Quantum Error Correction

The future of quantum information science and engineering promises computers with unprecedented speed and capabilities. Mikhail Lukin, an NSF-supported professor, discusses his work with neutral atom qubits, why error correction is important in quantum computing, and his journey through starting a company and moving into industry.

Episode Transcript

(00:03):
This is
the
Discovery
Files
podcast
from
the U.S.
National
Science
Foundation.
Collaborative
partnerships
between
academia,
government
and
industry
across
the country
are driving
breakthroughs
in quantum
information
science
and
technology,
taking
advantage
of the
physical
properties
of the
quantum
scale
universe.
Like
superposition
and
entanglement
for novel

(00:23):
approaches
and
applications,
these
breakthroughs
can enable
new areas
of
scientific
exploration
and
technologies
that
enhance
human
health,
national
security,
and
economic
competitiveness.
We are
joined
today
by Mikhail
Lukin,
co-founder
of QuEra
Computing,
Inc.,
and a
research
professor
whose group
made
headlines
when their work
was named
the Physics
World 2024
breakthrough
of the year

(00:43):
for
advances
in quantum
error
correction.
Professor
Lukin,
thank you
so much
for joining
me today.
Thank you
for
having me
Nate.
So
I want to
start by
asking you
the really
broad
question.
When
someone
hears
the term
quantum
computing,
what
are you talking
about?
So maybe
I start out
by telling
a little
bit of
a history
in the field.
And so
what happened
already?
Sometime,

(01:04):
like
several
decades
ago, people
realized
that
the
progress
in
developing
the
computing
technology
will
ultimately
result in
elements
of
computers
being of
quantal
size.
People
started
asking,
you know,
what
would happen
if
their transistor
will become
the size of
the atom?
Will
it mean

(01:24):
the end
of their
progress?
The end of
Moore's Law
or
something
else?
And then
very
quickly,
people
figured out
that
actually,
this is
of course
a challenge,
but
it's also
an
opportunity.
So in
particular
particles
and this
kind of
of the
microscopic
size set by
the laws of
quantum
mechanics,
which
by now
is hundred
years old.

(01:45):
And by now
these
laws are
very well
established.
But
they are
kind of
unusual.
They are
weird.
And also,
for
example,
a particle
can be a
different
places
or
different
states
at once,
like this
chair
can be here
in my room
and
your studio
at the
same time.
This is in
principle
allowed by
quantum
mechanics.

(02:06):
This is
very odd.
It
never happens
in
practice.
But never
the less.
This is the case,
and then
people
realized
that if
you now
apply these
laws of
quantum
mechanics
to a
computational device,
then this
kind of
computational
device
or devices
can really
have
extraordinary
properties.
In
principle,
they can solve

(02:26):
some
problems
exponentially
faster
than any
classical
computer
can do.
These ideas
were kind
of
pioneered
and
cultivated
by people
like
Feynman,
you know,
40
years ago.
But
of course,
at the
time,
the idea
that you
can store
information
in one atom
was a
complete
fantasy.
Fast
forward,
you know,

(02:47):
40 years
later,
we are
actually
experimenting
precisely
with
these systems
we can
create
arrays of
individual
atoms
which
we can
control,
fully
and encode
bits of
information
we call them
now
quantum
bits.
And
then what
we'll also
can do
we can
do logic
with these
bits,
creating

(03:07):
basically
this kind
of quantum
revolution
of
the processor
that people
have,
you know,
dreamt
about.
And
for now,
full
disclosure,
despite the
amazing
progress
in this
field,
you know,
we're still
in the very
early stages.
So we are
just now
starting
to think,
how will
actually
the
architecture
of real
computers
will
look like
this is
still

(03:27):
the
frontier
of science
and
technology
question.
But even
more
importantly,
we still don't
know what
will be
the killer
applications
of quantum
computers.
You can
say, wait
a second,
you know,
how
can this
be?
You know,
that is
actually
this
quantum
isn't
in use
every day.
But
actually,
maybe it's
not
so unusual
because
even this
more mature

(03:48):
areas
like AI,
we
still don't
know what
they will
do.
We of
course
will have
much better
Gmail
than ten
years ago.
But in many
of these
areas,
you know,
things
happen,
you know,
hand in
hand.
So to maybe
give you
a short
answer
to your
question
in some
way,
quantum
computer is
a new type
of
scientific
and
technological
instrument.

