So we can imagine that if you 0:09:15.509,0:09:19.730 have a large Qubit quantum quantum bit[br]register, then your number of quantum 0:09:19.730,0:09:26.290 states can be really, really large and[br]this can be very powerful for computation. 0:09:26.290,0:09:31.550 So in the rest of the talk, we gonna just[br]indicate this by like showing the register 0:09:31.550,0:09:37.129 as like a small rectangle to indicate that[br]it's not only a single value in there, but 0:09:37.129,0:09:41.309 that we have a superposition values of all[br]the possible input values to our function, 0:09:41.309,0:09:46.579 for example. And there is a condition and[br]so called normalization condition that 0:09:46.579,0:09:51.110 puts some constraints on these amplitude.[br]Because the sum of the squares of the 0:09:51.110,0:09:55.230 absolute values of these amplitude needs[br]to sum to one, which basically means that 0:09:55.230,0:09:59.029 the entire the probability of each of[br]these of all of these states together 0:09:59.029,0:10:07.439 needs to be 100 percent. So. And this is[br]the first ingredient that makes quantum 0:10:07.439,0:10:12.300 computers interesting for computation[br]because we can basically implement any 0:10:12.300,0:10:16.259 classical function that we can also run on[br]a classical computer, on a quantum 0:10:16.259,0:10:21.620 computer. The difference is that we cannot[br]only run it for one value at a time, but 0:10:21.620,0:10:25.260 we can call it can run it down on a[br]superposition of all possible input 0:10:25.260,0:10:29.140 values. So if you want, you have like this[br]massive paralellyzation where you run you 0:10:29.140,0:10:33.910 off computation on all possible inputs at[br]once and also calculate and all of the 0:10:33.910,0:10:39.260 possible output values. And that sounds,[br]of course, very cool and very useful. 0:10:39.260,0:10:44.040 There's a catch that we will discuss[br]later. So it's not as easy as that. But 0:10:44.040,0:10:50.110 this is one step off like the power that[br]makes quantum computing interesting. The 0:10:50.110,0:10:54.360 next thing that is different is that we[br]can on a quantum computer, not only run 0:10:54.360,0:10:58.619 classical gates or classical functions,[br]but we can also run so-called quantum 0:10:58.619,0:11:04.170 gates. And the quantum gates, they're[br]different in respect to the classical 0:11:04.170,0:11:09.490 functions because they cannot only like[br]classical operations like and or or on 0:11:09.490,0:11:15.000 like two Qubits in a predictable way. But[br]they can kind of like act on the whole 0:11:15.000,0:11:20.350 Qubit state at once and also create so-[br]called entangled states which are really 0:11:20.350,0:11:25.369 weird quantum states where we can't really[br]separate the state of one Qubit from the 0:11:25.369,0:11:29.100 state of or other Qubits. So it's kind of[br]like if we want to try to make a small 0:11:29.100,0:11:32.860 change to one of two Qubits in our[br]system, we also changing other Qubits 0:11:32.860,0:11:38.110 there. So we can never like separate the[br]bits, the Qubits out like we can with a 0:11:38.110,0:11:42.269 classical computer. And this is another[br]resource that we can use in quantum 0:11:42.269,0:11:49.970 computing to solve certain problems faster[br]than we could with a classical computer. 0:11:49.970,0:11:54.939 Now, the catch, as I said, is that we, of[br]course, do not, we do not want to only 0:11:54.939,0:12:00.670 make computation with our Qubits, Qubits[br]register, but we also want to read out the 0:12:00.670,0:12:05.580 result of our computation. And if we try[br]that. So we make like computation. And 0:12:05.580,0:12:10.339 when we want to measure the state of our[br]quantum register, we have a small problem 0:12:10.339,0:12:15.240 because, well, the measurement process is[br]actually quite complicated. But in a very 0:12:15.240,0:12:19.699 simplified way, you can just imagine, that[br]God is trying some dice here. And then if 0:12:19.699,0:12:23.709 we have a quantum vector, a quantum state[br]vector that has like this amplitude on the 0:12:23.709,0:12:28.730 left. So a one to a n. And then we will[br]pick. He or she would pick a state 0:12:28.730,0:12:34.170 randomly from the possible states. And the[br]probability of getting a given state as a 0:12:34.170,0:12:38.860 result is proportional, as is that before[br]to the square of the absolute value of the 0:12:38.860,0:12:44.099 amplitude. So that means we can perform[br]computation on all of the possible input 0:12:44.099,0:12:48.379 states of our function. But when we read[br]out the result, we will only get one of 0:12:48.379,0:12:53.990 the possible results. So it's kind of like[br]destroys at the first glimpse the utility 0:12:53.990,0:12:57.870 of quantum computing because we can do[br]like computation on all states in 0:12:57.870,0:13:02.029 parallel, but we cannot read out the[br]result. So not a very interesting computer 0:13:02.029,0:13:08.380 because we can't learn about the output.[br]So to say or not easily at least. But it 0:13:08.380,0:13:14.759 turns out that there's actually a way of[br]still using quantum computing to be faster 0:13:14.759,0:13:19.420 than a classical computer. And the first[br]kind of practical algorithm for a search 0:13:19.420,0:13:24.090 problem, notably the search problem that[br]we discussed before, was given by Love 0:13:24.090,0:13:30.490 Grover, who was a researcher at the Bell[br]Labs, and who found the Grover algorithm 0:13:30.490,0:13:36.619 that is named after him. That's basically[br]a search algorithm which can prove it can, 0:13:36.619,0:13:40.980 as we will see, solved the search problem[br]that we have in a much more efficient way 0:13:40.980,0:13:46.879 than any classical computer could. And in[br]my opinion, it's still one of the most 0:13:46.879,0:13:52.529 beautiful quantum algorithms because it's[br]very simple and it's very powerful and 0:13:52.529,0:13:56.140 does also prove, unlike for other[br]algorithms like the factorization 0:13:56.140,0:14:01.489 algorithms from Shor that the Grover[br]algorithm can be will be faster always 0:14:01.489,0:14:06.350 than any classical computer classical[br]algorithm. So in my opinion, it's a very 0:14:06.350,0:14:11.889 nice example of really a quantum algorithm[br]that is more powerful than a classical 0:14:11.889,0:14:20.609 one. Let's see how it works. So they're[br]three steps again and the algorithm. First 0:14:20.609,0:14:26.769 we initialize our Qubit register, our[br]state vector to a superposition of the 0:14:26.769,0:14:33.730 four possible output values, so 00 01[br]10 and 10, again, all with equal 0:14:33.730,0:14:40.440 probability in this case, zero amplitude.[br]Then we evaluate the function on this 0:14:40.440,0:14:44.619 input state here and what the function[br]then does. So we made some special 0:14:44.619,0:14:50.490 encoding here that basically marks the[br]solution of our problem by changing the 0:14:50.490,0:14:54.980 sign of the amplitude of the corresponding[br]state. We can see that in the output state 0:14:54.980,0:15:01.449 here, the 01 state has a sign which is[br]negative, which means that it's the 0:15:01.449,0:15:06.300 solution of the problem that we search.