Introduction to Quantum Computing | Koen Bertels | TEDxAntwerp



the wonderful world of quantum computing that's basically what I would like to talk about and if I can yeah use this device would be would be great what you see is I'm cheap that we've started developing in India no okay sorry in Delft with my colleague Leo DeCarlo he sounds Italian but it's actually Argentinean and with whom I'm working very intensively with the two teams that we have to actually build a first prototype of a quantum computer and I will hopefully explain a little bit in detail what we understand by that so let's let's start by looking at a classical machine yeah and here we see a diagram which which is being drawn by by someone where we have memory and processor and we have instructions in memory residing and of course we transport now instruction after instruction into the processor and we we decode the instruction which is an addition then we get the operands so that 3 and the 6 in this case from memory and we put them again in the local register we compute the result and we write back you know the result which again in this place will be in the same location as where I retrieved the 3 because that's what my instruction is actually telling me how it should be done so this is how in principle every single computer yeah what size and and when it was built basically works and I will try to explain a little bit what happened also let's say in the last 22 maybe even 50 years and it's it we see a clear trend in downsizing of the technology so that it's yep yeah as long I have to look around because I don't know what you know it's already then I'm sorry for this so I see I see that my scale is going up to the nanometer cell so 1 1 nanometer is 10 to the minus 9 divided by 1 meter so that that's pretty small and we see here that the compute power which we hold in our hands which can be an iPad or or your smartphone or whatever has more compute power than let's say supercomputers 20 30 years ago so that's kind of the trend we see in technology happening still today yeah we all know about Moore's law that's gonna end I will not explain about that but Intel says no it's not going to end it will end when we don't have any ideas anymore and that's basically what we try to explain today so at the far right you see Adams and that's indeed the scale at which we're using technology these days this is the scale at which we are developing things and even in quantum computing I will tell a little bit more later on yeah we're using single electrons for instance to represent one which we call a qubit but I will explain that in in a couple of minutes so where is this what kind of philosophy or yeah scientific discipline are we are we basing ourselves on is there evidently physics and I see here I show here three prominent physicists that you may well for sure you you've recognized Einstein because he already mentioned once or twice I think in previous talks yeah and we see the guy on the right which is Niels Bohr he says he's a Danish physicist and he lived more or less in the same time period as Einstein and but the scale at which they were reasoning the phenomena that they were looking at is completely different so I'm Stein you had the relativity theory and those kinds of things is about gravity and interaction between planets or large large scale phenomenon and Niels Bohr was more let's say at the sub atomic scale and he said but those those concepts do are not applicable yeah at that level and because quantum and that's quantum physics there is a fundamental principle which is randomness and I will I will explain a little bit later on and that's also why Einstein says now God does not play dice because he could not he could not imagine that if there is something like a god yeah that he would he would assume that there is some random behavior in in nature yeah and therefore also in in human in human beings and the third person at the bottom is Richard Feynman and he's he's an American he was one of the the most brilliant physicist that died I think that lived at least in the 20th century and there were quite a number yeah off of them and he is actually the initiator he was the one who formulated in 1982 so that's basically around six years before he died he said we should build a quantum computer which is based on this quantum phenomena that we want to understand in a very deep way and therefore he he was proposing that we should look at the technology and think about how to build a a device a computational device and I will I will explain those those concepts a bit more if I can switch to the next light which I can yeah so what is what makes quantum computer so special and I and I hope to give you at least an intuition a profound understanding that's maybe very difficult because I still have to find actually the first person that really understands at that level yeah how it works so we are more intuitive in into in intuitive reasoning rather than very analytical so what makes these these quantum computers so so special well again let's let's go back to how machines that we have in our daily life that are filled in your car yeah that you have on your on your on your wrist most likely or in your pocket your smartphone or whatever it is they are all based on bits zeros and ones and then we can combine these bits in multiple sequences so that we can represent bigger numbers yeah or we can represent like huge text on which we can then apply all kinds of algorithms and algorithms are like the standard protocols or their little programs yet that describe what should be done at each step in the in in the algorithm so that's what what classical bits are now when we move into the into a completely different world namely the quantum physical world we have to look at completely different phenomena and I'm just mentioning here the three that are really key and really fundamental and extremely difficult to make on a large scale and I will maybe not explain everything but this is this is really a big challenge so it's super position measurement and entanglement and I will explain those three in the next coming minutes so here you have a little animation from it a kinetic where I have let's say an electron that is now in the ground state and now it's in an excited state so you see the energy levels they move discretely between low and high and now it's in the super positions because you see the two levels are activated and indeed this electron in this is in these two states at the same time now I have an observation and there's so when the eye opens one of those two superposition states actually disappears and that's actually where the randomness comes from yeah and this is really key because that means that when you build a quantum computer you basically move away from what we what we know classically which is very deterministic very classical when you do the same algorithm and the same data you execute it a thousand times you will get thousand times exactly the same results in quantum this is not true and I will explain that in in a couple of minutes so we use qubits as I already said so we don't use these classical bits because they don't have this superposition and entanglement property that we need in order to to compute so we have these qubits which can be in a combined state so it can be 0 and 1 at the same time so classically it's either a 0 or a 1 or any combination of those but it's always in an exhaustive yeah state of a of one single combination of zeros and ones in quantum physics you know this is absolutely not true so what I see here is on the left the 0 part which is let's say the right arrow the the red arrow is pointing above and the 1 state is the the red arrow pointing pointing below and then you can have any position in between which is the