(04:08):
You might
want to
ask
a question
whether
it's
appropriate
to call it
a computer
now, but
it
certainly
can and
does
execute
some
computational
tasks.
And
these are
the tasks
which are
hard to
execute on
conventional
computing
devices.
So in
your group
you're
working
with
neutral
atom
qubits.

(04:29):
Can
you kind
of explain
what
those are
and why
they might
be
important?
These are
literally
individual
atoms
which are
we
call them
trapped.
They're
kind of
grabbed
and
maintained
by
something
which is
called
optical
tweezer,
the focused
beam
of light.
And it
literally
can
actually
capture
an atom
at the
point of

(04:49):
the highest
light
intensity.
And these
tweezers
are focused
in their
vacuum
chamber
where
the atoms
are
extremely
well
isolated.
There is
nothing
around
them,
and
they're
focused
so tightly
that each
tweezer
can grab
and hold
at most
one atom.
Then from
these
atoms, we

(05:10):
basically
build
artificial
kind of
materials.
If
you want,
you know
where
we can
position
these atoms
at will.

We can
encode
quantum
information
by using
what
we call
spin states
of the
atoms.
They are
kind of,
long lived,
states

(05:30):
which
can hold
or maintain
these
quantum
superpositions
for
very long
time,
for
seconds.
And what's
also kind
of special
is that
not only
we can
assemble
this kind
of
artificial
atomic
system
in
the beginning,
but
we can also
change
the
positions
of these
tweezers
during the
computation

(05:51):
process.
And
that way
we can
actually
change
the
connectivity
of the system.
So and
it's
actually
very
different
from
conventional
like for
example,
in your
smartphone,
the processor
right, is
something
which was
actually
designed
years ago.
And then
it was
actually
manufactured
by using
optical
lithography.

(06:11):
But
the positions
and the
connectivity
of this
processor
is
basically
determined
at the time
of design.
So this is
very
different.
In our case.
We can change
the system,
the
connectivity
as it
proceeds,
as the
computation
proceeds.
And one
I only have
to think
is a way.
How we do
quantum
logic is
we excite
the atoms
into the
so-called

(06:31):
Rydberg
states.
So
these are
the states
where
electron,
you know,
ventures
very far
away from
a nucleus.
And
that way
these atoms
become very
big in size
for a short
amount of time,
and then
they can
talk
to each other.
And that's
how we
execute
quantum
logic.
How do you
use the
neutral
atom qubits
to error
correct?
To answer
this
question,
we should
just

(06:51):
venture
into and
talk about
what is
error
correction
and why
is it
needed.
To answer
this
question,
I would
like to
return back
to kind of
this idea
that we
discussed
earlier,
where
the core
premise of
quantum
computer
is this
idea of
quantum
superpositions.
These
superpositions
do not
occur
or do not
exist
in real

(07:12):
life,
like this
chair,
for
example.
It's
always, I’m
sitting on it
now.
You can
ask, why
aren’t
these
superpositions
everywhere?
The reason
is
fundamentally
these
superpositions
are
extremely
fragile.
This kind of
implies
that you
involve
this
behavior
of these
kind of.
For example
one
electron
can be
in two

(07:32):
superposition
of two
states at
a time
is
something
that
you know
has been
seen in
many
experiments.
With the
large
scale,
These superpositions
just do
not exist
in the way
building
a quantum
computer
involves
creating
this type
of
superposition.
Of the
computing
register.
I mean,
this is
extremely
challenging,

(07:53):
and the way
how it
manifests
operationally
is that
if you
do the
computation,
you do this
quantum
logic,
which
we talked
about
a little bit
before.
You
typically
make this
small
probability.
And
this probability
can be
very small.
But
you make a mistake
and after
you execute
it many of
these
steps,
eventually
what
will happen

(08:13):
is that,
you know,
these
mistakes
pile up,
you know,
and
essentially
them
your
registers
in a state
which is a complete
kind of,
you know,
mess,
right?
There is nothing.
There is no
answer to
the
computational
problem
you're
trying
to solve.
And people
recognize
it very
early on.
In fact,
about
30 years
ago,
after
Feynman's

