Here, inside this refrigerator, at a temperature
just a tick above absolute zero, isolated from the rest of the universe...
...is a quantum computer. If you believe the hype, this nascent technology
embodies the promise of the future, and has the potential to revolutionize our lives with
its turbo-charged computation. But quantum computers aren’t the next generation
of supercomputers—they’re something else entirely. And before we can even begin to talk about
their potential applications, we need to understand the fundamental physics that drives the theory
of quantum computing. We’ll need to dive into another dimension,
smaller and more alien than anything we intuitively understand: the subatomic world of quantum
mechanics. Feynman’s Idea
In the 1980s, one of the most important physicists of the 20th century encountered a major roadblock. Richard Feynman was hungry for a window into
the quantum universe. But quantum systems, by nature, are fragile,
and the information they hold hides from us. Because Feynman couldn’t directly observe
quantum events, he wanted to design a simulation. It quickly became clear that his computer
wasn’t up to the task. As he added particles to the quantum systems
he was modeling, the cost of computation began to rise exponentially. Feynman concluded that classical
computers just can't scale up fast enough to keep pace with the growing complexity of
quantum calculations. Then he had a breakthrough. What if he could design a tool made up of
quantum elements itself? This instrument would operate according to
the laws of quantum physics, making it the perfect way to probe the mysteries of the
quantum realm. The idea of the quantum computer was born. And by dreaming it up, Feynman had started
to build a bridge between quantum physics and computer science. To understand how quantum computing works,
it’s essential to start by understanding what makes it quantum in the first place. This means that we need to talk about what’s
at the heart of quantum physics: a concept called amplitudes. Here’s what the classical rules of probability
tell us about getting tails if we toss a coin 20 times. We add up the probabilities for all the possible
outcomes resulting in tails. That’s just common sense. But common sense doesn’t govern the quantum
universe. Before you measure a subatomic particle, you
can think about it as a wave of probability that exists in a kind of black box—a quantum
system with many different chances of being in many different places. Quantum mechanics, at its core, is a change
to the rules of probability. (02:05) This is also where the power of quantum
computing comes from—from these different rules of probability than the ones that we
are used to.Amplitudes are closely related to probabilities. But they're not probabilities. A key difference is probability is always
a number from zero to one. But amplitudes are complex numbers. And what this means is that they obey different
rules. So, if I want to know the total amplitude
for something to happen, I have to add up the amplitudes for all the different ways
that it could have happened. But when I add up amplitudes, I see something
new, which is that a particle might reach a certain place one way with a positive amplitude
and another way with a negative amplitude. And if that happens, then those two amplitudes
can cancel each other out so that the total amplitude would be zero, and that would mean
that that thing would never happen at all. So the amplitudes are connected to the probability
that you actually see something when you look there. This is sort of the central thing that quantum
mechanics says about the world: that the way that you describe a physical system is by
a list of amplitudes. And the way that a physical system changes
over time is by a linear transformation of these amplitudes—by some change to these
amplitudes. But how can quantum computers use amplitudes
to store and manipulate information quantumly? This is a qubit. It’s the basic computational unit in quantum
computing. Qubits are like bits in a classical computer,
but with a crucial difference. A bit is binary—it stores information in
strings of binary digits that can only be 0 or 1. But qubits are made of subatomic particles,
so they operate according to subatomic logic. Qubits can be 0, 1, or what we call a linear
combination of 0 and 1. This fluid combination of amplitudes is at
the core of quantum computing. Before you measure a qubit, it exists in a
state called superposition. You can think about it as a quantum version
of a probability distribution, where each qubit has some amplitude for being 0, and
some amplitude for being 1. Superposition is the reason that quantum computers
can store and manipulate vast amounts of data compared to classical computers. When two or more qubits are in this closed
state of superposition, they relate to one another through the phenomenon of entanglement. This means that their final outcomes, when
we measure them, are mathematically related. Quantum entanglement is the word we use for
the characteristic correlations among parts of a quantum system, which are different from
the correlations that we normally encounter in the classical world, in ordinary experience. You could think of it as like a book. When you look at the pages one at a time,
you don't see any information—you just see random gibberish because the information isn't
encoded in the individual pages, but in the correlations among them. And to read the book, you have to collectively
observe many pages at once. But if you want to describe very highly entangled
states using ordinary bits, it's extremely expensive. Imagine that you had a primitive 10-qubit
computer. It could store 2^10 values in parallel. To describe this entangled configuration with
a classical computer, you’d need 16 kilobytes, or 16 thousand bits. Expand to a system with 500 entangled qubits,
and you now require more classical bits than there are atoms in the known universe. This is exactly what Feynman meant when he
said that classical computers weren’t scalable for simulating quantum mechanics. For a quantum computer to be of any use, you
need to measure information from the qubits to get an output. The problem is, when a quantum system is measured,
it collapses into a classical state. If you look at a qubit, let's say to ask it
whether it's zero or one, then you collapse its state, right? You force it to decide whether to be a zero
or one. Anything carries away information about whether
that qubit is zero or one—so for example, if that information gets recorded in some
radiation that's escaping from the quantum computer, then the effect on the qubit will
be exactly as if someone had measured it to see whether it was 0 or 1. When you look at the system, then the amplitudes
become probabilities. To extract an answer from the quantum system
that isn’t just a random outcome of probability, like the flip of a coin, we have to use interference. Interference can be seen in classical physics
… when waves in a pool hit each other, and one wave is above the surface, and the other
wave is below the surface, and they cancel each other. Interference is just what amplitudes do when
you add them up. … If something can happen one way with an
amplitude of a half and another way with an amplitude of minus a half, then the total
amplitude for it to happen would be zero. This is what you do in the famous double slit
experiment. You close one of the paths, and then you see
that now the thing that previously never happened, can happen. This is a quantum algorithm. Scientists can harness interference by creating
a deterministic sequence of qubit gates. These qubit gates cause the amplitudes to
add up constructively. This means that they’re mathematically guaranteed
to boost the probability of seeing one of the right answers. This is a quantum algorithm. Scientists can harness interference by creating
a deterministic sequence of qubit gates. These qubit gates cause the amplitudes to
add up constructively. This means that they’re mathematically guaranteed
to boost the probability of seeing one of the right answers. You might ask, how could you possibly concentrate
all this on the right answer when you yourself don't know in advance which answer is the
right one? This is exactly why designing quantum algorithms
is so difficult and why we have a whole field that's been studying it for decades. Since 1994, there have been a few major breakthroughs
in quantum algorithms, with theoretical applications in fields such as cybersecurity and search
optimization. But according to most experts in the field,
quantum computers are most likely to be useful for what they were born to do—when a curious
physicist wondered about the deep structure of our world. I find quantum computing exciting as a way
to explore physics. Now, whether that's going to make anybody
any money—whether there'll be practical applications in the near-term—that's still
very much an open question. But at least for physicists, it's an exciting
time. The truth is... that the most important application,
I believe, of quantum computers is something that we don't know yet. I'm sure that once we have a quantum computer
to play with, we'll find amazing applications that we can't yet foresee.