[br]Still, if we were to the read out now 0:15:06.300,0:15:10.410 directly, we wouldn't be able to learn[br]anything about the solution, because as 0:15:10.410,0:15:14.899 you can see, the amplitude is still equal[br]for all of the four states. So if you 0:15:14.899,0:15:20.050 would make a read out now, we would only[br]get like one of the four possible states 0:15:20.050,0:15:24.490 at random so we wouldn't learn anything[br]with a hundred percent probability about 0:15:24.490,0:15:29.350 the solution of our problem. In order to[br]do that, we need to apply another step to 0:15:29.350,0:15:35.249 so-called Grover or Diffusion, diffusion[br]operator, which now takes this phase 0:15:35.249,0:15:39.360 difference or the sign difference between[br]these individual quantum states and 0:15:39.360,0:15:45.310 applies a quantum operator to that, that[br]basically transfers the amplitude from all 0:15:45.310,0:15:49.129 of the states that are not a solution to a[br]problem to the states that is the 0:15:49.129,0:15:54.730 solution. And for on this case, with two[br]Qubits here and with four possible values, 0:15:54.730,0:15:58.959 there's only one step we need. And after[br]executing that, you can see that now the 0:15:58.959,0:16:04.119 amplitude of our solution state is one[br]versus very. But the amplitude of the 0:16:04.119,0:16:09.579 other states is all zero. So that's great,[br]because now we can just do a Qubit 0:16:09.579,0:16:14.389 measurement and then we will have a[br]hundred percent probability find a 0:16:14.389,0:16:18.839 solution to our search problem. And that's[br]where kind of like the magic of quantum 0:16:18.839,0:16:24.339 mechanics shows, because you can evaluate[br]its function only once. So remember that 0:16:24.339,0:16:27.920 in the first step we only call the search[br]function once of all of the values in 0:16:27.920,0:16:33.549 parallel. So from the computational[br]complexity, we are much lower than the 0:16:33.549,0:16:38.170 classical algorithm, but still we are able[br]to find 100 percent position in this case 0:16:38.170,0:16:45.209 to see which state is the solution to our[br]search problem. So and that's working not 0:16:45.209,0:16:49.799 only for the case of two Qubits, but also[br]with larger Qubit registers. So for 0:16:49.799,0:16:54.529 example, if you would take 10 Qubits, you[br]would need to execute a few more of these 0:16:54.529,0:16:59.549 steps, two and three. So instead of one[br]iteration, you would need 25 iterations, 0:16:59.549,0:17:04.760 for example, here, which is still much[br]better than the 1024 iterations that you 0:17:04.760,0:17:09.010 would need if you would really look into[br]every possible solution of the function in 0:17:09.010,0:17:15.870 the classical algorithm. So the speed up[br]here is very good for, so to say, all of 0:17:15.870,0:17:21.630 the like. It's quadratical for the[br]solution space. And if you like, look at 0:17:21.630,0:17:28.550 the complexity plot again, we can now[br]compare our classical algorithm with the 0:17:28.550,0:17:34.130 quantum algorithm on the Grover search.[br]And as you can see, the time complexity 0:17:34.130,0:17:39.690 or the number of variations of F that we[br]need is only a square root of N, where N 0:17:39.690,0:17:44.550 is the size of the search space,[br]which shows that that we have really a 0:17:44.550,0:17:48.420 speed advantage hier of the quantum[br]computer versus the classical computer. 0:17:48.420,0:17:54.180 And nice thing is the larger our search[br]space becomes, the more dramatic our speed 0:17:54.180,0:17:58.550 up will be, because for example, for a[br]search space with one million 0:17:58.550,0:18:02.510 elements. We will only have to evaluate[br]the search function 1000 times instead of 0:18:02.510,0:18:14.480 one million times. So that's quite so to[br]say a speed up in that sense. Now, how can 0:18:14.480,0:18:20.260 we build a system that realizes this[br]quantum algorithm? Here, I show on the 0:18:20.260,0:18:25.440 quantum processor that I built with my[br]colleagues at the Saclay during my PhD. So 0:18:25.440,0:18:28.730 if you want more information about this,[br]you should check out my last talk. I just 0:18:28.730,0:18:33.210 want to go briefly over the different[br]aspects here. So we use a so called 0:18:33.210,0:18:39.600 superconducting Qubits, transmit Qubits[br]for realizing our quantum computer. A 0:18:39.600,0:18:44.980 quantum processor. You can see the chip[br]here on the top. It's about one centimeter 0:18:44.980,0:18:49.870 across. You can see the two Qubits in the[br]middle. The other, like snake like 0:18:49.870,0:18:53.920 structures are coupling a wave guides[br]where we can manipulate the Qubits using 0:18:53.920,0:18:58.740 microwaves. So we use frequencies that are[br]similar to the ones that are used by 0:18:58.740,0:19:03.870 mobile phones to manipulate and read out[br]our Qubits. And if you look in the 0:19:03.870,0:19:09.240 middle, you can see the red area, which[br]contains the Qubit, each Qubit itself. And 0:19:09.240,0:19:13.000 then there's another zoom in here, which[br]contains the actual qubit structure, which 0:19:13.000,0:19:18.410 is just some two layers of aluminum that[br]have been placed on top of each other and 0:19:18.410,0:19:23.040 which create, when they are cooled, to a[br]very low temperature, a so-called 0:19:23.040,0:19:27.660 superconducting state, where we can use[br]the superconducting face between these two 0:19:27.660,0:19:34.400 values, layers to indicate to to realize[br]our Qubits. There's also coupler in the 0:19:34.400,0:19:39.120 middle. So this green element that you[br]see, which allows us to run quantum gate 0:19:39.120,0:19:49.110 operations between the two Qubits. To use[br]that in practice, we need to put this in a 0:19:49.110,0:19:53.021 delusion crisis that which is really like[br]just a very fancy refrigerator, you could 0:19:53.021,0:19:59.380 say, you cool it down to about 10 milli K.[br]So very low temperature just above the 0:19:59.380,0:20:04.300 absolute zero temperature. You can see the[br]sample holder here on the left side with 0:20:04.300,0:20:08.270 the chip mounted to it. So this whole[br]thing is put in the delusion fridge and 0:20:08.270,0:20:13.000 it's cool down to the temperature. And[br]then we can, as I said, manipulated by 0:20:13.000,0:20:18.950 using his microwave transmission lines.[br]And what we did is we implemented the 0:20:18.950,0:20:24.000 Grover search for the two Qubits. So we[br]ran this algorithm that I discussed 0:20:24.000,0:20:29.430 before. I don't want to go to too much[br]into the details. The results are obtained 0:20:29.430,0:20:35.090 by running this algorithm many times. And[br]as you can see, we have achieved not a 0:20:35.090,0:20:39.080 hundred percent success probability, but[br]over 50 percent for the most cases, which 0:20:39.080,0:20:44.180 is like, yeah, not perfect, of course, but[br]it's good enough to, in our case, show 0:20:44.180,0:20:49.350 that there was really a quantum speedup[br]possible. And if you ask why, okay, why is 0:20:49.350,0:20:53.560 not 100 percent probability possible or[br]why can't we build larger systems with 0:20:53.560,0:20:57.400 data, what kept us from, for example,[br]building a 100 or 1000 qubit quantum 0:20:57.400,0:21:03.230 processor? Well, there's several things on[br]this, of course, that we have like we make 0:21:03.230,0:21:07.540 errors when we manipulate the Qubits. So[br]the microwave signals are not perfect, for 0:21:07.540,0:21:11.520 example. So we introduce small errors when[br]like making two Qubit and single Qubit 0:21:11.520,0:21:16.550 interactions. We also need a really high[br]degree of connectivity if we want to build 0:21:16.550,0:21:20.170 a large scale quantum computer. So if[br]every Qubit is connected to every other 0:21:20.170,0:21:24.490 Qubit, for example, that would make one[br]million connections for 1000 Qubit code 0:21:24.490,0:21:28.290 processors, which processor which is just[br]on the engineering side, very hard to 0:21:28.290,0:21:33.790 realize. And then also our Qubits has[br]errors because they can the environment 0:21:33.790,0:21:39.170 that the Qubits are in, like the chip and[br]the vicinity there also introduces noise 0:21:39.170,0:21:43.241 that will destroy our quantum state and[br]that limits how many operations we can 0:21:43.241,0:21:50.160 perform on a single Qubit. So is possible[br]solution, which is the surface code 0:21:50.160,0:21:55.390 architecture which was introduced in 2009[br]already actually by David DiVincenzo from 0:21:55.390,0:21:58.770 the Jülich Research Center. And the idea[br]here is that we do not have a quantum 0:21:58.770,0:22:03.400 process of a full connectivity. So we do[br]not connect every Qubit to every other 0:22:03.400,0:22:08.780 Qubit. Instead, we only connect a Qubit to[br]its four neighbors via so-called tunable 0:22:08.780,0:22:12.060 coupler. And this is, of course, much[br]easier because you don't need so many 0:22:12.060,0:22:15.830 connections on a chip. But it turns out[br]that you can still run most of the quantum 0:22:15.830,0:22:19.840 algorithms that you could also run