(08:33):
pioneering
ideas,
there has
been
actually
a quite
famous
debate
between
Rolf
Landauer
who was,
a very
famous
condensed
matter
physicist
and
he's also
someone who
really made
major
contributions
to the
theory
of
classical
computation.
And,
at the
time,
a
young scientist

(08:54):
whose name
is Peter
Shor.
Basically
what
Landauer
and others
pointed out
at look,
I mean,
this
quantum
computation
is
essentially
kind of
an analog
device.
The reason
is, is
that
quantum
mechanics
is
described
by linear
equations,
and this
linear
equations
imply
analog
evolution.

(09:14):
And what
you that
in one of
his papers
Landauer,
he brought
he drew
this,
you know,
very famous
kind of
classical
transistor
curve
where,
you know,
classical
transistors
basically
is either
a 2.0
or a 2.1.
And this
transistor
graph
is
nonlinear.
So
even if your
voltage
fluctuates
slightly
in your
classical
processor,

(09:34):
right,
you know
that
you will
not flip
0 to 1.
So this is
this
non-linearity.
And
fundamentally
the
dissipation
which
actually
stabilizes
this.
And
this is
very much
counter
to this
kind
of idea
of quantum
mechanics,
which
actually
forced
linear.
But
moreover,
for
the system
to sort of
for the
state
to stay
quantum,
you have to
avoid
dissipation,
because

(09:54):
dissipation
is what
causes
quantum
systems
to cease
their
quantum
behavior
to become
classical.
That's
what kills
this
superposition
states.
At the
time, he
pointed out
that
this idea
of quantum
computers
and so on
is
fantastic
in
practice,
or
in theory.
But
it will never
work in
practice
because
of this
kind of
configuration.
It's
a linear

(10:15):
system,
right.
And that's
where
this idea
of quantum
error
correction
has been
upped.
And what
quantum
error
correction
does it
uses
redundancy
much
like many
classical
systems to,
for
example,
to make
something
robust,
you make
your
process
redundant.
For
example,

(10:35):
if you want
to encode
classical
bits zero,
you don't
just use
one
zero, use
three or
more.
Right.
And then
you do like
a majority
voting,
right.
This one is
going to
be flipped.
You know
you
basically
try to
correct
this error.
So kind of
early on
when people
started
thinking
about it,
it wasn't
clear that
this idea
can be
extended
to quantum

(10:55):
systems
for
multiple
reasons.
One is
that you
cannot copy
quantum
information,
so it's
called
no cloning
theorem.
You cannot
clone
quantum
state.
Moreover,
if you make
the
measurement,
you
typically
destroy
the quantum
state.
That's
another
property.
So it's
kind of
very much
related
to this
idea
of this
quantum
mechanics

(11:15):
being
linear
system,
which has
to be
isolated
from
all kind
of sources,
but
nevertheless,
it
turns out
it is
possible
to do
quantum
error
correction.
And the way
how you do
it,
you use,
something
which is
called
entanglement.
Entanglement
is very
much
related
to this
idea of
superposition.
Entanglement
is
basically
corresponds

(11:35):
to
superposition
of many
kind of
distinct
particles,
like for
example, in
a way,
that's
what gives
quantum
computers
a power
to begin
with.
But
it actually
turns out
that you
also can do
quantum
error
correction
by taking
your qubit,
and
delocalize
quantum
information
over many
physical
qubits.

(11:55):
And that's
the kind
of
ingenious
fundamental
idea
of quantum
error
correction.
And then
what
you can do
if you want
to detect
errors,
you can actually
make some
measurements
of some
quantities
which are
local
quantities,
where
you can
check
if errors
have
occurred
without
destroying
these
underlying
quantum
states.
And that's
actually
also
introduces

(12:16):
this
non-linearity
and
introduces
dissipation
that the
Landauer
has
been kind
of thinking
about.
So in
the way
it really
comes full
circle of
these
things.
And
actually,
I must
say on this
as a side
note,
this
has been
a
fascinating
debate
and
actually
someone to
talk to
about.
It could be
Peter Shor,
but okay.
But you
know, now
fast
forward

