Quantum Computers and Quantum Supremacy
hello
i’m john martinez and would like to talk
about our recent experiment on quantum
supremacy
computers are important for the modern
world in part because they continue to
become more and more powerful over time
as described by moore’s law however
physical
limits are slowing down such
advancements so that new computing ideas
are being researched
quantum computing is such a new
technology which promises to solve
certain problems
exponentially faster than current
computers
billions of dollars and decades of
research has been invested in quantum
computing
the aim of the quantum supremacy
experiment as google
was to explicitly demonstrate this speed
up
on at least one algorithm in particular
the quantum computer took data for 200
seconds
which then needed over a thousand years
to check on a supercomputer
thus showing the computational power of
quantum mechanics
now quantum panics was first developed
to describe how atoms work
for example take the hydrogen nucleus
which is a negatively charged electron
a electron circling a positive charge
proton nucleus
both particles attract each other and
classically you would expect them to
just stick together
this is not what happens since quantum
mechanics explains
the electron forms a small cloud around
the proton
whereas simultaneously at all parts
around the nucleus at the same time
physicists call this the superposition
principle
quantum explains why atoms have size an
important property of all the matter
around us the cloud is formed by a
precise physical theory
that can predict well the properties of
atoms
quantum computing abstracts the idea of
superposition to information
which is fundamentally described by bits
that are either
0 or 1. in a quantum computer each
quantum bit or qubit
can be both 0 and 1 at the same time in
a quantum superposition
if a qubit can be at 0 and one in the
same time
then is it possible to compute the
answer to zero
and the answer to one simultaneously
just like a parallel computer
this gives a factor of two speed up
which is interesting
but not necessarily useful it becomes
useful however
considering what happens to multiple
qubits
for example in two qubits it
simultaneously computes
the answer for 0 0 0 1 1
0 1 1 now a factor of 4 a parallelism
this increase in parallel computing is
doubled for each added qubit
and exponential increase versus size
thus at 50 cubits the parallelism is 2
to the 50th power
which then starts to compete with the
supercomputer
at 300 cubits 2 to the 300
is basically the number of atoms in the
universe
which is clearly a size that is gets to
be interesting
it’d never do classically we make our
quantum computer using artificial atoms
made out of superconducting electrical
circuits
here the macroscopic currents and
voltages in the circuit behave quantum
mechanically
it’s not obvious that nature allows
quantum mechanics to work this way
so that was my phd thesis in the
mid-1980s to prove it
the interesting idea here is that you
can build a complex quantum system
a quantum computer using integrated
circuit fabrication technology
just like that’s used for building a
computer chips
it has taken many people around the
world for about 30 years
to figure out exactly how to do this and
to get this
the circuits to work well for the
quantum supremacy experiment
we built at google a chip called
sycamore that had 53 qubits
where each qubit connected to each
neighbor in a square grid
the big step here was to precisely
control and calibrate
each qubit so we knew exactly what it
was doing
the quantum supremacy algorithm was
essentially building a quantum
random number generator which had a
quantum property similar to
laser speckle such that some num numbers
are more likely than other numbers
so we measured these random numbers from
a quantum computer
and then checked with a quantum
simulation on a supercomputer
that the numbers we measured were
actually the ones predicted
for 30 qubits one can check with your
laptop
at 40 you need a powerful workstation at
50
you need a super computer or a google
data center
at the 50 feet we eventually measure it
would take thousands of years to do the
check
now important science that was tested
here is that the quantum computer
worked just as predicted we found no new
physics that would prevent us from
making
even more powerful machines that’s
really good news
so now that we’ve shown a quantum
computer can be powerful the next step
is to show it can do
something useful for quantum algorithms
a big area of research is quantum
artificial intelligence
and quantum ominous optimization like
solving the traveling salesman problem
although this promises to have huge
commercial impact
useful algorithms have not yet been
invented and clearly in active area
research
i like the research into quantum
chemistry which is actually what richard
feynman first proposed in the 1980s
and now algorithms exist to solve such
problems
simplified versions are being run on
quantum computers now or giving sensible
results
it’s a good test of the theory what’s
going on
these algorithms can be thought of as
running let’s say a hundred to a
thousand
lines of code for for practical
applications we’re going to need larger
machines
maybe in the millions or billions of
lines of code
to do this we need to reduce qubit
errors by millions or billions
which requires quantum error correction
this is currently a frontier research
into making a quantum computer may take
maybe a decade to figure that all out
the field of quantum computing is
growing with billions of dollars of
research
funding now being planned although
building a quantum computer is
challenging
the quantum supremacy experiment has
shown that quantum computers can be
powerful as promised
and it’s the right time for investment
to now make them useful
okay thank you very much
you