with a[br]fully connected processor. You just have 0:22:19.840,0:22:25.000 to pay like a penalty for the limited[br]connectivity. And the nice thing is also 0:22:25.000,0:22:30.410 that you can encode a single logical[br]Qubit. So Qubit that we want to do 0:22:30.410,0:22:35.820 calculations with as for example, five[br]physical Qubits. And so all of these 0:22:35.820,0:22:40.700 Qubits here that are on the chip would[br]together form one logical Qubit, which 0:22:40.700,0:22:43.910 would then allow us to do error[br]corrections so we can, if there had been 0:22:43.910,0:22:47.750 some error of one of the Qubits, for[br]example, of relaxation or a defacing 0:22:47.750,0:22:52.350 error, then we can use the other Qubits[br]that we prepared in exactly the same of 0:22:52.350,0:22:56.520 same way to correct this error and[br]continue doing the calculations. And this 0:22:56.520,0:23:00.460 is quite important because in these[br]superconducting Qubit systems, there are 0:23:00.460,0:23:04.600 always error present errors present, and[br]we will not probably be able to eliminate 0:23:04.600,0:23:09.480 all of them. So we need to find a way to[br]correct the errors, while we perform the 0:23:09.480,0:23:18.270 computation. Now the Google processor[br]follows the surface code approach, here I 0:23:18.270,0:23:23.510 show you an image from the Nature article[br]which was released, I think, one months 0:23:23.510,0:23:28.420 ago. So it's a very impressive system, I[br]find, it contains 50 trees superconducting 0:23:28.420,0:23:34.340 Qubits, 86 couplers, tunable couplers[br]between those Qubits and they achieve 0:23:34.340,0:23:40.410 fidelity. So the success probability, if[br]you like, for performing one and two Qubit 0:23:40.410,0:23:45.880 gates, which is higher than 99 percent. So[br]this is already pretty, very, very good. 0:23:45.880,0:23:52.360 And almost enough fidelity to realize[br]quantum error correction as I discussed 0:23:52.360,0:23:57.520 before. And with the system, you can[br]really run quite complex quantum 0:23:57.520,0:24:03.850 algorithms, much more complex than the[br]ones that we run in 2012. So the paper, 0:24:03.850,0:24:07.730 for example, they run sequences with 10 to[br]20 individual quantum operations or 0:24:07.730,0:24:14.870 Quantum Gates. And just to give you an[br]impression of the crisis study, a 0:24:14.870,0:24:21.760 cryogenic engineering and microwave[br]engineering here, this is so to say, the 0:24:21.760,0:24:26.620 delusion crisis that where the Qubit ship[br]is mounted and you can see that it's quite 0:24:26.620,0:24:31.780 a bit more complex than the system we had[br]in 2012. So it really looks way more like 0:24:31.780,0:24:39.590 a professional quantum computer, I would[br]say. If you ask a physicist now, why would 0:24:39.590,0:24:45.130 you build such a system? The answer would[br]be, of course. Well, it's awesome. So why 0:24:45.130,0:24:51.060 not? But it turns out that if an[br]organization like Google gives like 100 or 0:24:51.060,0:24:55.510 200 million US dollars for realizing such[br]research, they also want to see some 0:24:55.510,0:25:02.710 results. So that's why the team, of[br]course, under John Martinez tried to use 0:25:02.710,0:25:08.940 this quantum process for something, that[br]shows how powerful or dead, so to say, can 0:25:08.940,0:25:17.980 outperform a classical computer. And this[br]sounds easy, but actually it's not so not 0:25:17.980,0:25:22.840 so easy to find a problem that is both[br]doable on this quantum computer, which has 0:25:22.840,0:25:28.440 like 50 Qubits and a bit more than 50[br]Qubits and like 80 couplers. But it's not 0:25:28.440,0:25:33.040 possible to simulate on a classical[br]computer. So we could think, for example, 0:25:33.040,0:25:38.830 about the factoring of numbers into prime[br]components, which is, of course, always 0:25:38.830,0:25:43.331 like the motivation of certain agencies to[br]push for quantum computing, because that 0:25:43.331,0:25:47.620 would allow them to read everyone's email.[br]But unfortunately, in both, the number of 0:25:47.620,0:25:53.150 Qubits that he would require for this and[br]the number of operations is much too high 0:25:53.150,0:25:58.100 to be able to realize something like this[br]on this processor. The next thing, which 0:25:58.100,0:26:01.730 would be very interesting is the[br]simulation of quantum systems. So if you 0:26:01.730,0:26:06.130 have like molecules or other quantum[br]systems that have many degrees of freedom, 0:26:06.130,0:26:10.610 it's very difficult to simulate those on[br]classical computers. On a quantum computer 0:26:10.610,0:26:14.930 you could do it efficiently. But again,[br]since the Google team did not do this, I 0:26:14.930,0:26:19.990 assume the quantum computer was just or[br]they didn't have like a feasible problem 0:26:19.990,0:26:24.240 where they could actually perform such a[br]simulation that would not be not be 0:26:24.240,0:26:28.830 performing well or like calculable on a[br]classical computer. So but in the near- 0:26:28.830,0:26:32.620 term, in the future, this might actually[br]be very relevant application of such a 0:26:32.620,0:26:38.150 processor. The last possibility would be[br]to run, for example, the search algorithm 0:26:38.150,0:26:42.430 that we discussed before. But again, for[br]the number of Qubits that are in the 0:26:42.430,0:26:47.690 system and the size of the search space,[br]it's still not possible because the 0:26:47.690,0:26:52.190 algorithm requires too many steps. And the[br]limited coherence times of Qubits in this 0:26:52.190,0:26:56.690 processor make it impossible to to[br]run this kind of like algorithm there, at 0:26:56.690,0:27:04.740 least to my knowledge. So what what they[br]did then, was therefore to perform a 0:27:04.740,0:27:09.410 different kind of experiment, one that was[br]doable with the processor, which is so- 0:27:09.410,0:27:15.320 called randomized benchmarking. And in[br]this case, what you do is that you instead 0:27:15.320,0:27:19.900 of like running an algorithm that does[br]something actually useful, like a search 0:27:19.900,0:27:24.200 algorithm, you run just a random sequence[br]of gates. So you have, for example, your 0:27:24.200,0:27:28.730 53 Qubits and then you run first like some[br]single Qubit gates. So you changed the 0:27:28.730,0:27:33.910 Qubit values individually. Then you run[br]two Qubit gates between random Qubits to 0:27:33.910,0:27:37.630 create like a superposition and an[br]entangled state. And in the end, it just 0:27:37.630,0:27:43.590 read out the resulting qubit state from[br]your register. And this is also very 0:27:43.590,0:27:48.850 complex operation. So you really need a[br]very high degree of like control of your 0:27:48.850,0:27:53.840 quantum processor, which the Martinez is,[br]the Google team was able to achieve here. 0:27:53.840,0:27:59.270 It's not it's just not solving a really[br]practical problem yet, so to say. But on 0:27:59.270,0:28:03.960 the other hand, it's the it's it's the[br]system. It's an algorithm that can be run 0:28:03.960,0:28:08.160 on the quantum computer easily, but which[br]is, as we will see, very difficult to 0:28:08.160,0:28:13.600 simulate or reproduce on a classical[br]system. And the reason that it's so 0:28:13.600,0:28:17.540 difficult to reproduce on a classical[br]system is that, if you want to simulate 0:28:17.540,0:28:21.200 the action of these quantum gates that we[br]run on the quantum computer using a 0:28:21.200,0:28:26.770 classical machine, a classical computer,[br]then for every Qubit that we add, roughly 0:28:26.770,0:28:31.990 the size of our problem, space quadruples.[br]So you can imagine if you have like two 0:28:31.990,0:28:36.500 Qubits, then it's very easy to simulate[br]that you can do it on like your iPhone or 0:28:36.500,0:28:42.000 like your computer, for example. If you[br]add more and more Qubits store, you can 0:28:42.000,0:28:46.760 see that the problem size becomes really[br]really big really fast. So if you have 0:28:46.760,0:28:51.280 like 20 Qubits, 30 Qubits, for example,[br]you cannot do it on a personal computer 0:28:51.280,0:28:55.700 anymore. You will need like supercomputer.