(12:36):
this year
is 100
years of
quantum
mechanics.
And
there is
a 40 years
of quantum
error
correction.
So and
you could
ask, oh,
wait
a second,
why has 40
years past,
you know,
why and
why don't
we have
quantum
computers.
So
it turns out
that
implementing
these ideas
in the lab,
in practice
is
extremely
challenging.
And for one
is that
you need
the
redundancy
for each

(12:56):
what
we call
logical
qubit,
protected
qubit.
We need to
have many
physical
qubits
to encode
the
information.
Moreover,
once
you encoded
this
information,
because
this
information
now is
delocalized
over
many
qubits,
it becomes
hard for
environment,
for
example,
to
manipulate
this
information
accidentally
let's say
geo error
correction,
but also

(13:17):
it becomes
hard
to
manipulate
this
information
deterministically
at will.
That is
the reason
why, up
until
very
recently,
the kind of
most
advanced
experiments
have been
limited
to a small
number
of small
meaning,
you know,
low
redundancy,
logical
qubits,
you know,

(13:37):
and only
will last
maybe 2
or 3 years.
We have
really seen
this field
of
experimental
quantum
error
correction
to
take off.
Now, with
all of this
big
introduction,
let me now
explain
how
in our approach,
we use,
neutral
atoms to do
error
correction.
So
the fundamental
idea is
related

(13:58):
to the fact
that we
keep
and hold
our atoms
in these
optical
tweezers,
and we can
change
the
positions
of these
tweezers
at will.
And
moreover,
what
actually
turns out
to be
very
important
with this
approach,
we can use
optical
holography
to
literally
multiplex
these
tweezers,
basically
just
starting

(14:18):
with one
beam of
light
and
passing it
through a
hologram,
which
or from
this time
dependent.
Or we can
actually
create
multiple
tweezers
and trap
multiple
atoms
in these
tweezers.
And then
just by
using
this one
control
knob,
for
example,
the
position
of this
laser beam,
we often
can
actually
move the
entire
block of
the atoms.

(14:38):
That
turns out
to be
very
instrumental
for
realizing
quantum
error
correction.
So in
particular,
in this
block of
the atoms,
we now
can encode
this
logical
qubit.
We can use
this
redundancy.
And
whereas
normally
you would
need to
control
each of
these
physical
qubits
individually.
By using
this
optical

(14:59):
tweezer
approach
and this
holographic
technique,
we can
control
all of
these
block
of atoms
at once.
And that
allows us
to
essentially
very
efficiently
control
all these
redundantly
encoded
logical
qubit.
And, for
example,
move these
logical
qubits
around
to make
them kind
of
interact.
Usually
it's
very hard
because now
this

(15:19):
quantum
information
is
delocalized.
And
that turns out
to be
a very
powerful
approach
to actually
I'll just,
you know,
implement
these
logical
qubits, but
start
building
algorithms
and
circuits
with these
logical
qubits.
And I think
that's
really kind
of
a
revolution
of the last
few years
that we
have moved
from
just kind
of trying
to task,

(15:39):
maybe
individual
components
of this
fault
tolerant
quantum
computers
to actually
building
circuits
and
starting
using
these
strategies
to
implement
these
small scale
algorithms.
And it's
very
exciting.
It's
very special.
So as
you're
developing
these
techniques
and getting
more
control
of things,
when
did you know
it was time
to start
a company?
Oh,
that's a

(15:59):
hard
question
to answer.
What is
special
about
this field
of quantum
computing
and quantum
information
is that it
uniquely
combines
now the
basic
science,
you know,
the cutting
edge
engineering
and
this kind
of
application.
What we
talked
about,
and it

(16:20):
became
clear
a few years ago
that,
you know,
with all of
these
ingredients
which we
develop,
we need to
really
start
thinking
about
building
systems
and also
kind of
deploying
these
systems
to
accelerate
application
development.
So that's
when
kind of
we thought
that maybe
complementing
the kind
of cutting
edge
science
work that
we have

(16:40):
been doing
are still
doing
in our labs.
It would make sense
to actually
start
an effort
in
industry.
You know,
I would say
that
still
remains
the case
today.
I think
it's
kind of
unique
because,
you know,
you
basically
would like
to bring
all of
these
forces
together
to really
advance.
And you
actually,
it's
necessary
to
bring
all of