superposition state and then depending on what I'm going to how my algorithm is computing what it actually is doing it is manipulating what I'm showing you know is from behind it that the projection shows I'm sorry so what it what we do in in a quantum algorithm is that we influence the alphas that we see there that alpha 0 in the alpha 1 and in principle what it means is that alpha 0 in a way represents the probability when I do the measurement then the superposition disappears then at what with what probability am I going to get a 0 and with what probability do I get a 1 that's what those alphas actually mean and so all of the quantum algorithms and so now I will show you that it can be the 1 so that the Alpha 1 is bigger than the alpha 0 yeah then you will when you measure it you will measure the 1 results now assume that I have here 80% white and 20% black so that means I have to execute my algorithm multiple times so that then I can aggregate all of the results and then I can simply count what result gives me most is computed in most of the cases and that's going to be the final result of your quantum algorithm and that's why we call it it's a non deterministic computing reverse due to classical computing so I will briefly say something then about the the two other phenomena which is measurement as I already explained it basically means that when I when I'm in a superposition state yeah and I will measure it I'm gonna force it into either the 0 or in the 1 state always we have not yet figured out a way to overcome that and maybe in 20 30 40 50 years maybe we will but for now we don't yeah so that is that is that is where this non-deterministic element comes from and then the third property is entanglement and that's also a very crucial way of combining these qubits because the qubit that is in the superposition state basically has two states as two values namely the 0 and the 1 and if it's 50/50 then then it can you can you can get any kind of result at any point in time by measuring but you can never predict when it's 8020 then you know that in 80% of the cases you will get the 0 and the 0 case it for instance this case so the entanglement is again I'm combining two qubits they're in a superposition here yeah so I basically have when I combine two qubits I have four positions and if I have 3 it's 2 to the power 3 I cannot show that with my with my fingers yeah but you have eight different states that's what 2 to the power 3 so 2 by 2 by 2 right means so this and then the entanglement is a way of combining them and then when you when you apply whatever quantum algorithm on such an entangled state then you apply it at all of those combinations at the same time yeah it's not only about the entanglement there but this is this is really key so I will briefly show you yet to give you a little bit of an idea where the compute power now possibly comes from because yeah sure now that depends on how many qubits you will have and whatever it is yeah but if I give me five five or ten supercomputers I can do the same well this is not true and I'll give you an example later on so here I plot the graph of having two three up to ten cubits combined and then you see here these are 10 yeah 2 to the power 10 combinations so parallel States and what quantum physics is giving me is giving me massive parallelism for free and the for free should be taking a little bit of yeah salt because there's no free lunch here either but that's that's the principle idea that we're that we're using and why is that important because if I didn't execute any quantum gate and I will show one simple example on those on this matrix I am I apply all that particular gate on all of those states at the same time in one shot whereas in a classical machine I would have to do it sequentially yeah and two to the power 100 that's already a big number so it's sequential versus massively parallel that's what it that's what it means that also implies that if I'm actually building a new machine and I want to build said okay I'm gonna make it more powerful because I need to have more data that I need to process what I need to do is simply add one more qubit because b by adding one more qubit it's 2 to the 2 to the 3 to the 4 to the n n plus 1 I each time double the entire number of superposition states that I can manipulate at any point in time so that's what the the theory of course tells us right in the and we have to work on building a machine which is capable if we're if I have let's say 300 which was the point on the far right of this graph if I have 300 of those let's say perfect qubits I can represent more data than there are atoms in the known universe just just think about how big such such a huge amount of data would be so that's why I showed this this little yeah universe picture now let's let's start and talk a little bit about how to build a a quantum machine so just like classically we have to build transistors yeah and the transistors we combine yeah in particular ways well we have to do the same thing physically we have to work on making those qubits and we have to make sure that these qubits are stable and live long enough in time yeah because we're here yeah there's this one Austrian and Italian working in Australia and he was capable of making one qubit yeah in silicon 20 in 28 silicon whatever that is and it could live for five seconds it was incredible and how can you make to know maybe three no okay so he doesn't know it has no clue how to scale this up and most likely that's the direction that we are basically not looking at so in Delft yeah because I'm I'm Flemish but I work already for quite some time in in the Netherlands yeah we were basically on quantum dots NV centers semiconductors and superconductor kind of technology we do not work on on ion traps which was very popular and is still very popular especially in the US but we don't we don't have anybody in Delft doing that and I personally am more on the quantum dots and and the superconducting qubits now how do you start doing such a work because I'm a classical engineer yeah I'm a computer engineer so I'm used to building machines and and then we had we had the the the request from the rector yeah of our university said yeah but we start anew if we start a new research center called qt you have quantum technology whatever is in the name yeah and I want I want you from that faculty from the electrical engineering faculty to be responsible for yeah making the architecture actually ok quantum physics yeah do I know anything yeah the basic principles like I just explained and then then I started looking at ok I have to start somewhere and the first drawing I made was actually this one was not that that sophisticated as I was it here right it was just with pen and on paper and this is a system stack of a classical machine and I said well if I simply write the Q in front of each layer maybe good enough yeah well turns out so far this is the case yeah this is the case yeah well we can be proven wrong yeah and then I will come here and say I was completely wrong excuse me for that yeah but for now this really seems to be yeah the keypad and turns out yeah just it's a little bit of off of no I don't know so turns out that Intel was at some point every four five years this American that you most likely know yeah revises its roadmap technology roadmap and quantum was always out there's an added sewer that is too early yeah we're absolutely not interested yeah it hasn't it hasn't matured in any particular way yeah until until the last time and this was like three years ago there was a guy Jim Clark yeah I can mention his name now right he was traveling all over the world literally all over the world visiting every single research lab working on quantum technology quantum computing and that