[br]And then if you keep increasing the number 0:28:55.700,0:29:00.510 of Qubits, then at some point in this[br]case, 50 Qubits or 53 Qubits, it would be 0:29:00.510,0:29:04.870 impossible even for the fastest[br]supercomputers that we have right now. And 0:29:04.870,0:29:09.090 that's what is called the so-called[br]quantum supremacy regime here for this 0:29:09.090,0:29:15.180 randomized gate sequences, which is[br]basically just the area here on the curve. 0:29:15.180,0:29:20.300 That is C, that is still doable for this[br]quantum processor that Google realized. 0:29:20.300,0:29:25.760 But it's not simulatorable or verifiable[br]by any classical computer, even like a 0:29:25.760,0:29:32.620 supercomputer in a reasonable amount of[br]time. And if we can run something in this 0:29:32.620,0:29:37.640 regime here, it proves that we have a[br]quantum system that is able to do 0:29:37.640,0:29:41.860 computation, which is not classically[br]reproducible. So it's something that 0:29:41.860,0:29:46.220 really can only be done on a quantum[br]computer. And that's why running this kind 0:29:46.220,0:29:50.750 of experiment is is interesting, because[br]it really shows us that quantum computers 0:29:50.750,0:29:55.050 can do things that classical computers[br]cannot do, even if there are for the 0:29:55.050,0:30:01.550 moment not really useful. And the gate[br]sequence that they run looks something 0:30:01.550,0:30:06.670 like this. We can see again, like here[br]five, four of the Qubits that the Google 0:30:06.670,0:30:11.420 team has. And they run sequences of[br]operations of different lengths, then 0:30:11.420,0:30:14.960 perform a measurement and then just sample[br]the output of their measurements. So what 0:30:14.960,0:30:20.750 they get as a result is a sequence of long[br]bit strings, so zeros and ones. For each 0:30:20.750,0:30:26.710 experiment, they run and to reproduce the,[br]to check that a quantum computer is 0:30:26.710,0:30:30.830 actually doing the right thing, you have[br]to compare it to the results of a 0:30:30.830,0:30:37.310 classical simulation of this algorithm.[br]And that's, of course, a problem now, 0:30:37.310,0:30:42.870 because, we just said that we realized the[br]quantum computer, a quantum processor, 0:30:42.870,0:30:48.440 which is able to do this computation on 53[br]Qubits and that no classical computer can 0:30:48.440,0:30:54.670 verify that. So the question is now, how[br]can they prove or show that what the 0:30:54.670,0:30:58.240 quantum computer calculates is actually[br]the correct answer or that he does not 0:30:58.240,0:31:01.790 just produce some garbage values? And[br]that's a very interesting question, 0:31:01.790,0:31:07.320 actually. And the way they did it here is[br]by extrapolation. So instead of, for 0:31:07.320,0:31:11.820 example, solving the full circuits, so[br]that contains all of the connections and 0:31:11.820,0:31:17.240 all of the gates of the full algorithm,[br]they created simplified circuits in two 0:31:17.240,0:31:21.570 different ways. So, for example, they cut[br]they cut some of the connections between 0:31:21.570,0:31:26.330 the Qubits and the algorithms, so that the[br]problem space would become a bit smaller 0:31:26.330,0:31:30.350 or in the other case, with the allied[br]circuit, they just changed the operations 0:31:30.350,0:31:34.710 in order to allow for some shortcuts in[br]the classical computation of the classical 0:31:34.710,0:31:39.820 simulation of the algorithm. So in both[br]cases, they were able to then verify the 0:31:39.820,0:31:43.600 result of the quantum computation with this[br]classical simulation performed on a 0:31:43.600,0:31:48.850 supercomputer. And then they basically[br]just did this for a larger and larger 0:31:48.850,0:31:53.770 number of Qubits. They plotted the[br]resulting curve and they extrapolated that 0:31:53.770,0:31:58.960 to the supremacy regime to see that. OK.[br]Based on the error models that they 0:31:58.960,0:32:02.760 developed, based on the simulation, they[br]can with a certain confidence, of course, 0:32:02.760,0:32:06.930 say that probably the quantum computer is[br]doing the right thing even in the 0:32:06.930,0:32:11.920 supremacy regime, even though we can't[br]they cannot verify it using the classical 0:32:11.920,0:32:18.720 simulations. And in case the quantum[br]computer did wrong still, they have also 0:32:18.720,0:32:22.730 archive to the results. And maybe ten[br]years when we have better supercomputers, 0:32:22.730,0:32:28.240 we might be able to just go back to them[br]and then verify them against the 53, 53 0:32:28.240,0:32:31.920 Qubits processor here, by which time, of[br]course, they might already have like a 0:32:31.920,0:32:39.610 larger quantum processor again. So the key[br]results of this, I would say, are that for 0:32:39.610,0:32:43.940 the first time they show that really[br]quantum computer can beat a classical 0:32:43.940,0:32:49.220 computer, even though it is at a very[br]artificial and probably not very useful 0:32:49.220,0:32:53.280 problem. And what the experiment also[br]shows is that really, I would say an 0:32:53.280,0:32:59.060 astounding level of control of such a[br]large scale on medium size quantum 0:32:59.060,0:33:05.410 processor, because even five years ago,[br]six years ago, 2012, 2013, the systems 0:33:05.410,0:33:10.610 that we worked with mostly consisted of[br]three or four Qubits and we could barely 0:33:10.610,0:33:16.000 fabricate the chips and manipulate them to[br]get like algorithms running. And now if I 0:33:16.000,0:33:21.090 see like a 50 tree Qubit processor with[br]such a high degree of control and fidelity 0:33:21.090,0:33:25.530 there, I can really say that is really an[br]amazing progress in the last five years 0:33:25.530,0:33:30.600 that what was achieved, especially by the[br]Google Martinez team here. And I think it 0:33:30.600,0:33:34.200 is a very good wild milestone on the way[br]to fully work on quantum computer because 0:33:34.200,0:33:38.950 it nicely shows the limitations of the[br]current system and gives a good direction 0:33:38.950,0:33:44.590 on new areas of research, for example, an[br]error correction, where we can improve the 0:33:44.590,0:33:50.150 different aspects of the quantum[br]processor. The research has also been 0:33:50.150,0:33:55.330 criticized from various sides, so I just[br]want to iterate a few of the points 0:33:55.330,0:34:00.050 here. One of the criticisms is, of course,[br]that it doesn't do anything useful. So 0:34:00.050,0:34:05.840 there's really no applicability of this[br]experiment and why that's true. It's, of 0:34:05.840,0:34:11.450 course, very difficult to go from like a[br]basic, very simple quantum process of two 0:34:11.450,0:34:15.450 Qubits to a system that can really [br]factorize prime numbers or do anything 0:34:15.450,0:34:20.109 useful. So we will always need to find[br]problems that are both hard enough so that 0:34:20.109,0:34:24.210 we can solve them in a reasonable[br]timeframe. A couple of years, for example, 0:34:24.210,0:34:28.540 that still proved the progress that we[br]make on the road to quantum computing. So 0:34:28.540,0:34:33.070 in this sense, while quantum supremacy[br]does not really show anything useful in 0:34:33.070,0:34:37.690 terms of computation that is done. I think[br]it is still a very good problem as a 0:34:37.690,0:34:41.580 benchmark for any kind of quantum[br]processor, because it requires that you 0:34:41.580,0:34:46.110 have very good control over your system[br]and that you can run such a number of 0:34:46.110,0:34:50.621 gates at a very high fidelity, which is[br]really currently, I would say, the state 0:34:50.621,0:34:56.899 of the art. The research also took, took [br]some shortcuts. For example, they used 0:34:56.899,0:35:00.290 like a two Qubits, quantum gates, which [br]are not, as we call them, canonical 0:35:00.290,0:35:04.230 gates, which might be problematic [br]because if you want to run a quantum 0:35:04.230,0:35:07.900 algorithm on the system, you need[br]to implement certain quantum gates that 0:35:07.900,0:35:12.240 you need for that. And since they only[br]have like non canonical gates here, which 0:35:12.240,0:35:16.740 are