(17:00):
these
forces
together
to advance,
this
frontier.
What
have been
the main
challenges
and
benefits
working
into the
quantum
industry
end of it?
The effort
in the
industry is
kind of
different
from what
you can
do in
the
academic
labs.
So
in a way,
in the
industry,

(17:20):
you can
really
deploy
quite
substantial
resources.
You can
also push
and perfect
engineering.
And also
develop new
tools,
new
technology
tools,
which are
necessary
to sort of
combine
all of these
things, to
bring this
product
to a
customer.
Right.
And that's
the kind
of thing
which
one can
do in

(17:41):
Academic
Lab.
As an
example,
our startup
company,
QuEra
Computing,
deployed,
a system
in Japan,
you know,
in the
Japanese
Institute
of
Standards
and
Technology,
and it's
kind of
building
a community
around
these machine.
I mean,
this is
something
that we
can not do
in our
university
labs.
At the same
time,
our kind of
university

(18:01):
work,
you know,
still
is
proceeding
and in
many ways
has been
accelerated
with
the help of
the
technology,
which
has been
developed
at QuEra.
And what's
distinct
is that,
you know,
what we do
in our
university
still,
we are doing
pathfinding.
You know,
we're
exploring
different
things
and we are
trying to
do things
which
would be
very hard
to do
in

(18:21):
a company
setting.
So that's
why
I view this
as a kind
of very
complementary
effort.
How has
NSF
impacted
that
effort?
This field
of neutral
atom
quantum
computing
have seen
truely
a
revolution
in oh,
the last
five years,
I would
say that's
probably
one of
the most
remarkable
things
which

(18:41):
came out
from this
national
quantum
initiative.
This revolution
has been
like
basically
from the
point
of view
of where
these
neutral
atoms have
not even
been on
our radar
to actually
becoming
kind of
leading
quantum
computing,
quantum
information
technology.
So our NSF
impacted it
in multiple
different
ways.
First of
all,
the reason
for this

(19:02):
fast
progress
is that it
relies on
the
foundational
knowledge
on this
atom
trapping
control.
And
manipulation
which has
really been
developed
over
the last
40 years,
it
resulted in
multiple
Nobel
Prizes.
This is
the reason
why we can
now, in
a way,
the speed
of light

(19:22):
kind of,
you know,
because we
are
utilizing
all of
these
technology
tools.
So at
the same
time, NSF
also quite
directly
impacted
some of
this
development
because it
basically
provided
the
foundational
support
for, for
example,
developing
this atom
trapping

(19:42):
technology,
realizing,
for
example,
quantum
logic by
exciting
atoms in
Rydberg
states
more
broadly,
these kind
of
innovations,
and in
terms
of
developing
control
tools
and
techniques,
which happened,
you know,
will last
two
years, saw
the in
combination.
I think
these are
the things
which
really
makes

(20:03):
us as
a community
as it is
as
a country,
a leader
in this
field.
And,
this,
support
is still
essential
to keep
pushing the
frontier
and,
and to,
you know,
eventually
figure out
what we
can use
these
quantum
computers
for, which
I think is
still
is a big
outstanding
challenge.
And maybe
I'll just
add that
the future

(20:24):
applications
of quantum
computers
will be
almost
certainly
scientific.
That is
what happened
with
many other
technologies.

This is
what in
the next
maybe five,
maybe even
ten years
is really
the most
exciting
frontier.
So, I mean,
there
are some
really
challenging
scientific
problems
for which,
quantum

(20:44):
computing.
And more
broadly,
quantum
technology
could
really
offer
some very
special
solutions.
How
do you think
quantum
technologies
might
impact
the average
person's
day to
day life
in that
next
ten years?
My
anticipation
is that
the first
applications
will be
scientific,
but often
the path

(21:05):
from
thank you
for
discovery
to real
world
application
is actually
not very
long.
And in
particular
what
quantum
computers
are
really good
at.
They are good
at solving,
you know,
hard
problems
which are
fundamentally
quantum,
and that,
involves