means you know the Ivy League universities in India Wes right from MIT Princeton Stanford those kind of things to Australia to Canada all over Europe and he also came at some point to TU delft and they asked me together with my colleague around the shop oh hey can't you guys say something about quantum computing now happens to be this is a little bit of interior information three weeks before we were completely killed because of this idea by our my own rector by by a physicist say like what I helped are you doing go back to your office and think again and come back so we say that what sorry we're gonna sorry gonna bring to Intel and I said you know what we don't know anything else we'll just give this one yeah and remember what happens they said no I was gonna say dirty word survivor I have to filter myself sometimes right said like we are we have never heard this story I've been traveling to every single research lab in the u.s. in Canada in Australia in Europe and this is the only place where I get a systematic and a coherent view on how to build a quantum computer language so and that was a start that was a start now I'll show you yeah vanity of course has its price so I'll show a little picture yeah where we started actually working with with the Intel people so again the system stack we have and then you have to again start at the lowest bottom so the dark blue the point that you saw was basically where all the physicists anywhere in the world are working yeah okay time is out okay so this is a this is a this is a quantum circuit so on the right you see yes you're right yeah you see a quantum circuit consisting of a couple of operations that classically you do not really use and that's fine we know how to use them and that's I think for now the most important part then we came up this was also a drawing that I showed this Intel person I said yeah I think we need all of these things and again the little box the pink box on the on the right is where the physicists are so that's 99% of the people that are working on it or in just this this pink box box at the at the right and we do everything else yeah and and well actually that's that's a team in in Delft that that is working on that and so you have to define also you need to you need to have programming languages you need to have algorithms of course so you need a language you need a compiler you need you need to build all these things you need to have a microarchitecture and again as you can see we're working on three different kinds that I'm currently working on all three different kinds of physical technologies and so we have prototypes at least this year for the for the spin qubits and the trans months yeah we're not so good in making publicity so maybe being here today is a good is a good exercise yeah and this is my team that works on all of these aspects that I've been that I've been telling you the people at the bottom are our PhD students at the top are our postdocs or assistant professors and and and the old guy on the left is that that's me yeah and then who else is working on as I said already it's it's huge it's a pretty big field these days and so yeah on the right you see the American companies we know Microsoft was on if you if you read the red except in ba I don't know whether you do that yeah I do that from time to time and when they were thirsty much like Microsoft is busy announcing the big quantum computer uh-huh yes it like we're running behind it's absolutely not true right so the most famous one is maybe d-wave which is a Canadian company and they claim to have already more than 2000 cubits is it in a powerful no my laptop run smarter than there than their quantum machine oh and here you do see the picture of Intel when we started yeah that's I'll just now what are we yeah yeah I'm running a bit late but I think that's fine right because the people listening to me said yeah but you have to talk much more about what are you gonna do with it so this is the part that I will insist on a little bit right so what why do you need quantum computers yeah I will no of course the compute power now allow yeah we believe you yeah sure that's fantastic but how are we gonna use that yeah my my the language that you develop wow great yeah but what kind of problems and then you see an entire domain yeah and the big data is a big is a big field right I mean big data they slap us around the years with big data every day yeah and and and we don't have solutions for that well wait until we have a quantum computer of course yeah now what I personally believe and I will not go into too much detail but I personally believe and we start working it we started working on this already now is working on what is in the top left you know what is in the in chemical in chemical applications and mostly yeah what we what we start now is on genome sequencing genome sequencing what I feel you're talking about physics and now it's only biology again right but genome sequences is the future of medicine so whatever is going to happen let's say in ten years when you go to your physical doctor yeah it's gonna take a bit of blood or whatever and he will send you through a such a such a sequencing machine and then it needs to be analyzed now this is billions billions of bytes that you need to process so it takes for one person yeah it takes a very very powerful computer network for more than a week for one person to do one analysis so that's that's one one thing that we think is is an important application domain so just to give you another example in then I will really wrap it up yeah is is is this one is about factoring a large number and that this is something that we all use on a daily basis whenever you contact your bank yeah it's basically in an encrypted way so that means that the day that you don't send zeros and ones you send them in an encrypted way and the encryption basically boils down to decode finding the the prime numbers yeah when you multiply all those prime numbers it gives you your big number and that's kind of the key yeah to do it so again if I'm assuming that I have the most powerful supercomputer which is Chinese by the way right at the chancre here and I can build a data center the size of Germany and the Netherlands so that's around four hundred thousand square kilometers yeah and just assume that every single meter is filled with this the supercomputer it's still gonna take me let's say about a hundred years a trillion euros and basically it can never work but it's going to consume so much power that the earth is running out of energy in one month so that's not something you can actually see as a realistic approach now if I can build a quantum computer yeah and look at the five billion cubits you know the physicists they're very proud they're at 7 and 17 yeah so going to five billion is still quite a way to go but in principle at that size with the factoring the Shor's factoring algorithm it should be done in a bit more than a day so just to give you an example from 100 years yeah which is in consumer with in consumable amounts of energy you can downsize it to two to one day and it really uses much much less energy I did not insist on that yeah because I am already running out of time I will be very short here in the conclusion yeah that of course there are a number of huge challenges that we still need to overcome and and scaling it up is actually the biggest one yeah I can I can for maybe I can make seven cubits nine cubits 17 yeah and they're always like odd numbers so don't ask me about those odd numbers but yes they are all odd numbers yeah but yeah the the potential benefits of building such a machine are so huge that I personally think in many people you know that the investment is really worth worth the effort thank you very much [Applause]