still universal, by the way, they[br]could not do that directly, but with some 0:35:16.740,0:35:21.060 modification of the system, that should[br]also be possible. And the last criticism 0:35:21.060,0:35:25.640 might be that, okay, here you have a[br]problem that was engineered to match a 0:35:25.640,0:35:31.540 solution, which is of course that, okay,[br]we need some solution, as I said, some 0:35:31.540,0:35:36.520 problem that we can't realistically solve [br]on a such a system. I think, though, also 0:35:36.520,0:35:40.230 like the other points, if you want to[br]build a large scale quantum processor, you 0:35:40.230,0:35:45.030 need to define reasonable milestones and[br]having such a benchmark that other groups, 0:35:45.030,0:35:49.720 for example, can also run that process[br]against is a very good thing because it 0:35:49.720,0:35:54.200 makes the progress visible and also makes[br]it easy to compare how different groups 0:35:54.200,0:35:59.820 or are companies or organizations [br]are are at competing on the 0:35:59.820,0:36:12.040 number of Qubits under control they have[br]about them. So, if you want to make a more 0:36:12.040,0:36:16.770 kind of Moore's Law for quantum computing,[br]there would be several possibilities that 0:36:16.770,0:36:22.550 you could do. Here I show you, for[br]example, the number of Qubits that have 0:36:22.550,0:36:28.620 been realized for superconducting systems[br]over the years. This is, of course 0:36:28.620,0:36:32.340 incomplete because it could like the[br]number of Qubits alone doesn't tell you 0:36:32.340,0:36:36.810 much about your system. I mean, we could[br]do a Qubit chip of 1000 or 10000 Qubits 0:36:36.810,0:36:41.230 today. But if you don't have the[br]connectivity and don't have to controllability 0:36:41.230,0:36:45.119 of individual Qubits, then this[br]chip wouldn't be good. So there are other 0:36:45.119,0:36:49.210 things, that we also need to take into[br]account here. As I said, just as like the 0:36:49.210,0:36:53.550 coupling between individual Qubits and the[br]coherence time and the fidelity of the 0:36:53.550,0:37:00.010 Qubit operations. So this is really just[br]one one very small aspect of this whole 0:37:00.010,0:37:03.470 whole problem space. But I think it shows[br]nicely that in the last years there was 0:37:03.470,0:37:07.790 really tremendous progress in terms of the[br]power of the superconducting systems, 0:37:07.790,0:37:15.350 because the original Qubit, which was[br]developed in at NYC in Japan by 0:37:15.350,0:37:20.710 Professor Nakamura, was done in like[br]around 2000. So that very, very bad 0:37:20.710,0:37:24.920 coherence time, very bad properties. But[br]still it showed for the first time that he 0:37:24.920,0:37:28.970 could coherently control such a system.[br]And then it didn't take long for other 0:37:28.970,0:37:32.390 groups, for example, to Quantronics[br]Group and so Saclay, to pick up on this 0:37:32.390,0:37:37.480 work and to do to keep improving it. So[br]after a few years, we already had Qubits 0:37:37.480,0:37:42.250 of a few hundred or even a microsecond of[br]coherence time, which was like in like 0:37:42.250,0:37:46.100 three or orders of magnitude better than[br]what we had before. And there were other 0:37:46.100,0:37:51.490 advances then made by groups in the US,[br]for example, in Yale, the ShowCoupLab, 0:37:51.490,0:37:56.010 which developed new Qubit architectures[br]that allowed us to couple the Qubits more 0:37:56.010,0:37:59.570 efficiently with each other and to again[br]have better control of them manipulating 0:37:59.570,0:38:04.910 them. And then there's also groups like[br]the research group at IBM or companies 0:38:04.910,0:38:09.410 like WeGetty that took again these[br]ideas and that added engineering and their 0:38:09.410,0:38:14.130 own research on top of that in order to[br]make the systems even better. So in 2018, 0:38:14.130,0:38:19.610 we already had systems with 17 or 18[br]Qubits in them. And now with this Google 0:38:19.610,0:38:25.440 and UC Santa Barbara work, we have the[br]first systems with more than 50 Qubits 0:38:25.440,0:38:32.510 after not even 20 years, which I think is[br]quite some progress in this area. And of 0:38:32.510,0:38:38.520 course, if you ask me how close we are to[br]and actually working quantum computer, 0:38:38.520,0:38:44.860 it's still very difficult to say, I find,[br]because we've proven the group prove the 0:38:44.860,0:38:49.840 quantum supremacy for its randomized[br]algorithm. But in order to do something 0:38:49.840,0:38:56.370 applicable or useful with such a quantum[br]system, I think we need like at least 0:38:56.370,0:39:03.440 again, 50 maybe 200 additional Qubits and[br]a larger number of Qubit operations. But 0:39:03.440,0:39:07.299 it's really hard to say. That's why I also[br]say don't believe in this chart because 0:39:07.299,0:39:11.890 there's also, of course, a lot of work in[br]the theory of quantum algorithms, because 0:39:11.890,0:39:16.360 up to now we are still discovering new[br]approaches of doing quantum simulations 0:39:16.360,0:39:19.800 for examples. And right now, there are a[br]lot of research groups that are looking 0:39:19.800,0:39:24.120 for ways to make these medium scale[br]quantum computers. So quantum computers 0:39:24.120,0:39:29.940 with 50 or 100 Qubits already useful for[br]using quantum simulations. So it's really 0:39:29.940,0:39:35.420 an interplay between what the theory can[br]give us in terms of quantum algorithm and 0:39:35.420,0:39:40.070 what in terms of experimental realization[br]we can build as a quantum processor. So in 0:39:40.070,0:39:44.390 my opinion, quantum simulation will[br]definitely be something that where we will 0:39:44.390,0:39:49.390 see the first applications in the next. I[br]would say three to five years. Other 0:39:49.390,0:39:54.900 things, optimizations. I have to admit I[br]am less an expert and I think they're a 0:39:54.900,0:39:58.810 bit more complex. So we will probably see[br]the first applications in those areas a 0:39:58.810,0:40:04.011 bit later. And the big motivation for like[br]the three letter agencies always is, 0:40:04.011,0:40:10.350 of course, the factoring out the breaking[br]of cryptosystems, which is the most 0:40:10.350,0:40:14.619 challenging one, though, because in order[br]to do that, you would both need very large 0:40:14.619,0:40:19.850 numbers of Qubits. So at least 8000[br]Qubits for an 8000 bits RSA key, for 0:40:19.850,0:40:23.831 example. And you would also need a very[br]large amount of Qubit operations because 0:40:23.831,0:40:29.520 you need to run the sure operation. And[br]that involves a lot of steps for the 0:40:29.520,0:40:34.350 quantum processor. And so to say the most,[br]I would say from my perspective 0:40:34.350,0:40:39.550 unrealistic application of superconducting[br]quantum processes in the next year. But I 0:40:39.550,0:40:43.320 think, if somebody would build a quantum[br]computer, maybe we would also not just 0:40:43.320,0:40:52.749 know about it. So who knows? So to[br]summarize, quantum computers, quantum 0:40:52.749,0:40:57.020 processors are getting really, seriously[br]complex and very impressive. So we have 0:40:57.020,0:41:02.260 seen tremendous progress in the last five[br]years. I still think that we are like five 0:41:02.260,0:41:06.840 years away from building really practical[br]quantum computers and there are still some 0:41:06.840,0:41:11.510 challenges. For example, an error[br]correction in the Quantum Gatefidelity and 0:41:11.510,0:41:15.500 indeed, again, general architecture of[br]these systems that we need to overcome. 0:41:15.500,0:41:18.690 And they might also be some challenges[br]which we haven't even identified yet with 0:41:18.690,0:41:22.750 which we might only encounter at a later[br]stage when trying to build really large 0:41:22.750,0:41:28.390 scale quantum processors. And as a last[br]point, I just want to stress again, that 0:41:28.390,0:41:33.700 quantum computing research is not only[br]done by Google or by IBM, there a lot 0:41:33.700,0:41:37.540 of groups in the world involved in this[br]kind of research, both in theory and an 0:41:37.540,0:41:43.030 experiment. And as I said before, a lot of[br]the breakthroughs that we use today for 0:41:43.030,0:41:47.410 building quantum processes were done in[br]very different places like Japan, Europe, 0:41:47.410,0:41:52.690 USA. So it's really, I would say, a global[br]effort. And you should also, when you 0:41:52.690,0:41:58.281 look, when you see this marketing PR that[br]companies like Google and IBM do, maybe 0:41:58.281,0:42:03.869 not believe all of the hype they're[br]creating and keep on down to earth views, 0:42:03.869,0:42:10.950 so to say, of the limits and the potential[br]of quantum computing. So that's it. And I 0:42:10.950,0:42:13.940 would be happy to take on your questions[br]now. And if you have any 0:42:13.940,0:42:19.160 feedback, there's also my Twitter handle[br]and my email address. And I think we also 0:42:19.160,0:42:23.540 have some time for questions here right[br]now. Thank you. 0:42:23.540,0:42:32.430 Applause 0:42:32.430,0:42:37.050 Herald: Thank you, Andreas. We have almost[br]20 minutes for Q and A. If you're leaving 0:42:37.050,0:42:42.560 now, please do so very quietly and if you[br]can avoid it, just don't do it. Thank you. 0:42:42.560,0:42:47.390 Okay. Q and A. You know the game. There's[br]eight microphones in this room, so just 0:42:47.390,0:42:53.580 queue behind them and we will do our best[br]to get everyone sorted out sequentially. 0:42:53.580,0:42:58.490 We will start with a question[br]from the Internet. 0:42:58.490,0:43:01.859 Signal-Angel: Thank you. Do you have[br]information about the energy consumption 0:43:01.859,0:43:07.840 of a quantum computer[br]over the calculation power? 0:43:07.840,0:43:11.729 Andreas: Yeah, that's an interesting[br]point. I mean, for superconducting quantum 0:43:11.729,0:43:16.490 computers, there are like several costs[br]associated. I think right now the biggest 0:43:16.490,0:43:20.970 cost is probably of keeping the system[br]cooled down. So as that you need very low 0:43:20.970,0:43:25.570 temperatures, 20 or 10 millikelvin. In[br]order to achieve that, you need the so- 0:43:25.570,0:43:29.490 called delusion crisis that and these[br]systems that consume a lot of energy and 0:43:29.490,0:43:36.300 also materials like helium mixtures, which[br]are expensive and like maybe not so well, 0:43:36.300,0:43:40.510 kind of like a real material right now. I[br]think that would be the biggest 0:43:40.510,0:43:46.290 consumption in terms of energy use. I[br]honestly don't have so much of an idea. I 0:43:46.290,0:43:50.060 mean, the manipulation of the Qubit system[br]is done via microwaves and the power that 0:43:50.060,0:43:54.070 goes into the system is very small[br]compared to any of the power that we use 0:43:54.070,0:43:58.330 for cooling the system. So I would say for[br]the foreseeable future, the power 0:43:58.330,0:44:02.370 consumption should be dominated by like[br]the cooling and the setup costs and the 0:44:02.370,0:44:06.180 cost of the electronics as well. So the[br]classical electronics that that controls 0:44:06.180,0:44:10.440 the Qubit, which can also be quite[br]extensive for large system. So the Qubit 0:44:10.440,0:44:14.690 chip itself should be very should be[br]really negligible in terms of energy 0:44:14.690,0:44:18.200 consumption.[br]Herald: Thank you. Microphone number one 0:44:18.200,0:44:22.560 please.[br]Mic 1: Hello. I have a question in regards 0:44:22.560,0:44:28.320 to quantum simulation. So I would have[br]thought that with 53 Qubits, 0:44:28.320,0:44:34.900 there would already be something[br]interesting to do, since I think their 0:44:34.900,0:44:41.680 border the limit for more or less exact[br]quantum chemistry calculations on 0:44:41.680,0:44:46.930 classical computers is that there are 10[br]to 20 particles. So is there a more 0:44:46.930,0:44:53.720 complicated relation from particles to[br]Qubits that's missing here or what's the 0:44:53.720,0:44:57.570 problem?[br]Andreas: Yeah. So in the paper I couldn't 0:44:57.570,0:45:02.920 find an exact reason why they choose this[br]problem. I think there are probably two 0:45:02.920,0:45:09.960 aspects. One is that you don't have in the[br]system the like arbitrary Qubit control. 0:45:09.960,0:45:14.940 So to say you cannot run like any[br]Hamiltonian or quantum algorithm that you 0:45:14.940,0:45:18.660 want. You are like limited in terms of[br]connectivity. So it's possible that they 0:45:18.660,0:45:25.100 were not able to run any quantum algorithm[br]for simulation, which was not easy to run 0:45:25.100,0:45:28.810 also on a classical system, you know, so.[br]But I'm really not not sure why they 0:45:28.810,0:45:33.000 didn't. I think just if they would have a[br]have had this chance to do a quantum 0:45:33.000,0:45:36.630 simulation, they would probably have done[br]that instead, because that's, of course, 0:45:36.630,0:45:41.530 more impressive than randomization or[br]randomized algorithms. So because they 0:45:41.530,0:45:45.869 didn't do it, I think it was just probably[br]too complicated or not possible to realize 0:45:45.869,0:45:49.949 on the system. Yeah. Okay. So it's this.[br]But again, I don't know for sure yet. 0:45:49.949,0:45:52.789 Thank you.[br]Herald: Yes, and also speaking as a sometimes 0:45:52.789,0:45:58.270 quantum chemist, you can't directly map[br]Qubits to to atoms. They're not two level 0:45:58.270,0:46:02.880 systems. And you don't I mean, you usually[br]also simulate electrons and not just 0:46:02.880,0:46:07.310 atoms, but I'm not a speaker. We can[br]discuss later. Maybe microphone number two 0:46:07.310,0:46:10.869 please.[br]Mic 2: Thanks. Can you compare this 0:46:10.869,0:46:16.530 classic or general quantum computer to the[br]one by D-wave? That's one of the quantum 0:46:16.530,0:46:22.480 computers by a AWS offered. They have two[br]thousand Qubits or something. 0:46:22.480,0:46:26.050 Andreas: Yeah, that's a very interesting[br]question. So D-wave system is this so- 0:46:26.050,0:46:32.020 called adiabatic quantum computer, to[br]my knowledge. So this in this case the 0:46:32.020,0:46:36.750 computation works a bit differently. It's[br]the normal with this quantum computer that 0:46:36.750,0:46:40.560 Google produced. You have a gate sequence[br]that you run on your input Qubits and then 0:46:40.560,0:46:44.290 you get a result that you read out. With[br]the D-wave system it's more that you like 0:46:44.290,0:46:48.560 engineer like in Hamiltonian, which is[br]also which also consists of local 0:46:48.560,0:46:53.160 interactions between different Qubits. And[br]then you slowly changed this Hamiltonian 0:46:53.160,0:46:58.320 in order to like change to the ground[br]state of the system to a solution of a 0:46:58.320,0:47:02.680 problem that you're looking for. So. So[br]it's a different approach to quantum 0:47:02.680,0:47:09.520 computation. They also claimed that they[br]can can achieve what I did, achieve a 0:47:09.520,0:47:14.270 quantum supremacy, I think in a different[br]way for like an optimization problem. But 0:47:14.270,0:47:19.859 to my knowledge, the proof they have is[br]less rigid probably than, what the Google 0:47:19.859,0:47:23.610 Group produced here. So but again, I'm not[br]like an expert on that, a bit of quantum 0:47:23.610,0:47:29.800 computing. So I'm more like a gate based[br]person. So, yeah, I think though, the 0:47:29.800,0:47:34.570 proof that here the Google show is more[br]convincing in terms of like reproduce 0:47:34.570,0:47:40.300 reproducibility and really make the proof[br]that you are actually doing something that 0:47:40.300,0:47:47.490 cannot be done on a classical computer.[br]D-Wave will see the different view though. 0:47:47.490,0:47:53.850 Herald: All right. Let's go to the back.[br]Number seven, please. Hello. 7. You just 0:47:53.850,0:47:58.630 waved to me.[br]Mic 7: Hey, uh, hello. Uh, I was reading 0:47:58.630,0:48:06.369 that earlier this year IBM released the[br]first commercial Q one system or whatever 0:48:06.369,0:48:11.760 the name is. And you were mentioning[br]before to keep our expectations down to 0:48:11.760,0:48:18.520 Earth. So my question is, what kind of[br]commercial expectations is IBM actually 0:48:18.520,0:48:22.921 creating?[br]Andreas: Mm hmm. So I spoke to some 0:48:22.921,0:48:30.369 companies here in Germany that are[br]collaborating with IBM or D-Wave or Google 0:48:30.369,0:48:35.290 as well. And to ask what they're actually[br]doing with the quantum computers. They are 0:48:35.290,0:48:41.090 the the companies offer. And I think the[br]answer is that right now, a lot of 0:48:41.090,0:48:45.500 commercially, a lot of companies are[br]investigating this as something that could 0:48:45.500,0:48:50.670 potentially be very useful or very[br]relevant in five to 10 years. So they want 0:48:50.670,0:48:54.751 to get some experience and they want to[br]start collaborating. I don't think, at 0:48:54.751,0:49:00.040 least I don't know any reproduction use of[br]these systems where the quantum computer 0:49:00.040,0:49:05.130 would do some calculations, that would not[br]be doable on a classical system. But 0:49:05.130,0:49:08.560 again, I don't have a full overview of[br]that. I think now it's mostly for 0:49:08.560,0:49:12.890 experimentation and forgetting to notice[br]systems. I think the companies or most of 0:49:12.890,0:49:17.260 the customers there probably expect that[br]in five years or 10 years, the system will 0:49:17.260,0:49:21.030 systems will really be powerful enough to[br]do some useful computations with them as 0:49:21.030,0:49:24.520 well.[br]Herald: Thanks. All right. The Internet, 0:49:24.520,0:49:27.270 please.[br]Signal-Angel: With a quantum computer, you 0:49:27.270,0:49:32.260 can calculate things in parallel. But[br]there is this usability requirement. So 0:49:32.260,0:49:36.820 how much faster is a quantum[br]computer at the end of the day? 0:49:36.820,0:49:42.440 Andreas: Mm hmm. Yeah, it's true, so that[br]if you want to and, if you want to realize 0:49:42.440,0:49:46.700 classical algorithm, you have to do it in[br]a reversible way. But to my knowledge, you 0:49:46.700,0:49:51.920 can from an efficiency perspective,[br]implement any classical non reversible 0:49:51.920,0:49:59.210 algorithm as a reversible algorithm[br]without loss in complexity. So you can 0:49:59.210,0:50:02.510 have also like for a reversible[br]computation, you have universal gaits like 0:50:02.510,0:50:07.291 the control not gate that you can use to[br]express any logic function that you 0:50:07.291,0:50:12.580 require. You might need some additional[br]Qubits in compared to the amount of the 0:50:12.580,0:50:16.220 classical bits that you need for the[br]computation. But in principle, there is 0:50:16.220,0:50:20.009 nothing that keeps you from implementing[br]any classical function on a quantum 0:50:20.009,0:50:24.670 computer. In terms of actual runtime, of[br]course it depends on how fast you can run 0:50:24.670,0:50:28.760 individual operations. Right now, a single[br]Qubits operation, for example, on this 0:50:28.760,0:50:34.820 Google machine takes about I think 20 to[br]40 nanoseconds. So in that sense, the 0:50:34.820,0:50:39.010 quantum computers are probably much slower[br]than classical computers. But the idea is 0:50:39.010,0:50:43.109 anyway that you do only really the[br]necessary computations that you can't do 0:50:43.109,0:50:46.450 on a classical machine, on a quantum[br]computer and anything else you can do on a 0:50:46.450,0:50:52.410 normal classical system. So the quantum[br]process in this sense is only like a like 0:50:52.410,0:50:56.990 inside a core processor, like a GPU, in[br]that sense, I would say. 0:50:56.990,0:50:59.850 Herald: All right. Microphone number four,[br]please. 0:50:59.850,0:51:05.270 Mic 4: On the slide that shows Richard[br]Feynman, you said that quantum computers 0:51:05.270,0:51:14.020 were invented to simulate quantum systems.[br]And can you please elaborate on that? 0:51:14.020,0:51:17.760 Herald: You went past, huh?[br]Andreas: Yeah. So I don't have to link to 0:51:17.760,0:51:21.830 the lecture here. Unfortunately, the link[br]is broken, but you can find that online. 0:51:21.830,0:51:27.130 It's a 1982 lecture from Feynman, where he[br]discusses like how you would actually go 0:51:27.130,0:51:32.780 about simulating a quantum system, because[br]as we have shown like the if you want to 0:51:32.780,0:51:36.849 simulate a full quantum system, you need to[br]simulate the density matrix of the system 0:51:36.849,0:51:42.414 and that takes about that take, it takes[br]an exponential amount of memory and 0:51:42.414,0:51:46.849 computation in terms of like the number of[br]Qubits or quantum degrees of freedom that 0:51:46.849,0:51:51.590 you want to simulate. And with a classical[br]Turing machine, you couldn't do that in an 0:51:51.590,0:51:55.980 efficient way because every time you add a[br]single Qubit, you basically quadruple your 0:51:55.980,0:52:00.320 computational requirement. And that's[br]really where the idea came from. I think 0:52:00.320,0:52:04.860 from Feynman to think about a computing[br]system that would use quantum mechanics in 0:52:04.860,0:52:09.109 order to be able to do these kind of[br]simulations, because he saw probably that 0:52:09.109,0:52:13.260 for large quantum systems it would never[br]be possible to run, at least with our 0:52:13.260,0:52:16.190 current understanding of classical[br]computing. It would never be possible to 0:52:16.190,0:52:20.390 run a quantum simulation of a quantum[br]system on a classical computer in an 0:52:20.390,0:52:23.550 efficient way. Does that answer the[br]question? 0:52:23.550,0:52:25.380 Mic 4: Yeah.[br]Andreas: Okay. 0:52:25.380,0:52:28.290 Herald: All right. Microphone eight,[br]please. 0:52:28.290,0:52:35.820 Mic 8: As a physicist who's now doing[br]analog circuit design. I'm kind of 0:52:35.820,0:52:41.620 wondering why all the presentations about[br]quantum computers always use stage zero 0:52:41.620,0:52:45.880 and 1 and not multiple states. Is that a[br]fundamental limitation or is that just 0:52:45.880,0:52:49.530 just a simplification for the sake of the[br]presentation? 0:52:49.530,0:52:52.730 Andreas: So you mean why you don't use[br]like higher Qubit states or like... 0:52:52.730,0:52:57.540 Mic 8: Multi valued logic or even[br]continuous states? 0:52:57.540,0:53:01.330 Andreas: So in principle, the quantum bits[br]that we're using, they don't they're not 0:53:01.330,0:53:05.090 really two level systems. So there is not[br]only level zero and one, but also level 0:53:05.090,0:53:10.520 two tree and so on. You could use them, of[br]course, but the computational power of the 0:53:10.520,0:53:15.609 system is given as the number of states,[br]or like m for example, race to the power 0:53:15.609,0:53:20.240 of the number of Qubits. So M to the power[br]of N. So in that sense, if you add like 0:53:20.240,0:53:26.750 another state, you only change like. Like[br]a small affected and adding another Qubit. 0:53:26.750,0:53:30.550 So it's usually not very interesting to[br]add more states. What he would do instead, 0:53:30.550,0:53:35.380 is just add more Qubits to your system.[br]And for continuous variable quantum 0:53:35.380,0:53:39.660 mechanic quantum computation. I think[br]there is some use cases where this might 0:53:39.660,0:53:43.870 outperform like the digital quantum[br]computers, especially if you can engineer 0:53:43.870,0:53:48.980 your system to like mimic the Hamiltonian[br]of the system that you want to simulate. 0:53:48.980,0:53:54.050 So I think in this sense, in these cases,[br]it makes a lot of sense. For other cases 0:53:54.050,0:53:57.720 where you say, OK, you want to run a[br]general quantum computation, then like 0:53:57.720,0:54:01.179 such a digital quantum computer is[br]probably the best solution. And you could 0:54:01.179,0:54:07.820 also just add that run like a continuous[br]simulation of a quantum system on such a 0:54:07.820,0:54:13.460 gate based quantum system, just like the[br]linearly in the same order of complexity, 0:54:13.460,0:54:17.850 I would say. Does that answer the[br]question? 0:54:17.850,0:54:23.470 Mic 8: I think I delude myself to have[br]understood that the non diagonal elements 0:54:23.470,0:54:28.450 in the density matrix grow much faster[br]than the number of states in any and any 0:54:28.450,0:54:32.170 diagonal matrix element.[br]Andreas: I guess you could say like that. 0:54:32.170,0:54:37.530 Yeah, I have to think about.[br]Herald: All right. Number three, please. 0:54:37.530,0:54:43.320 Mic 3: What do you have to say about the[br]scepticism of people like Nikolai that 0:54:43.320,0:54:51.100 claim that inherent nice will be a[br]fundamental problem in scaling this 0:54:51.100,0:54:54.820 quantum computers?[br]Andreas: I mean, it's a valid concern, I 0:54:54.820,0:55:01.410 think. As of today, we don't have even for[br]a single Qubit shown error correction. 0:55:01.410,0:55:04.840 There are some first experiments, for[br]example, by the ShowCoup Lab in Yale that 0:55:04.840,0:55:08.970 showed some of the elements of error[br]correction for a single Qubit system, but 0:55:08.970,0:55:15.230 we haven't even managed today to keep a[br]single Qubit alive indefinitely. So that's 0:55:15.230,0:55:19.220 why I would say it's an open question.[br]It's a valid criticism. I think the next 0:55:19.220,0:55:23.160 five years will show if we are actually[br]able to run this quantum errors and if our 0:55:23.160,0:55:26.310 error models themselves are correct[br]because they only correct for certain 0:55:26.310,0:55:30.809 errors or if there's anything else that[br]keeps us from like building a large scale 0:55:30.809,0:55:35.400 system. So I think it's a totally valid[br]point. 0:55:35.400,0:55:40.990 Herald: Microphone five, please.[br]Mic 5: There has been a study on 0:55:40.990,0:55:48.830 factarising on adiabatic machines, which[br]requires a lock squared N Qubits while 0:55:48.830,0:55:58.260 Shor requires Log N. But as the adiabatic[br]systems have much higher Qubit numbers, 0:55:58.260,0:56:05.140 they were able to factorize on these[br]machines, much larger numbers than on the 0:56:05.140,0:56:11.870 normal devices. And that's something that[br]never shows up in the discussion. Do you 0:56:11.870,0:56:17.320 want to comment on that? Have you read the[br]study? What do you think? Are adiabatic 0:56:17.320,0:56:23.200 machines, bogus? Or, is that worth while[br]resolved? 0:56:23.200,0:56:26.130 Andreas: I'm not. Yeah, as I said, like an[br]expert at adiabatic quantum 0:56:26.130,0:56:31.980 computing. I know that there were some[br]like studies or investigations of the 0:56:31.980,0:56:37.690 D-wave system. Like I haven't read this[br]particular study about factorization. I 0:56:37.690,0:56:40.940 think adiabatic quantum computing is a[br]valid approach as well to quantum 0:56:40.940,0:56:48.520 computing. I just I'm not just just not[br]sure if currently like the results were 0:56:48.520,0:56:54.849 like shown with the same amount of like[br]rigidity or like rigid proves like for the 0:56:54.849,0:56:58.380 gate based quantum computer. But yeah, I'm[br]I really would have to look at the study 0:56:58.380,0:57:02.730 to to see that.[br]Herald: Can you maybe quickly say the 0:57:02.730,0:57:09.270 authors. So it's on the record. Yeah. If[br]your mike is still on number five. 0:57:09.270,0:57:14.140 Mic 5: Sorry, I don't.[br]Herald: Okay, no problem. Thank you. All 0:57:14.140,0:57:15.790 right.[br]Andreas: But yeah, I don't think adiabatic 0:57:15.790,0:57:19.660 quantum computing is like and I[br]think adiabatic quantum computing is a 0:57:19.660,0:57:24.190 valid choice or valid approach for doing[br]quantum computation as well. 0:57:24.190,0:57:29.339 Mic 5: So I can give you that. I can[br]search for the authors later and give it 0:57:29.339,0:57:30.759 to you.[br]Andreas: Okay. Okay. It would be great. 0:57:30.759,0:57:33.029 Thank you.[br]Herald: Thank you. Microphone four, 0:57:33.029,0:57:36.199 please.[br]Mic 4: What do you say about IBM's claim 0:57:36.199,0:57:41.100 that Google's supremacy claim is invalid[br]because the problem was not really hard? 0:57:41.100,0:57:45.840 Andreas: Yeah. So basically IBM, I think[br]said, okay, if you do some optimizations 0:57:45.840,0:57:49.810 on the way you simulate the systems, then[br]you can reduce this computation time from 0:57:49.810,0:57:55.290 10000 years to like maybe a few hours or[br]so. I think it's, of course, valid. It 0:57:55.290,0:57:59.910 might be a valid claim. I don't know if it[br]really invalidates the result because as I 0:57:59.910,0:58:04.760 said, like the computational power of like[br]the classical systems, they will also will 0:58:04.760,0:58:09.700 also increase in the coming years. Right[br]now, you could say that maybe if we 0:58:09.700,0:58:14.839 haven't achieved quantum supremacy in[br]regards to elect 2019 hardware, then maybe 0:58:14.839,0:58:19.220 we should just like look at the 2015[br]hardware and then we can say, okay, there, 0:58:19.220,0:58:24.369 probably we achieved that. In any case, I[br]think the most what's most impressive 0:58:24.369,0:58:28.930 about this result for me is not like, if[br]we are really in the supremacy regime or 0:58:28.930,0:58:34.200 maybe not. That's really the amount of..,[br]the degree of controlability of the 0:58:34.200,0:58:37.290 Qubits system that this group has[br]achieved. I think that's really the 0:58:37.290,0:58:40.920 important point here, regardless of[br]whether they actually achieved the 0:58:40.920,0:58:46.190 supremacy or not. Because it shows that[br]these kind of systems seem to be a good 0:58:46.190,0:58:50.250 architecture choice for building large[br]scale quantum processes. And this alone is 0:58:50.250,0:58:54.680 very valuable, I think, as a guide to[br]future research direction, regardless of 0:58:54.680,0:58:59.750 whether this is actually, you know, they[br]achieved this or not. Yeah, but yeah, I 0:58:59.750,0:59:05.760 can understand, of course, the criticism.[br]Mic 4: OK. One thing. The article is 0:59:05.760,0:59:11.099 called Quantum Annealing for Prime[br]Factorization appeared in Nature in 0:59:11.099,0:59:18.739 December 18. Authors Jiang, A. Britt, Alex[br]J. McCaskey, S. Humble and Kais. 0:59:18.739,0:59:22.190 Andreas: Okay, great. I think we'll have a[br]look at that again. Thanks. 0:59:22.190,0:59:25.359 Herald: All right. Microphone 6, do you[br]have a short question? 0:59:25.359,0:59:33.990 Mic 6: Yeah, hopefully. It is known that[br]it is not very easy to understand how 0:59:33.990,0:59:41.000 large quantum superposition goes into a[br]macroscopic state. So in the macroscopic 0:59:41.000,0:59:47.340 physical description. So apparently there[br]are a couple of things not understood. So 0:59:47.340,0:59:51.970 is there anything you know about when you[br]go two thousand, ten thousand, million 0:59:51.970,1:00:00.310 Qubits, could you expect the quantum[br]behavior to break down? Are there any 1:00:00.310,1:00:07.150 fundamental argument that this will not[br]happen or is this not a problem considered 1:00:07.150,1:00:09.760 recently?[br]Andreas: Huh, Okay. I'm not sure if I 1:00:09.760,1:00:13.089 fully understand the question. It's mostly[br]about like if you say like quantum 1:00:13.089,1:00:17.820 mechanics or some like scale variance so[br]that if you go to a certain scale and some 1:00:17.820,1:00:22.160 time, at some point you have like a[br]irreversibility or like a something like 1:00:22.160,1:00:27.010 that. Yeah. I mean, I think that a large[br]quantum systems that occur naturally, I 1:00:27.010,1:00:30.210 don't know. I like Bose Einstein[br]condensate, for example, has a lot of 1:00:30.210,1:00:33.640 degrees of freedom that are not[br]controlled, of course, but that also 1:00:33.640,1:00:39.420 quantum mechanical and there it seems to[br]work. So personally, I would think that 1:00:39.420,1:00:43.780 there is no such limit. But I mean, who[br]knows? It's like that's why we do like 1:00:43.780,1:00:47.960 experimental physics. So we will see as if[br]we reached it. But from like the theory of 1:00:47.960,1:00:52.099 quantum mechanics right now, there is no[br]indication that this should be such a 1:00:52.099,1:00:55.670 limit to my knowledge. [br]Herald: All right, so maybe we will see 1:00:55.670,1:00:57.870 you again in five years.[br]Andreas: Yeah. 1:00:57.870,1:01:00.490 Herald. So please thank Andreas, until I[br]ask once again. Thanks. 1:01:00.490,1:01:02.184 Applause 1:01:02.184,1:01:05.654 36c3 postroll music 1:01:05.654,1:01:28.000 Subtitles created by c3subtitles.de[br]in the year 2020. Join, and help us!