Here is the full transcript of UNSW quantum physicist Michelle Simmons’ TEDx Talk presentation on Quantum Computation at TEDxSydney Conference. Professor Simmons is the Director of the Australian Research Council Centre of Excellence for Quantum Computation and Communication Technology, a Laureate Fellow and a Scientia Professor of Physics at the University of New South Wales in Sydney. Read the full bio here.
Right click to download the MP3 audio:
Michelle Simmons – UNSW quantum physicist
Every year computers get smaller and smaller, and faster and faster. Have you ever wondered when is it ever going to end?
Well, one person that’s been looking at the miniaturization of computers over the last several decades has been Gordon Moore. And he’s the co-founder of Intel back in the 1960’s. And he noticed that the number of components on a silicon chip doubled roughly every 18 months to two years.
Now for this to happen, it means that the smallest feature size on a silicon chip has to decrease at the same rate. And he came up with something called Moore’s Law and here it is represented on the screen.
Now, this law has been going now for approximately 4 to 5 decades. And what started out as an observation by Gordon Moore has now become a law after his name, Moore’s Law. This actually continued in time.
The interesting thing is that the industry has now set this as their roadmap of how to make computers smaller and smaller, and faster and faster. So you have multi-trillion dollar industries, the semiconductor industries, pouring money in every year to try and beat that law. Until now it’s become a self-fulfilling prophecy.
See, if we have a look at where we are at the moment. Here is a cross-sectional Scanning Electron Microscope image of a single transistor. Now, the smallest feature size in this transistor is the distance here between the source and the drain. It’s about 30 nanometers. It’s 5,000 times smaller than the width of a human hair.
What’s amazing about that is if you look around you now, we all carry around our personal electronics. And within one silicon chip you have over 3 billion of these transistors. And they all have to work reliably so that your computer, your mobile phone, whatever you’ve got with you, actually works. That’s quite amazing. Just think about that now. Everybody in this audience has got billions of transistors. There are trillions of transistors in this room.
But one of the nice things about Moore’s law is, you can actually predict with time what’s going to happen. And eventually you’ll see out here, in roughly 2020, less than 10 years away from where we are now, the size of a transistor will get down to the size where it’s a single atom. That’s the smallest component of nature. It’s very difficult to imagine that you could make a transistor any smaller than that.
But this is the world of digital information. So let’s just understand how that transistor works. Here, we have a silicon substrate. That’s what the transistors are made of. And above that we have an insulating oxide and then a metal gate. What we do is we apply a positive voltage to this top gate here, and that sucks up — attracts all the electrons that are in the silicon up towards this gate. But they can’t get there due to this insulating oxide. So they form this two dimensional sheet which forms a conducting channel between source and drain, and that turns the transistor on. That is our “1” of digital information.
If we now put a negative voltage on this gate, we repel the electrons down here and we push them away from that channel. So there is no conducting sheet and, as a consequence, you get the “0” of digital information. So that’s the ones and zeros as we go down. For everything that works around us now, everything is coded in either a “1” or a “0”.
And what happens as we go smaller and smaller in size, is we actually cross over from what we call the “Classical Age” to the “Quantum Age”. And there, things really start to change.
In the classical world, we understand how things work. So if I had a tennis ball now and I was to throw it at a wall, it would hit the wall and it would bounce back and I’d understand and I’d see it and be able to write equations of motions to describe that. But as I miniaturize things down and imagine that tennis ball being the electron in my transistor, if I made it very, very small and I threw that electron at the wall, instead of it bouncing back, it actually behaves more like a wave than a particle, and it can tunnel through the wall and it can come out the other side.
Now, that’s something that’s quite scary. As we make our devices smaller and smaller, the wonderful world of quantum mechanics comes in. Electrons behave like waves and they no longer go in the computer where we want them to go. So a lot of people have predicted that this would herald the end of Moore’s law. But in reality, it’s the start of something new. We’re now transitioning to quantum mechanics.
And if we control quantum physics, we could actually build computers in the quantum regime that are predicted to have exponential speedup over classical computers. So one of the questions a lot of people ask me is, aren’t computers fast enough already? Can’t they do all the things that we need them to do? Obviously, everyone wants things to be faster all the time. But there are some problems out there that just cannot be solved efficiently using a classical computer. And one of those is something called the traveling salesman problem.
So here we have a salesman, we want him to travel to lots of different cities, and we want to work out where the shortest possible route is. That sounds like an easy problem. But it’s actually one of those intractable exponentially hard problems. So, here we have on the screen the number of possible routes he can take as a function of the number of cities. It’s something that grows very very quickly. So by the time you have 14 different cities, there are now already 10 to the power of 11 possible roots that he can take.
So if I take a classical computer, it works in the gigahertz regimes, 10 to the 9 operations per second. And it can work out the shortest possible route in about 100 seconds. Well that’s no big deal.
But now what happens when I go to 22 cities? There are now 10 to the 19 possible routes that salesman can take. And with that same classical computer it would take 1,600 years. This is amazing. And if you look, by 28 cities, it’s longer than the lifetime of the universe to work out what the shortest possible route is. I heard this problem many years ago and I just couldn’t quite believe it. This is a real problem, it exists out there.
So how can we make a computer that can somehow solve those kind of problems? We have to look at how a classical computer works. A classical computer is very fast. But it searches through all the possibilities one after the other. Rather like a recipe. So if I was to write down a telephone number on a piece of paper and I’d forgotten whose telephone number it is, I’d get my classical computer to start looking through all the A’s, then all the B’s, then all the C’s. And eventually it would find whose number it is and tell me. If I wanted to go faster, I could put 2 computers onto the problem. Get one searching between A to L, the other between M to Z, and it would go faster. To go faster, I’d have three computers. Well, that’s the digital world.