10 Comments

  1. Василий Немцов said:

    Я ничего не понимаю, но я так люблю эти видео.

    June 27, 2019
    Reply
  2. Kyle Langston said:

    You can never be in two places at once yet these very small particles can yet quatum physics cant explain it which means the alpha 1 and 0 do not mean anything because there is nothing you can do with them to be effective enough to help any situation because you can never contain what you cant control and I see no control who knows what's really going on

    June 27, 2019
    Reply
  3. bert havermout said:

    What a nerd… Sorry man, who needs these quantum computers? Human Genome??? What an answer!!! Develop AI? Get us enslaved by AI? Think before you get excited about something that nobody is waiting for. Only big companies…

    June 27, 2019
    Reply
  4. ututura said:

    the hurry in the air of all ted talks is almost as pointless as the talks themselves

    June 27, 2019
    Reply
  5. Naimul Haq said:

    Bell Laboratory devised the transistor with three terminals, revolutionizing the computer employing semi-conductors circuit, producing the on-off switching systems to give the two states 0 and 1.
    For the intermediate states of super position of the other intermediate Qubits, what is perhaps needed is a similar device with more than three terminals, providing the intermediate states.Which may make it possible to make QC at room temperature with no cooling necessary.

    Nature employs QC and we can study these systems, as in our five senses, the brain and in all our cells, how nature solves the problem of simulating the intermediate states. I have not seen any attempt in this direction, although molecular biologists have figured out intricate process how cells produce protein and transfers them employing small motors. However, sending small signal inputs and studying the outputs may lead to a breakthrough.

    June 27, 2019
    Reply
  6. Elek said:

    You must remember that you are talking to everybody , not only to a native speakers .

    June 27, 2019
    Reply
  7. bcmasur said:

    all these ted talks about quantum computing but nothing about how it works, still haven't had a qgasm

    June 27, 2019
    Reply
  8. DocStomp said:

    what is the date of the talk?

    June 27, 2019
    Reply
  9. Michael Fimian said:

    Great presentation, Koen… Don't let the timekeepers bug you!

    June 27, 2019
    Reply
  10. NumChuck Lee said:

    most likely not a good idea to have computer engineers define quantum machinery because they do not live the math !

    June 27, 2019
    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *