blog banner

Quantum computing explained with a deck of cards | Dario Gil, IBM Research


We have a very exciting
last talk coming up. Dario Gil will take us
into a quantum world. Dario is the Vice President
of Science and Solutions at IBM research, where he
leads over 1,500 engineers that are researching in technologies
and physics, math, health care, life sciences and others. And while some of
you will think, a quantum world,
that’s too far out, I’m very sure Dario
will tell us otherwise. So come up here
on stage, please. Thank you. Thank you. I was joking with
Mark that we couldn’t pick an easier topic to end
the day, on quantum computing. But I’ll try to make it
entertaining, and hopefully easy to understand. I’m going to start
with a reference to this term of beautiful ideas. And it came from hosting
a filmmaker about a year and a half ago, in the
laboratory I just showed you. At the TGA Watson Research
Center in Yorktown Heights. And he was a filmmaker
that directed this documentary called
Particle Fever, that I don’t know if you’ve had
a chance to watch, but I highly recommend it. It’s about the team
that was pursuing the discovery of
the Higgs boson, in the largest physics
experiment ever conducted. And a major
character in the film is a professor from Stanford. And at the beginning
of the film, he said something that
really captivated me. He said, “The thing that
differentiates scientists is a purely artistic ability to
discern what is a good idea, what is a beautiful idea,
what is worth spending time on, and most importantly,
what is a problem that is sufficiently interesting,
yet sufficiently difficult, that it hasn’t yet been solved,
but the time for solving it has come now.” ” So I want to tell you about
this beautiful idea, whose time for solving it has come now. And that is the possibility
to create quantum computers. If you look at how
we have created the basis of the
information revolution, and you trace it back to
other beautiful ideas, like what Shannon
taught us, to think about the world of
information abstractly. If you look at an old
punch card and DNA, we’ve come to appreciate that
both carry something in common. They carry information. And Shannon told us
that this world of bits could be decoupled from its
physical implementation. That was really interesting. But in fundamental
ways, it went too far. Leaving too much physics out. So here is two scientists that
work at IBM Research, Charlie Bennett on the right, continues
to work in our laboratory, And is an IBM fellow. And they asked the
question, at the time, of is there a fundamental
limit to how efficient number crunching can
be, computing can be? And when they asked that
question as physicists, they ended up with a
very surprising answer. And they found the
answer to be no. It turns out, that
number crunching can be thermodynamically reversible. These led to an
exploration of, what is the relationship between
physics and information? And there was a
now-famous conference that was jointly organized
between IBM research and MIT at Endicott house,
where this topic was explored in more detail. And the plenary speaker was
none other than Richard Feynman. And Feynman proposed
in that conference, that if you wanted
to simulate nature, we should build a
quantum computer. And I’m gonna explain
you what that means, and how it’s created, and the
problems that it will solve. But first I’ve got to tell you,
what is a fundamental idea? The fundamental
idea, just like we have bits in the
classical world, that can be a zero or a one. In a quantum computer,
you have qubits, which stands for quantum bits. Now, the difference
is that there can be a zero, a one, or
both at the same time. That exploits a principle
of quantum physics called superposition. And it sounds weird and
crazy, but it’s true. Now to give you this unease that
you should feel when you talk about quantum information,
and quantum computing, I’m gonna give you a
very simple example. A thought experiment that
also happens to be true. So let’s imagine that we’re
going to solve this problem. The problem involves,
you have four cards, three are identical, one is
different, one is a queen. We shuffle the cards, and
we put them face down. And the problem we’re
going to solve together, is find the queen. We’re going to be
assisted by two computers. One is a classical computer,
one is a quantum computer. So what we do, is
we turn them down, and we load them into memory. So we use four memory slots. The cards are
identical, we put zeros. The one that has a
queen, we put a one. So in our four slots, we
will have three zeros, and one is a one. We load them on
the two computers. Now we has to write a program
to find the queen, find the one. How would it be
done classically? You would go and
pick a random number, you don’t know where it is. You go look under that memory
slot, see if it’s a one, if not, you go to the next
slot, and so on, and so on. On average, it would take you
the equivalent of 2 and 1/2 turns to find it. It turns out, that with
a two-qubit quantum computer for this
problem, you can always solve it in one shot. So that uneasy feeling
that you have now, should be an explanation that
quantum computer is not just about building a
faster computer. It is building something
that is fundamentally different than a
classical computer. Now, a way to think about
it, an abstraction of it, is that a quantum
computer is always going to have a classical
computer next to it. They have to go together. So you have a classical
set of bits, right? The problem that you’re
trying to explore. And what that quantum computer’s
gonna allow you to do, is to explore these
exponential number of states. These 2 to the n, where n is a
number of qubits that you have. So now, we have relatively
small quantum computers, with few qubits. But just think of the
number, that by the time you have 50 qubits, you
have 2 to the 50 states. That’s a phenomenally
large number. But in the end, after you
explore these number of states, you go back to a
classical output. A string of zeros and
ones, that you interpret with a normal computer. So why is this interesting? And I think in this
audience, I don’t need to explain in
great detail, you know, what exponentials mean,
and why 2 to the 50 is a very large number. But it’s still, I think
it’s an interesting way to communicate
the power of this, and I like to map
it to some problems. But I like to go after
this apocryphal story that actually, IBM
used in the 1960s to explain to people the
power of exponentials. And it had to do
with the person who invented chess, that goes
to the emperor, and says, well here’s his wonderful game. And asks, what do
you want in return? And the person who
invented it says, give me a grain of
rice on the first day, for the first square,
and the second day you give me twice as much. And on the third square, third
day, you give me twice as much as the day before. And the emperor agrees
promptly that that seems quite reasonable. And after a week you
only have 127 grains. After a month,