(21:26):
solving
some
material
science
problems,
designing
new
materials,
designing
new drugs
that
involves
solving
problems
in nuclear
physics.
These are
scientific
problems.
But
if you have
a new tool
which, for
example,
allows you
to kind
of engineer
new
solutions
for these
problems,
often
we will see
the impact

(21:47):
of that
in a way
which
is very
hard to
predict.
But
you know,
most likely
will be
quite
transformative.
I have two
more
questions
for you
today.
I wanted
to throw
a kind
of fun
one
in there.
As a
hockey fan,
I have to ask
you about
quantum
hockey.
Yes.
So I,
I see
on the
whole
probably
referring
to our,
you know,
famous
or infamous
quantum
supremacy
hockey game
at the

(22:07):
March
meeting.
You can read
about it
by now
in science
magazine.
You know,
notice
that it was
April
1st issue.
But
I should say
this is a
very
special
tradition.
So just
to explain
to the
listeners,
so starting
about
maybe 5
or
6 years ago
and
at every
March
meeting
with have a
hockey game
where,

(22:27):
you know,
basically
atomic
physics
team
is playing
against
solid state
qubits.
Basically
superconducting
teams were
you know
versus
atomic
qubits.
And
this was,
you know,
started to
be a game
for quantum
supremacy.
And it's
a lot of fun,
I should say
that we
won all of
them so
far....
So the
outcome is

(22:47):
very clear.
But I think
it's a
great way
to build
the
community.
I mean,
this is a
very
competitive
field,
right now.
And people
are working
very hard.
And I
think,
you know,
at the end
when we
advance
our,
neutral
atom
qubits,
we learn
something
which
impacts
other
areas.
You know,

(23:07):
likewise,
part of
this kind
of most
recent
spectacular
advances
is
that they
will build
on
the foundations
which was
established
by,
for
example,
trapped
ion quantum
computers.
I mean,
to close
the system
to us
to share
a lot
of
technology,
but also by
by
superconducting
qubits.
I think
this
kind of,
you know,
building
a community
where
we learn
from each
other
or benefit

(23:28):
from each
other.
And to
Frank,
you know,
both on the
academic
side,
but also on
the
industry
side,
it's very,
very
special.
And this
phase of
the
progress
is really
accelerating.
And
that's why
this
quantum
supremacy
hockey game
is very
important
enough.
So for the
for the
last
question
today,
I want to close
out by
asking you
about
the future
a little bit.
What are the
most
important

(23:49):
problems
to solve
next to
bring
quantum
computers
really
to
fruition?
And so
I am now
becoming
increasingly
convinced
that
quantum
computers
of
sufficiently
large scale
will
eventually
be built.
So it has
been
a question
for
the past,
you know,
20,
40 years

(24:09):
whether
these ideas
of error
correction
can really
work.
And I think
it's very
clear
that they
actually
work.
And
it's also
very clear
that
they will
work
at scale.
So
I think the
central
challenge
and the
most
interesting
question
now is
what
will be
the first
applications
of quantum
computers.
And
as already
mentioned,
most likely
they will be

(24:29):
scientific,
but then
most likely
again
from the
scientific
applications,
something
will emerge
which will
affect
the real
life of
people.
We are now
just
entering
this kind
of era
where
we can
really
start
using these
machines,
which
we are
building,
to
really kind
of ask
and answer
these
questions,
and that's
very,
very
special.
That's
a very,
very
special

(24:49):
time.
It's kind
of
comparable
to the time
when first
classical
computers
were built
and,
you know,
kind of
deployed.
Talking
about
industry,
we now have
a way
to
interface
with
customers,
with people
who don't
actually
care
what's
inside
these
quantum
computers,
but
they would
like
to know
some
answers

(25:09):
to
the problems
which
we have.
And I think
this is
the era
in which
we are
now
entering
and I
think
really
very
special,
very
exciting.
Special
thanks to
Mikhail
Lukin
and
Alexander
Cronin.
For
The Discovery
Files.
I'm Nate
Pottker.
Watch video
versions
of these
conversations
on our

@NSFscience
YouTube
channel.
Please
subscribe
wherever
you
get podcasts
and
if you like

(25:29):
our
program,
share
with
a friend
and
consider
leaving
a review.
Discover
how the
U.S.
National
Science
Foundation
is
advancing
research
at NSF.gov.