you have more rice then you’ll eat in your
lifetime, for sure. But just by the time you get
to the end of the chessboard, you have more rice
than Mount Everest. So there are a large
number of problems in the world that have this
characteristic, that they blow up exponentially. And a dirty secret in
the world of computing is that we obviously talk
a lot about all the things that computers can solve, and
can solve a lot of things. But then, there’s
a lot of things that computers can not solve. And very interestingly, they
cannot solve it now, nor ever. And the reason is because they
have this exponential built into them. So take as an example, this
fairly simple equation. Factoring. So if I have a number,
M, that is made out of the multiplication of
two large prime numbers. And I only give you M, and
I ask you find me p and q. It turns out, that that
is phenomenally difficult to solve. There’s no other way but to
divide it sort of sequentially, by prime numbers. So in fact, it’s
so difficult, we use it as the basis
of all encryption. But, if you had a very large
universal fault-tolerant quantum computer, which
is many, many years away, you could solve that
problem in seconds, what would take billions of
years in a classical computer. That tells you something
about the power of what is going to be possible. Take chemistry, as a problem. Because it also has
this characteristic, that it blows up exponentially,
if you try to calculate it. This equation that you see
here is very interesting, because it’s predicted
to occur at the ocean floor near volcanic
sites, and famously has been hypothesized to be the
basis of the formation of life on Earth. But if you take a molecule
like iron sulfide, and you try to do relatively
simple calculations with a normal
machine, it turns out, that we’re not very accurate. And the reason is
that molecules form when electron orbitals
overlap, and the calculation of each orbital requires a
quantum mechanical calculation. So for that simple
molecule, you have on the order of 76 orbitals,
and two to the power of 76, is intractable with a classical
computer, so we can not solve it. Again, on this theme of our
assumptions that computers solve everything,
but they don’t. If you look at calculating
for example, the bond length of a simple molecule
like calcium monoflouride, we still get it off
by a factor of two, even using the largest
supercomputers in the world. To me, this has been
very interesting, this recognition of all these
problems we cannot solve. It’s also true in
optimization problems, that are the basis of
logistics and routing, and you know,
portfolio optimisation. There’s tons and tons of
problems in which at best we do approximations, but
we’re far from optimal, because a number of
possibilities is enormous. So if there’s one message I
want to be able to come across, it’s that we have these
easy problems, which is the world where
classical computers fit, and the problem it’s solved. But then there these other
hard problems, that go outside. And if you don’t
believe that p equals np, which I would say the
majority of mathematicians don’t believe that that is the
case, that those problems are hard for a reason, the only
avenue to go and tackle that, aside from approximations,
will be to the creation of quantum computers. So where are we? We believe that small
practical quantum computers are going
to be possible, and we’re building them now. It requires reinventing
the whole stack. The device is different. It’s not the
traditional transistors. As an example,
this is the device we use for that
quantum computers that we create at IBM, based
on superconducting Josephson junctions. And you’re seeing an example
of one of these device, is superconducting device. And because it’s
superconducting, you have to cool it. So this is what a small
quantum computer looks like. What you’re seeing
here is something called a dilution refrigerator. And this quantum processor
sits at the bottom of this refrigerator,
at the nice temperature of 15 millikelvin. So that is colder
than outer space, where we have to put this
quantum processor in. This is what, for example,
a 16-qubit quantum processor looks like. And you know, inside,
you see the square where the qubits are, and you
see these squiggly lines, which is these coupling
resonators that allow you to send information
uncoupled to the qubits, To send the information. This is what the
wiring looks like, into the refrigerator going
into a quantum processor. There’s these coaxial
cables, because the way you send information
to a quantum processor, is through a series of
microwave pulses, that go in, and then you’re
able to take it out. Now, if you look at pictures
of what computers were like, right, in the ’40s
and the ’50s, it’s kind of like where
we are today, right? That’s what, you know,
quantum computer, that’s the signal processing
required to actually send all those signals down
the coaxial cables, it looks like that. But we’ve also seen
this movie before, in the sense that we know
how much progress we have made from those early system. And while we don’t anticipate
that quantum computers will be on your phone, because they
require cryogenic cooling, we definitely
believe that access to quantum computers
in the cloud will be something that people
will be able to leverage, behind the scenes,
even not knowing. Because we believe that,
we created a small quantum computer last year, and we
made it available to the world. In something called the
IBM Quantum Experience. And all of you can go and log
in and have access to this. It’s available for free. It’s a 5-qubit machine. And since we launched it,
we have over 36,000 users from over 100 countries
that have been doing it. And 15 scientific
publications have gone on it, and people are learning how
to program, and to learn about this new world, and
what is being created. And you can actually
run things on this. So I was telling you about
these chemistry problems. So this is an example of
the expected theoretical calculation, and the actual
calculation, on a small quantum machine, of hydrogen.
So we’re starting to solve small problems. And what is coming in the years
ahead, in the next few years, will be machines that
no classical computer will be able to emulate. Because by the time you
have order of 50 qubits, think about that, that’s
2 to the 50 states. And no classical machine
will be able to emulate what that can do. And that is new territory. And that’s the territory
we’re all going to enter. And now is the most
interesting part, because it’ll be the path of
discovery of what we can do, and what value we can
create, on problems we couldn’t solve before. So I’ll close with
Feynman, who proposed this original idea of creating
these quantum machines. In his inimitable
style, he said, “Nature isn’t classical,
dammit, and if you want to make a
simulation of nature, you better make it
quantum mechanical, and by golly, it is a
wonderful problem, because it doesn’t look so easy.” Thank you.

Leave a Reply

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