►
From YouTube: QIRG Interim Meeting, 2021-03-24
Description
QIRG Interim Meeting, 2021-03-24
A
B
And
I
see
the
little
icon
that
says:
recording
in
progress
so
for
the
54
or
50
or
so
of
you
who
are
here.
We
are
recording
the
seminar
and
I
hope,
we'll
be
able
to
post
the
link
to
the
qrg
mailing
list
later.
So
if
you
speak
up
and
or
turn
your
video
on,
you
may
be
recorded.
A
Yes,
okay,
so
with
that,
let
me
hand
over
to
maddie
now
a
final
reminder:
if.
A
If
you
still
haven't
filled
out
the
blue
sheets,
please
fill
them
out
and
now
maddie
the
floor
is
yours.
C
Awesome
awesome,
okay,
so
I
I
cannot
find
the
I.
I
don't
think
if
the
turning
the
video
option
is
still
on
so
yeah,
because
that's
why
I
stopped
sharing.
I
thought
this
is
something
that
doesn't
show
okay.
A
In
that
case,
we'll
just
have
to
live
without
the
video
something
to
figure
out
for
the
next
seminar,
at
least
we'll
know.
A
C
Okay,
I
hope
you
all
are
seeing
my
screen
yeah.
I
see
your
screen
yes
very,
very
hard
if
you're
not
seeing
this
screen
awesome,
I'm
very
very
happy
to
be
here.
I
hope
everyone
is
having
a
good
day
afternoon
night,
depending
on
where
you
are
I'm
guessing,
I'm
the
only
one
in
the
east
coast
in
the
morning
time
so
for
the
next
like
30
minutes,
40
minutes
or
so
I'm
just
gonna
talk
a
little
bit
about
the
the
work
that
we're
doing
here
at
cunyx.
It's
a
startup
in
brooklyn,
hi.
C
C
A
Okay
and
some
people
has
already
printed
out
that
the
audio
is
fine
for
them,
I'd
say
continue
and
if
somebody
has
a
problem,
can
they
contact
me
in
the
chat?
Okay,
fantastic
yeah.
C
So
so,
yeah
for
half
an
hour
or
40
minutes
I'll
talk
a
little
bit
about
the
things
we're
doing
at
qnx
to
develop
practical
the
elements
up
necessary
for
a
practical
quantum
network
that
operates
at
room
temperature.
What
I
mean
by
practical
quantum
network
is
that
not
only
devices
that
operate
and
show
the
quantum
performances
you
need
in.
In
our
cases,
mainly,
we
are
focused
on
quantum
memories
and
entanglement
sources,
but
also
devices
that
are
robust
and
feel
deployable
enough
that
they
have
a
perspective
to
go
out.
C
There
be
integrated
into
different
networks,
eventually
into
the
telecom
infrastructure
and
and
can
be
reliably
used
for,
let's
say
five
to
ten
years
without
damage
maintenance.
It's
a
big
difference
compared
to
how
quantum
computation
scheme
is
so
we
we
of
course
avoid
technologies
that
needs
carogenic,
cooling
or
even
laser
cooling,
magneto,
optically,
trapped,
atoms
and
things
like
this.
So
the
focus
of
this
talk
is
going
to
be
mainly
on
the
challenges
that
it
takes
to
to
develop
these
technologies
at
room
temperature
and
make
them
work
robustly
for
for
extended
amount
of
time.
C
So
we
don't
need
to
keep
sending
grad
students
across
the
country
to
to
fix
things
every
time.
Something
goes
wrong.
Okay,
let's
see
okay,
yeah
awesome,
so
you
all
are
in
this
this
panel.
So
I'm
guessing
this
slide
is
a
little
bit
redundant
for
you.
Why
we
care
about
quantum
networking?
I'm
gonna
do
my
best
to
entire
entirety
of
this
talk
to
avoid
the
use
of
the
word
quantum
internet.
So
when
I'm
saying
quantum
network,
I
mean
a
much
much
much
simpler.
C
Everything
excluded.
All
classical
layers
is
excluded
of
the
future
quantum
internet
networks,
if
they
ever
happen,
but
but
quantum
networking
is
just
just
all
the
quantum
elements
of
the
the
network.
One
of
the
key
reasons
people
really
like
these
is
that
it
gives
us
this
option
of
eventually
connecting
quantum
computers
to
each
other,
because
that
really
gives
us
this
ability
to
harvest
the
full
potential
of
these
computers
as
they're
getting
developed.
C
Of
course,
one
of
the
main
applications
or
most
known
applications
of
quantum
networking
is
the
secure
communication
which
to
scientists
is
one
of
the
most
boring
application
of
quantums
as
as
basic
as
it
gets.
But
but
it
works
and
it
and
it's
great
so
I
always
put
them
into
the
slides,
but
quantum
networking
can
be
used
for
so
many
applications,
not
only
pretty
much
connecting
anything
that
has
a
quantum
nature
in
it,
and
you
want
to
preserve
that
quantum
nature.
C
If
you
want
to
have
a
network
quantum
sensing
and
a
telescope
areas
of
optical
area
of
optical
telescopes,
quantum
computers
pretty
much
the
the
infrastructure
for
connecting
quantum
systems
to
each
other,
and
the
field
is
fortunately
moving
forward
very
fast
from
the
time
that
these
things
got
a
little
bit
more
heated.
And
then
china
sent
the
first
quote-unquote
quantum
satellite
up
there
till
this
last
year
that
there
are
like
so
many
of
the
things
that
I
put
here
from
celtic.
All
the
way
to
china
are
all
happening
in
2020.
C
There
are
so
many
researchers
across
the
airport
that
are
trying
to
implement
this
network,
so
not
only
to
realize
the
devices
but
also
like
much
more
focused
than
individual
devices.
Now
there
are
much
more
focus
into
the
network
elements
of
how
these
devices
are
going
to
come
together
and
work
together,
which
is
which
is,
of
course,
a
very
happy
site
for
people.
C
Like
me,
usually,
when
we
talk
about
quantum
networks,
I
don't
want
to
use
the
word
generations
either,
but
most
of
the
networks
to
date,
especially
on
the
commercial
side,
are
just
direct
qubit
transmission
ones.
These
are
like
qkd
quantum
key
distribution
networks,
things
that
you
can
even
right
now
buy
commercially
from
companies
like
id
quantique
or
so
just
the
directing
alice's
as
a
bunch
of
qubits
to
bob,
mainly
with
the
focus
on
this
security.
Quantum
secure
networks.
C
C
But
the
good
thing
about
this
is
that
there
are
a
lot
of
protocols
in
distributed
entanglement,
including
entanglement
swapping
that
that
gives
you
this
prospect
and
this
ability
to
eventually
extend
these
distances
and
and
be
able
to
use
multiple
entanglement
sources
to
to
to
share
this
entanglement
between
alice
and
bob
the
further
away
they
are
from
each
other.
So
many
experiments
have
been
done.
C
C
Now
what
I
mean
with
entanglement
distribution,
there
are
so
many
different
protocols
and
different
ways:
people
people
trying
to
envision
this,
this
distribution
of
entangled
photons,
but
basically
that
the
concept
is
a
little
bit
simple.
The
general
concept
is
that,
let's
say
I
start
with
two
entanglement
sources.
These
infinity
signs
and
each
of
them
are
generating
a
pair
of
entangled
photons.
Here
I
don't
care
what
type
of
entanglement
they
have.
Let's
say
polarization
and
these
sources
are
100
kilometers
or
so
apart
from
each
other.
C
I
want
to
end
up
sharing
a
pair
of
entangled
photons
with
alice
and
bob
who
are
200
kilometers
apart
from
each
other.
Of
course,
the
most
obvious
answer
is
to
put
an
entanglement
source
right
in
between
them
send
one
up
on
down,
but
we
want
to
find
a
way
that
the
overall
loss
is
less
than
the
loss
you
have
to
endure
when
you
send
things
to
long
fibers,
because
the
loss
in
optical
fibers
are
exponentially
increasing.
So
after
a
certain
kilometers,
it's
just
a
meaningless
thing
to
do.
C
If
you,
the
direct
transmission
of
these
photons,
so
so
the
trick
is
relatively
simple.
Although
the
implement
is
incredibly
hard,
but
there
are
a
few
in
a
few
elements,
you
usually
need
in
this
networks,
for
example,
not
all
entanglement
sources
are
naturally
made
for
telecom.
Oh,
let
me
turn
on
my
laser
pointer
yeah.
So
not
all
entanglement
sources
are
made
to
operate
at
telecom
wavelengths,
but
obviously
fibers
would
much
rather
have
those
wavelengths
to
to
at
least
give
you
the
minimum
loss
possible.
C
So
you
either
usually
need
to
do
a
frequency
conversion
before
entering
your
fibers
or
if
you
build
your
entanglement
sources
at
telecom
like
1550
or
something
usually
after
it
comes
out
of
the
fiber.
You
need
to
do
some
some
conversion,
because
it's
not
that
simple
to
have
just
entanglement
sources.
You
need
elements,
for
example,
like
quantum
memories
in
your
network,
quantum
memories
do
play
a
very
critical
role
in
terms
of
buffering
and
synchronization.
C
So
what
quantum
memories
do
for
these
protocols
is
that
they
take
entanglement
sources
that
are
not
deterministic,
so
you
don't
know
when
you're
generating
this
entanglement
purse
and
synchronize
this
protocol.
So
after
the
quantum
memories,
you
always
know
that
you
have
these
photons
coming
out
and
the
main
reason
you
care
about
this
is
that
the
the
main
process
of
swapping
the
entanglement
quote
and
unquote
is
happening
in
this
joint
photon
measurement
device.
This
ballistic
measuring
device
here
all
this
device
is
doing
is,
is
a
very,
very
interesting
process.
C
It
grabs
one
photon
from
one
source.
One
photon
from
the
other
source
does
a
measurement
on
them
that
results
in
entangling
the
other
remote
photons
that
they
never
met
each
other.
So
this
entanglement
swapping
can
happen
as
long
as
the
photons
arrive
to
the
ballistic
measuring
pretty
much
overlapping
at
the
same
time.
So
within
like
a
two
nanosecond
interval,
or
so
quantum
memories
play
as
the
role
of
the
the
buffering.
C
The
quantum
frequency
converters
are
the
interface
between
the
entanglement
sources,
frequency
telecom,
fibers
and
the
quantum
memories,
and
eventually
you
have
an
entangled
photon
between
alice
and
bob.
I
make
it
sound
simple,
but
but
in
practice
we've
been
trying,
and
by
v
I
mean
the
whole
science
community
been
trying
for
decades
and
we're
still
here
to
show
that
this
network
as
sophisticated
as
this
can
be
the
loss
of
a
typical
fiber
transmission.
C
Now
at
qnet
we
look
at
how
can
we
make
something
like
this?
The
most
practical
version
of
it,
and
one
of
one
thing
that
makes
things
practical
is
simplicity.
So
we
slightly
have
changed
these
protocols
and
adopted
to
a
protocol
that
at
least
we
hope
has
give
us
the
the
result.
Much
faster
and
much
more
robust,
the
the
main
difference
that
is
happening
here
is
is
at
these
entanglement
sources
yeah.
So
one
of
the
things
we're
doing
here
is
that
we
are
developing
these
entanglement
photos.
C
I
will
talk
about
them
during
this
conversation,
also
that
generate
the
pairs
of
photons
that
are
entangled
in
polarization
space,
but
these
are
by
chromatic
sources.
C
So
in
this
way
we
can
have
an
entanglement
source
that
is
simultaneously
compatible
with
both
the
quantum
memories
and
the
and
the
telecom
fibers.
So
the
way
we
can
have
these
nodes
set
up,
for
example,
in
a
version
that
you
see
right
here,
anything
that
you
see
blue
is
a
telecom
wavelength.
Anything
that
you
see
red
is
at
795
wavelengths
near
ir
wavelength,
so
in
in
a
protocol
like
this,
these
two
entanglement
sources
on
the
right
side.
Gonna
release
the
photons
that
goes
through
the
lung
fibers.
C
These
are
the
very
telecom
fibers,
while
the
memories
next
to
them
are
just
gonna
store,
the
795
photons,
so
they're
gonna
store
the
other
protons
and
simultaneously
the
other
left
entanglement
sources
are
gonna,
also
release
anytime.
A
successful,
build
ballistic,
swapping
happens
here
in
these
two
knots
they're
going
to
send
the
signal
to
the
central
node
and
the
quantum
memories
are
going
to
release
their
photons
you'll.
C
Do
a
bell
estate
measuring
here,
which
will
result
in
entangling
the
two
further
out
quantum
memories,
and
this
is
very
desi
chainable
you
can
have
well,
you
can't
have
infinite
of
them,
but
as
long
as
your
devices
are
low
loss,
you
can
keep
increasing
the
number
of
these
nodes
you
have
in
your
network
and
and
keep
swapping
the
entanglement
further
and
further.
C
So
there
is
no
quantum
memory
in
this
link
or
frequency
converter,
which
gives
me
a
better
future
of
mixing
these
photons
using
multiplexing
or
something
with
with
everything
else,
that's
going
through
those
fibers.
If
you
want
to
have
hybrid
digital
quantum
networks.
C
Another
very
good
thing
about
this
is
that,
as
you
can
see,
of
course,
we
don't
need
frequency
conversion,
anything
which
really
really
simplifies
things.
There
is
a
huge
different
level
up
like
having
three
quantum
devices
and
two
quantum
devices
that
need
to
work
together,
and
this
is
also
a
conversation
I'm
not
going
to
dive
into.
But
this
this
network
structure
needs
the
minimum
amount
of
heralding
so
a
lot
of
times.
I
don't
need
necessary
my
quantum
memories
to
be
heralded.
I
don't
need
my
quantum
memories
to
have
very
insane.
C
Like
a
storage
times,
you
can
adjust
these
networks
for
a
quantum
memory
that
has
a
reasonable
stress
time
of
like
a
millisecond
or
so,
and
as
long
as
I
have
a
few
anchor
nodes
across
the
network
that
can
herald
the
system,
I
can
always
implement
these
protocols
with
high
fidelities
and
and
defeat
the
loss.
Another
good
thing
about
this
is
that
the
devices
that
we're
building
are
designed
to
eventually
all
be
multiplex.
So
so
you
don't
need
to
have
just
one
layer
of
these
things:
the
quantum
memory
itself,
the
entanglement
sources.
C
Everything
can
be
multi
and
multiplex,
so
you
generate
many
pairs
of
photons
simultaneously.
Your
quantum
memory
can
store
many
many
of
these
photos
simultaneously,
so
you
can
scale
everything
up
by
just
simply
multiplexing
and
throughout
this
talk,
if
I
forget
this
to
mention
this,
everything
here
works
at
room
temperature.
C
But
the
whole
main
point
of
here
is
to
just
show
that,
of
course,
as
you
expect
for
for
any
with
any
basic
calculation,
you
can
show
that
a
protocol
like
this
has
a
much
much
better
transmission
or
or
can
defeat
the
loss
over
long
distances,
much
better
than
direct
transmission,
which
I'm
showing
here
with
this
blue
line.
So
so
something
like
this
can
these
repeaters
can
allow
you
to
defeat
the
exponential
loss
and
turn
into
something
much
more
manageable,
a
polynomial
loss
versus
the
distance
of
the
fibers.
C
C
C
Of
course,
for
these
plots
there
have
been
like
insane
number
of
assumptions
and
simplifications
here,
but
but
at
the
end
of
the
day,
this
is
always
what
you
expect
to
get
a
plan
nominal
loss
compared
to
an
exponential
loss.
Another
thing
that
I
want
to
talk
about,
but
it
won't
be
that
much
focus
of
this
conversation
is
the
at
qnc:
not
only
we're
building
these
quantum
hardware
devices,
so
we
are
building
the
entire
sources,
the
quantum
memories
and
the
much
more
simpler
devices
these
these
ballistic
swapping
devices.
C
But
we
also
want
to
have
this
prospect
of
eventually
integrating
these
devices
into
telecom
infrastructure,
at
least
for
our
own
field
test
tests.
So
we
can
put
them
out
there
so
we're
developing
some
additional
hardware.
We
call
them
our
quantum
support
hardware
that
will
help
make
the
everything
works
better
in
terms
of
making
the
infrastructure
a
little
bit
more
quantum
friendly
and
and
also
bring
everything
together.
So
just
the
three
of
them
to
to
mention
is
quantum
memories.
These
entanglement
sources,
they
all
have
their
own
pump
lasers.
These
are
different
transitions.
C
We
usually
use
rubidium
atoms
I'll
talk
a
little
bit
later,
so
these
are
all
need
to
be
these
lasers
that
are
comes
with
these
devices.
They
need
to
be
blocked
precisely
to
the
atomic
transitions,
and
so
so
we're
building
these
devices,
and
these
these
are
just
normal
locking
schemes,
but
but
every
device
can
block
four
to
eight
different
lasers
simultaneously,
so
you
don't
need
to
have
multiple
of
these
things
everywhere
in
a
rackmount
design
as
time.
C
Synchronization
is
incredibly
important
in
these
networks
because,
as
I
was
mentioning
the
ballistic
swapping
stations
really
care
about
when
the
photons
are
arriving,
because
these
are
joint
measurements.
So
if,
if
your
photons
are
nanosecond
aligned
with
photons,
you
need
them
to
arrive
within
the
same,
and
you
need
the
whole
network
to
be
synchronized
time
wise
everything
that
you
do,
the
pulsing
of
your
palms
they're,
posing
up
your
control
field
that
stores
these
photons.
Everything
needs
to
be
done
within
nanosecond
time
frame.
C
So
we
are
building
these
devices
that
you
that
can
be
used
as
to
trigger
all
the
electronics
down
to
sub
nanosecond
precision
and
another
element
that
we're
building
are
the
automated
polarization
compensating
devices
all
the
telecom
infrastruct
telecom
fibers
are
made
for
optical
communications,
but
they're
not
made
for
like
actual
single
photon
carrying
quantum
information
type
of
communications.
C
So
they're,
really
not
that
great
in
preserving
the
the
quantum
information
you
include
on
your
photons
and
regardless
of
what
type
of
polarization
what
type
of
encode
you're
using
and
quantum
encode
you're
using
for
your
photons,
you
always
have
to
maintain
the
polarization
these
fibers
are
not
polarization.
Maintaining.
You
always
have
to
compensate
for
it,
because
these
stations
really
care
about
indistinguishability
of
your
photons.
So,
even
if
you
have
time
being
qubits,
but
your
polarization
of
photons
are
different,
you
start
feeling
a
very
significant
loss.
C
I'll
show
this
actually
in
one
of
the
slides
when
you
want
to
do
these
swappings,
and
so
so
our
other
medicalization
compensating
device
can
do
the
job
for
you
that
it
calibrates
the
system
using
some
reference
photons
to
make
sure
that
they
compensate
for
what
the
fibers
are
doing
to
your
photons,
and
our
eventual
goal
is
to
be
able
to
put
all
these
devices
into
what
we
call
our
eventual
product.
The
quantum
rack
modules
where
everything
is
recommended.
C
Everything
is
ready
for,
let's
say
a
fiber
hub
or
something
you
can
just
put
one
of
these
racks
in
and
depending
on
the
node
that
you
have.
Of
course,
this
structure
of
the
rack
is
a
little
bit
different,
but
everything
is
built
to
be
modular
and
very,
very
easy
to
mix
and
match,
based
on
your
need
in
different
nodes
kind
of
ignore
this.
C
The
list
of
the
challenges
are
pretty
much
infinite
between
where
we
are
right
now,
until
we
get
to
to
a
robust
network
that
works
out
they're
integrated
into
the
telecom,
so
I'm
just
gonna
categorize
everything
in
in
four
different,
like
main
challenges
that
I
at
least,
we
could
think
of
one
of
the,
of
course,
the
main
ones
is
that
not
only
you
want
very
quote
and
unquote
strong
quantum
devices,
devices
that
are
very,
very
efficient
and
have
very
high
fidelity
at
quantum
levels
like
a
quantum
memory
that
can
restore
the
photons
with
very
high
fidelity.
C
But
you
also
want
something
to
be
very
very
reliable
for
years,
and
this
is
not
that
easy.
As
a
person
who
has
spent
the
past
eight
ten
years
of
his
life,
doing
experimental
quantum
physics
stuff,
a
lot
of
the
experiments
that
I
did
I
needed
to
during
my
phd,
especially
I
needed
to
stop
the
experiment,
literally
every
15
minutes
or
so
to
readjust
something
to
fix
the
temperature
and
something
like
that
and
continue.
Now
we
are
talking
about
five
to
ten
years
of
robustness,
so
the
skills
are
gently
different.
C
The
the
other
thing
that
matters,
even
if,
when
you
build
these
devices,
is
that
in
quantum
quantum
devices
are
super
needy
a
lot
of
them,
they
need
their
own
kind
of
input,
whether
like
in
terms
of
wavelength,
the
line
without
the
bandwidth
of
the
photons,
what
type
of
photons
you're
using.
So
it's
very,
very
important
to
have
a
product
suit
that
are
compatible
with
each
other.
C
Otherwise,
even
if
you
have
individual
quantum
devices
that
are
highly
efficient,
you're
going
to
massively
endure
losses
when
you
link
them
to
each
other,
let's
say
you
have
both
of
these,
and
now
you
have
to
deal
with
the
fact
that
I
was
just
mentioning
that
these
are
not
directly
gonna
kind
of
be
connected
to
each
other.
A
lot
of
times
there
is
gonna,
be
a
telecom
fiber
in
between
there's
gonna,
be
a
lot
of
fiber
back
patches.
So
many
things
in
between
that
we
as
researchers,
especially
quantum
researchers,
are
not
that
many
family
orbit.
C
So
so
it's
very
very
important
to
have
hardware
that
can
make
the
telecom
infrastructure
more
quantum
friendly.
So
when
you
use
them
to
plug
your
devices,
you
you
don't
suddenly
go
to
efficiencies
of
like
pretty
much
zero,
and
if
you
have
all
those
things,
then
then,
of
course,
when
you
go
towards
the
more
sophisticated
networks,
you
have
to
talk
about
the
issue
that
we
simply
right
now
don't
have
any
proper
protocol
that
can
connect
classical
quantum
interface
networks.
This
is
this
is
a
big
big
issue.
C
Well,
right
now,
no,
but
very
soon,
then,
then
you
build
these
elementary
networks.
Then
the
next
question
is
that
there's
a
huge
gap
between
what
we
call
a
simple
or
elementary
quantum
network
and
what
is
an
actual
quantum
internet
that
is
going
to
be
integrated
into
the
telecom.
So
this
this
the
last
challenge
pretty
much
requires
five
to
maybe
ten
years
of
collaborations
between
scientists,
networks
and
network
scientists,
network
engineers,
everyone
together.
C
So
so
we
can
adopt
these
protocols,
so
they
can
be
integrated
into
the
already
existing
protocols,
and-
and
this
is
we've
all
seen-
a
lot
of
proposals
nowadays
come
out
from
physics
labs
in
q-lan,
cube
and
stuff
like
this,
but
but
they
are
not
that
realistic
when
it
gets
into
terms
of
what
is
actually
out
there
and
and
these
devices
need
to
work
with
the
infrastructures
and
the
protocols
that
we
are
currently
using
for
internet.
C
Eventually,
I'm
talking
way
way
in
the
future,
but
but
this
is
a
research
that
needs
to
be
pushed
forward,
rather
rather
soon.
If
you
want
these
things
to
ever
happen
now,
everything
I
said
so
far
was
introduction.
So
hopefully
I
don't
bore
you
guys
I'll
talk
a
little
bit
more
about
our
quantum
memories
and-
and
I
just
going
to
mention
quantum
memories
and
entanglement
sources,
with
a
focus
on
quantum
memories,
because
that's
one
of
the
things
we're
getting
pretty
good
at
after
a
decade
or
so
research.
C
There
are
plenty
of
ways
to
create,
like
matter
interfaces
that
would
allow
you
to
store
temporary,
a
single
photon
with
the
polarization
or
the
information
encoded
on
them,
and
there
are
infinite
number
of
researchers
out
there
that
are
doing
infinitely.
Amazing
research
in
the
field
from
using
cold
atoms,
magneto
optically,
trapped,
atoms
or
literally
a
single
atom
trapped
in
a
cavity
all
the
way
to
a
little
bit
more
solid
status.
Stuff
like
silicon
vacancy
centers,
nitrogen
vacancy
centers
or
the
rarer
stops.
C
Crystals
are
very,
very
different,
very,
very
interesting
protocols
to
to
restore
photons
the
ones
that
we
are
more
focused
on
are
the
one
down
here:
the
room,
temperature
and
samples
to
be
able
to
soar
and
successfully
retrieve
the
photons
and
the
information
encoded
on
them,
using
a
rubidium
vapor
cell
that
is
operating
at
room
temperature,
of
course,
designed
with
above
room
temperature
they're
like
60
celsius
or
so,
but
but
nothing
cooled
nothing.
Laser
trap,
nothing
magnetically,
trapped,
very,
very
simple
system
to
design.
C
Technique
eit
is
a
very
elegant
method,
discovered,
I
don't
know
even
how
many
years
ago,
like
30
years
ago
or
so
in
which
you
can
induce
transparency
to
a
medium
that
is
supposed
to
be
naturally
absorbed
into
your
photons
by
that.
What
I
mean
is
that
we
use
rubidium
atoms.
C
So
if
I
grab
my
photons
and
I
make
sure
that
my
photons
are
sitting
on
the
transition
of
b
to
a
transition
of
rubidium
atoms,
I'm
showing
them
with
e,
of
course,
if
I
send
those
photons
into
the
medium
into
the
vapor
cell,
they're
going
to
get
absorbed
because
they're
literally
on
resonance
with
my
atoms
yeah.
But
the
good
thing
is
that
if
I
use
a
strong
control
field
omega
here,
that
is
also
on
resonance
with
my
with
my
atoms.
Through
these
land
schemes,
I
can
control
the
transparency
of
the
medium
under
proper
conditions.
C
Whether
this
control
field
is
on
or
off
will
decide
whether
the
medium
is
transparent
or
absorb
them
for
these
these
photos.
So
I
can
change
this.
By
having
my
control
field
on
and
off,
I
can
change
the
behavior
of
the
medium.
This
is
called
electromagnetic
use,
transparency,
and
this
allows
me
to
create
a
coherent
quantum
memory
in
which
this
is
my
vapor
cell.
Here
I'll
have
the
control
field
on
the
medium
is
transparent
and
the
control
plating
on.
C
So
when
the
photon
arrives,
it
can
enter
the
cell,
and
while
this
the
photon
is
inside
the
cell,
I
turn
up
the
control
memory
off
adiabatically
slowly
store
my
photon
cut
an
uncut.
I
have
the
photon
here
for
for
a
certain
interval
of
time,
and
then
I
can
turn
on
my
control
field
again
and
allow
the
photon
to
leave
the
medium
and
continuous
journey,
of
course,
you're
working
at
room
temperature
so
and
and
you're
working
at
quantum.
So
everything
has
a
limit.
C
For
example,
we
are
here
targeting
for
these
photons
to
be
a
stored
in
a
millisecond
time
of
scale.
Nothing
really
further.
It's
it's
incredibly
hard
to
increase
these
things.
Although
people
have
done
some
very
awesome,
experiments
at
room
temperatures
up
to
second
longest
storage,
but
but
for
a
robust
device.
It's
not
that
easy.
But
good
thing
is
that
in
a
lot
of
these
protocols
for
quantum
men's
networking,
we
don't
need
anything
further
than
a
millisecond
or
so
storage.
C
Now
there
is
a
downside
of
using
room
temperature
stuff
beside
the
limitation
of
a
storage
time
and
that's
the
the
giant
amount
of
noise
you
have
to
fight
if
you're
working
at
room
temperature.
So
one
of
the
things
that
I
mentioned
is
the
control
field
itself
has
to
be
very
strong,
and
in
this
case,
because
of
the
doppler
broadening
it
has
to
co-propagate
with
your
input
qubits.
C
So
every
time
that
I
successfully
store
and
retrieve
one
good
photon.
Now
I
need
to
deal
with
two
very
large
categories
of
noises
and
very
vastly
different,
the
first
one
I
call
them
usually
technical
noises.
These
are
noises
directly
coming
from
your
laser
itself,
for
example,
the
fact
that
that
it's
it's
a
laser
field,
the
control
field,
so
you
have
to
be
able
to
filter
10
to
the
power
of
14,
different
or
unwanted
photons,
without
destroying
your
one
good
photon.
This
is
a
massive
amount
of
filtering
required.
C
It's
doable
and
we're
getting
very
good
at
it.
Thanks
to
the
fact
that
these
photons
are
6.8
gigahertz
different
in
frequency,
compared
to
your
good
photons
and
a
lot
of
times,
you
can
design
them
to
be
perpendicularly
polarized
to
your
good
photons.
Another
thing
that
you
have
to
get
rid
of
is
the
the
ase
noise
of
your
laser
itself.
This
is
the
amplified
spontaneous
emission
noise,
by
which
it
means
nothing
is
perfect.
C
Even
if
I
have
a
laser
that
is
lazing
at
this
very,
very
specific
wavelength,
it's
still
doing
it
with
a
certain
efficiency,
so
there
it
still
emits
photons
at
very,
very
low
powers
that
are
several
nanometers
or
so
wide.
C
You
need
to
be
able
to
also
filter
that
the
reason
these
are
different
is
because
you
can't
just
simply
use
optical
cavity
source,
or
things
like
that,
filter,
something
that
is
like
20,
nanometers
wide
and,
of
course,
that
there
are
counts
and
everything
comes
into
play,
you're
working
at
singapore
levels,
even
if
you
do
all
those
things
you
have
to
deal
with
the
fact
that
you're
at
room
temperature,
you're
shining
a
strong
control
field
into
a
medium
that
is
hot.
C
So
your
medium
itself,
this
control
field
can
interact
with
your
atoms
and
create
a
bunch
of
new
sources
of
noise.
For
you,
the
most
easiest
one
to
to
to
talk
about
is
the
spontaneous
ram
and
scattering
is
this
case
that
when
I
have
the
control
field
here,
if
my
my
states
are
not
perfectly
pumped
out,
if
not
all
these
atoms
are
empty.
C
Even
if
I
have
a
few
atoms
left
in
these
states,
that's
enough
to
start
pumping
them
up
and
create
a
spontaneous
emission
of
these
photons
that
are
now
exactly
at
the
same
wavelength
as
my
good
photons.
So
my
my
optical
cavities
are,
for
example,
useless
for
those
things.
There
are
several
atomic
processes
that
can
happen
that
can
result
in
photons
that
are
much
harder
to
filter
than
your
just
technical
noise
photons.
C
C
We
also
quantum
engineer
the
the
interaction
itself
to
suppress
all
the
possible
atomic
noises
to
to
minimize
these
noises,
because
because
our
because
they
are
pretty
much
immune
to
to
our
filtering
system,
so
we
do
need
to
engineer
them
inside
the
diaper
cell
to
make
sure
they
are
not
happening
now.
Another
thing
that
we
do,
of
course
this
matters
quite
a
lot
is
that
these
quantum
memories
should
be
able
to
not
only
restore
the
single
photons
themselves,
but
also
the
the
information
encoded
on
them
be
really
like
polarization.
C
So
these
quantum
memories
are
built
to
to
store
polarization,
and
we
can
do
that
simply
with
the
trick
of
a
dual
rail
quote
and
encode
quantum
memory.
All
I
need
to
do
I'll,
make
it
sound
simple,
but
but
pretty
much
all
I
need
to
do
is
to
grab
my
qubit,
that
is
in
a
superpolis
position
of
polarization
state
and
map
that
superposition
into
a
superposition
of
special
superposition.
C
So
the
up
and
down
red
superposition
using
optical
elements
like
a
beam
displacer
and
now
I
just
store
them
separately
and
identically
inside
the
vapor
cell
at
the
same
time
and
then
eventually
recombine
them
back
into
or
map
them
back
into
the
polarization
state.
So
this
way
I
can
preserve
the
polarization
and
just
to
show
you
how
it
looks
like
one
of
the
things
that
we
are
very,
very
proud
of
is
how
low
noise,
our
quantum
memories,
especially
the
prototypes
that
we're
building
right
now
are.
C
This
is
an
example
on
the
left
side
is
an
example
of
a
storage
of
a
pulse
with
average
one
photon
per
pulse.
That
is
a
swearing
for
like
a
micro,
two
microsec,
a
microsecond
or
so
for,
for
some
reasons,
we're
doing
this
at
low
efficiencies
to
to
match
with
some
previous
data.
So
this
part
that
you're
seeing
here
is
the
part
that
we
could
successfully
store,
but
with
five
percent
efficiency,
but
even
at
those
low
efficiencies,
you
can
just
compare
it
to
the
noise.
The
background
noise.
C
We
can
achieve
signal
to
background
ratios
of
20,
30
plus,
which
is
which
is
very
high
because
that
corresponds
to
to
fidelity
the
storage
facilities
of
96
plus,
which
at
room
temperature
is.
Is
I'm
still
fascinated
honestly
that
these
things
work
and
and
on
the
on
the
right
sides?
I
have
the
poincare
sphere
so
just
just
to
show
that
you
can
send
any
arbitrary
polarizations
to
these
quantum
memories.
C
As
your
input
signals,
and
if
you
measure
the
output
signal
you
you,
you
get
the
same
photons
but
a
little
bit
shrank
because
you're
you're,
you
still
have
noise
and
you
still
have
to
deal
with
those
things.
So
you
can
really
have
a
quantum
memory
that
destroys
polarization
pretty
much
with
very,
very
high
efficiency.
C
C
We
started
with
these
experiments.
These
are
the
things
I
did
during
my
phd
at
the
stony
brook
university.
So
there's
this
massive
optical
table
with
so
many
elements.
That
is
the
quantum
memory.
This
is
our
very,
very
first
prototype
of
quantum
memory
coming
out
of
qnx.
It's
right
now
actually
ready
for
our
very
first
cell
to
to
research
customers
significantly
smaller
and
has
all
the
elements
in
it
and
now
we're
developing
the
next
generation.
C
I'll
tell
you
why
we're
developing
this
in
a
second
which
we
expect
to
have
it
really
by
before
the
end
of
this
year.
These
are
rank
mounted
u2
size,
so
very,
very
suitable
to
just
put
them
in
a
rack
setup
and
use
them
much
much
smaller
and
robust.
So
we
have
a
very
clear
path
of
not
only
making
things
smaller,
but
cheaper
and
more
robust
and
and
in
the
engineering
side
and
everything
now.
C
I
think
I'm
talking
in
a
way
that
this
this
talk
is
gonna,
go
till
like
four
or
five
hours
from
now,
so
I'll
try
to
pick
up
the
pace
and
and
jump
over
a
couple
of
slides.
So
I
can
also
talk
a
little
bit
about
the
entanglement
sources,
but
this
is
a
little
bit
about
the
versions
that
are
different
like
generations
or
different
marks
of
the
quantum
memories
that
we're
building
at
qnx.
C
C
The
overall
transmission
of
this
system
is
going
to
be
high.
These
are
not
easy
to
increase
the
transmission,
because
you
deal
with
so
many
optical
cavities
and
everything
we
are
aiming
for
coherence
times
of
above
500
microseconds,
which
will
gives
us
the
performance
that
we
need
for
for
short-term
quantum
networks,
medium-length
quantum
networks
and-
and
these
devices
are
now
designed
to
be
a
stable
for
for
more
than
a
month.
We
don't
need
any
without
the
need
of
any
like
adjustment
on
our
side.
C
C
We
know
engineering
engineering
alone
is
not
going
to
take
us
that
far
we're
talking
about
devices
that
are
sensitive
to
temperature
change,
down
to
millikelvin
they're
sensitive
to
the
adjustment
of
the
lasers,
where
the
laser
heats
down
to
like
a
few
10
micrometer
or
so
so
we
are
trying
to
use
some
basic
machine
learning
algorithms
to
make
our
devices
self-diagnose
and
self-optimize
themselves
during
the
downtime
of
the
network.
So
the
quantum
memory
can
just
light,
run
a
light
through
different
elements
and
recognize.
C
Now
I
am
gonna
go
to
quantum
source.
I
do
see
some
messages
popping
up
people
talking
in
in
the
background,
if
the
questions
are
for
me
feel
free
to
ask
at
any
point,
especially
now,
because
I'm
gonna
talk
a
little
bit
about
the
quantum
sources
and
it's
just
that
they
come
to
my
screen
and
then
they
disappear
before
I
can
read.
So
I
don't
know
so.
A
In
that
case,
maybe
we
can
feel
the
question
or
or
two
so
there.
There
were
two
questions,
one
of
from
olaf,
which
was
answered
by
rod,
but
there's
also
a
question
from
thank
you.
D
You
can
hear
me
yes,
okay,
yeah!
No,
I
I
was
just
actually
wondering
in
this
specific
scheme
why
you
were
so
sort
of
putting
emphasis
on
that.
You
needed
these
memories,
but
I
guess
I
now
we're
getting
to
your
sources,
and
I
think
that
completes
the
story
as
to
why
you
want
to
store
your
entangled
state
in
something
else
than
what
you
use
to
make
your
entanglement.
C
Okay,
so
that
that's
a
very
good
question
I'll
just
quickly
go
back
to
it,
to
explain
where
the
quantum
memory
exactly
came
in
and
and
then
we'll
talk
about
the
entanglement
sources
yeah.
So
this
is
where
the
quantum
memory
comes
in
is
in
this
one,
critical
step
that
I
said
all
it
does
is
very
simple
yeah.
It
allows
you
to
first
wait
for
these
two
remote
blasted
swapping
to
happen
successfully
before
you
release
this
and
then
do
this.
C
C
C
Then
I
do
the
other
ones
I'll
end
up
with
this
problem
that
I
have
like
100
real
estate,
swapping
that
all
need
to
somehow
happen
successfully
simultaneously
for
me
to
be
able
to
to
implement
these
networks
when
I
scale
them
up
which
which
is
statistically
speaking
impossible.
So
all
these
memories
are
doing
is
by
breaking
these
networks
into
sub
sections.
It
increased
the
probability
because,
right
now,
if
if
something
goes
wrong
here,
these
memories,
these
two
memories
can
keep.
C
If
this
one
goes
wrong,
for
example,
but
the
other
one
can
do
it
successfully,
I
can
keep
my
photons
while
this
guy
repeats
again
and
again
and
again
until
it
does
successfully,
and
then
I
release
my
photons,
but
if
I
don't
have
the
quantum
memories,
if
the
left
one
does
it
successfully
the
right
one
fails.
I
have
to
get
rid
of
the
discard
the
left,
one
repeat
again
repeat
again
until
it
happens,
so
it
makes
it
much
more.
C
The
ink
affects
the
rate
significantly
to
have
the
quantum
memories
in
a
protocol
like
this
exactly
and
it's
really
not
about
the
type
of
entanglement
sources
we
build
is
pretty
much
about
all
the
tangamore
sources
are
the
is
the
intent
in
the
deterministic
nature
of
the
sources
that
cause
this
issue.
This
requirement,
okay,
so
the
quantum
sources.
So
I
said
a
lot
of
things
that
you
you
would
want
to
have
up
for
these
quantum
memories,
but
for
the
sources
you
also
have
a
bunch
of
requirement
requirements.
C
First
of
all,
you
want
something
that
is
easy
to
eventually
integrate
into
the
telecom
infrastructure.
At
least
we
care
about
that,
because
we
want
to
build
something
that
has
the
prospect
of
being
integrated
into
the
telecom
infrastructure,
but
you
also
want
something
that
is,
you
can
integrate
it
into
a
multi-modular
quantum
networks
with,
for
example,
quantum
memories
or
eventually,
like
quantum
sensors
quantum
processors.
So
if
you're
using
ion
traps
quantum
computers,
atomic
quantum
computers,
quantum
sensors,
a
bunch
of
them-
are
actually
based
on
these
ruby
domes
atoms.
C
C
You
want
to
be
also
eventually
have
this
concept
of
hybrid
free
space
free
space
network,
so
it's
good
to
develop
a
source
that
works
for
that.
Obviously,
you
want
something,
the
more
bright
it
is,
and
the
more
efficient
the
source
it
is
the
better,
but
at
the
same
time
you
need
it
to
be
very
robust
field
deployable.
So
you
pretty
much
need
everything
for
these
sources
yeah.
They
really
need
to
satisfy
so
many
conditions
and
before
you
can
take
them
out
there
and
implement
them
now.
C
Entanglement
sources
compared
to
quantum
memories,
are
devices
that
right
now
you
could
commercially
buy
out
there.
They
are
companies
that
sell
them
and
the
research
is
much
more
further
along
the
most
common
way
of
building
them
is
using
a
spontaneous
parametric
down
conversion.
So
you
use
these
crystals
that
these
are
not
non-linear
crystals,
that
they
absorb
a
photon
and
then
they
re-emit,
two
photons
with
the
total
energy
the
same,
so
each
of
them
are
half
of
the
energy
of
your
initial
proton,
but
they
are
correlated
in
the
polarization
space,
for
example.
C
So
you
can
use
these
techniques
to
create
pairs
of
photons
that
are
entangled
to
each
other.
If
you
start
with
a
blue
pump,
you
can
get
two
red
photons
that
are,
for
example,
frequency
entangled
with
each
other.
It's
a
very,
very
simple
process,
but
it
has
a
couple
of
downsides
that
makes
them
pretty
much
useless
for
our
applications.
C
The
biggest
biggest
downside
that
they
have
is
that
the
photons
that
come
out
are
very
wide
themselves,
so
the
photons
are
like
10
nanometers,
a
few
nanometers
wide
that
they're
in
line
between
the
frequency
of
space,
which
means
they
are
like
two
to
ten
terahertz
in
frequency.
Space
wide
are
quantum
memories
and
a
lot
of
other
different
quantum
devices
ion
traps
and
stuff
like
these.
They
do
not
want
anything
that
is
beyond
the
natural
line
of
the
line,
that
of
your
atoms.
The
quantum
memories
are
even
more
picky.
C
They
want
like
a
megahertz
or
so
so
a
lot
of
these
devices
when
you
interface
them.
That's
one
of
the
things
that
I
was
mentioning
before.
If
you're
interfacing
an
internal
source
with
a
quantum
device,
you
you
need
a
photons
that
is
at
the
right
wavelength,
but
also
the
line
right
line,
width
and
in
in
this
case
the
centennial
sources
are
a
factor
of
like
10,
000
or
so
off.
C
So
if,
if
I
have
a
source
right
now,
I'm
connected
to
my
quantum
memory,
even
if
the
quantum
memory
has
like
a
normal
efficiency
of
100
percent.
The
interface
efficiency
between
these
are
gonna
be
practically
zero.
Yeah,
it's
gonna,
be
a
meaningless
number
and
the
the
other
downside
of
them
is
that
they're,
usually
single
wavelength
yeah.
C
So
you
can
either
tune
this
depending
on
your
pump
to
create
the
photons
at
795
or
they
create
the
photons
at
1550,
whatever
you
want,
usually
depending
on
your
pump
requirements
but
limitations,
but
but
they
usually
come
at
the
same
as
the
same
color
yeah.
So
these
are
the
issues
that
makes
them
a
little
bit
useless
for
us.
There
are
ways
to
use
these
crystals
to
make
them
much
more
narrow
than
blind
bit.
This
is
a
paper.
This
is
a
photo
from
a
paper
in
in
2016
from
andrew
white's
group
in
australia.
C
You
can
use
a
cavity
to
enhance
this
systems.
You
can
do
these
optical
parametric
cavities,
which
pretty
much
is
forces
the
photons
to
come
out
within
the
line
width
of
the
cavity,
so
people
can
create
photons
that
are
megahertz
or
even
narrower,
but
the
downside
is
that
first
of
all,
your
rate
is
always
very,
very
limited,
and
these
devices
are
incredibly
complicated,
complex
in
terms
of
for
for
for
what
system
robustness
for
for
something
that's
field
deployable.
C
So
actually,
during
my
phd
in
our
research
lab
at
sunnybrook,
we
are
developing
these,
but
but
we
kind
of
a
little
bit
gave
up
on
on
making
making
them
a
commercial
product
that
can
interface
out
there.
The
method
that
we
like
better,
is
these
light
matter
sources.
The
good
thing
about
rubidium.
Is
that
not
only
you
can
use
these
atoms
for
for
for
a
surge
of
light?
You
can
also
use
them
for
generation
of
lights.
C
A
lot
of
this
initial
quantum
repeater
protocol,
blcz
protocols
and
stuff
like
that
they're
based
on
atomic
sources
anyway,
and-
and
you
can
just
create
these,
so
you
can
create
this
spontaneous
four-way
mixing
processes
which
are
pretty
much
atomic
spontaneous
parametric
down
conversion
processes
yeah.
So
I
can
use
these
techniques
to
create
pairs
of
photons
using
my
rubidium
atoms
like,
for
example,
the
one
that
I
have
here
from
a
jqi
paper
in
2010.
C
Is
it
uses
a
structure
like
this?
So
you
can.
I
can
pump
my
photons
using
two
different
pumps
to
an
excited
state.
Double
excited
states,
two
photon
reference,
exactly
the
states
and
then
this
decays
and
coherently
create
and
gives
me
two
photons.
Not
only
these
two
photons
are
entangled
to
each
other,
mainly
because
of
the
degeneracies
of
these
levels,
but
because
these
are
atoms,
I
can
tune
them
to
come
up
with
two
different
wavelengths
because
they
have
to
follow
the
atomic
transitions
yeah.
C
We
want
to
do
the
exact
opposite
of
these
people,
so
we
are
generating
the
photons
at
795,
but
but
also
the
land
that
is
much
much
more
narrower.
So
now,
I'm
talking
about
photons
that
are
sub
50
megahertz
and
in
the
range
of
like
15
acres
or
so
compared
to
10
terahertz
wide.
So
I'm
a
factor
of
10
000
times
right
now,
more
or
less
better,
with
interfacing
with
my
quantum
memories,
then
it
gets
to
the
efficiency
yeah
and
there
are
several
like
different
schemes,
I'm
not
going
to
go
through
this.
C
C
I'm
saying
these
are
around
like
18
kilohertz,
but
that's
a
very
high
number,
because
the
photons
are
very,
very
narrow,
so
this
18
kilohertz
makes
a
lot
of
more
use
for
us
compared
to
a
200
kilohertz
spot
on
that
gives
you,
terra
has
line,
beats
the
one
that
we
are
now
developing
at
tunic
and
and
it's
at
very,
very
earliest
stage
of
its
developments
connects
a
very
recent
lap
is
a
modified
version
of
what
I
just
showed
you,
but
with
a
little
bit
of
more
multiplexing
and
we
can
and
redesigning
of
the
atomic
layers
that
we're
gonna
use,
we
can
generate
entangled
photons
that
are
at
795
and
1324
nanometer
line
bits,
but
but
our
aim
is
to
create
a
source
that
can
have
somewhere
around
a
generation
rate
of
one
megahertz
or
so
so
we
really
can
boost
everything
by
multiplexing,
my
our
photo
and
generation
rates,
without
really
causing
like
a
lot
of
photon,
bunching
and
stuff
like
that.
C
So
these
devices-
hopefully
when
we
have
them,
which
will
be
the
preliminary
results
for
them-
they're
gonna,
come
out
in
in
summer
these
years,
and
hopefully
they
have
them
in
a
much
more
modular
version
and
next
year
in
q3
or
something
and
these
devices
can
can
operate
much
much
better
with
our
quantum
memories
and
are
much
more
suitable
for
integration
into
the
quantum
networks.
C
Now
I
think
I've
been
certainly
over
talking
so
I'll.
Just
finish
this
with
just
two
more
minutes
of
talking
that
we
not
only
we're
building
these
quantum
devices,
we
also
really
really
care
about
being
able
to
test
them
into
the
vtvd
in
infrastructure.
So
we
do
a
lot
of
partnerships
with
stony,
brook
university
and
brookhaven
national
lab
and
because
they
are
local
and
also
one
of
our
co-founders
is
my
ph.d
advisor.
Who
is
a
scientist
in
both
these
institutes
and
he's
still
working
on
these
networking
aspects
of
the
research?
C
And
so
so
we
do
a
lot
of
we
have
plans
and
currently
a
certain
level
of
field
testing
using
buried
fibers
that
are
already
between
dna
sony
group.
These
are
the
ones
we
have
access
to
them.
Right
now,
which
allows
us
to
to
integrate
our
devices
into
these
networks,
for
example,
some
one
of
the
things
he
just
recently
put.
C
So
we
are
slowly
getting
to
a
place
that
we
are
integrating
or
we
are
using
our
devices
in
different
field
tested
tests,
and
now
we
are
talking
with
different
networks
like
es
nets
and
and
fiber
providers
to
bring
the
fibers
all
the
way
to
where
we
are
at
connect.
So
we
can
create
this
long
island
wide
network
and
in
different
places
we
can
put
our
quantum
devices.
Okay,
I'm
just
gonna
wrap
this
up
by
thanking
our
team
here
at
connect.
C
We
are
a
very,
very
new
team,
so
we
are
expanding
our
team
as
as
I'm
speaking,
we
are
hiring
more
engineers
and
technicians.
If
you
know
anyone
send
us
our
way
and
of
course,
a
lot
of
things,
especially
the
last
two
slides
have
been
happened
by
much
much
larger
teams
of
scientists
at
stony
brook
and
brookhaven
national
lab
as
well.
Thank
you
all
so
much
for
for
your
time,
so
far
feel
free
to
ask
me
anything.
I'm
sorry.
This
meant
a
little
bit
over
time.
A
Thank
you
maddie.
I
guess
there
would
be
a
round
of
applause
if
we
were
in
an
actual
room.
Well,
thank
you
very
much.
There
have
been
some
questions
accumulating
in
the
chat,
so
I'll
just
go
in
the
order
that
they
have
been
asked,
but
not
answered
so
I'll.
Ask
those
that
have
been
answered
so
bruno.
You
have
a
question
about
end-to-end
network
performance.
Would
you
like
to
ask
it?
A
C
Oh,
that's,
that's
a
very
good
question
yeah.
So
at
the
end
I
went
a
little
bit
faster,
so
so
these
things
were
not
clear,
so
we
have
been
researching
this
quantum
memories
for
literally
like
well
since
my
phd.
So
we
started
the
research
on
that
in
2013
or
so
at
the
stony
brook
university,
and
then
we
moved
the
research
here
at
qnec
and
our
lab
is
quite
recent
new,
so
our
lab
became
operational
last
august.
C
So
a
lot
of
things
that
I
showed
about
the
quantum
memory
progress
has
happened
in
the
past
seven
six
months
or
so,
unfortunately,
because
we
are
very
new,
we
still
are
working
on
these,
so
so
the
entanglement
sources
are
much
more
literally,
are
setting
up
the
setup
and
right
now
to
to
achieve
them.
So
the
best
I
can
tell
you
is
the
ones
that
are
in
ed
and
is
building
in
his
lab
that
are
very
similar
to
this.
The
rate
of
the
sources
are
still
low.
C
There
are
around
like
five
to
ten
kilohertz
and
in
terms
of
the
best
publications
that
I
could
find
on
these
atomic
sources.
The
base
rate
was
18
kilohertz
that
2019
paper
that
they
follow
up
in
the
paper
in
2020
also,
so
these
rate
of
these
entanglement
sources
if
they
are
not
multiplex,
the
rates
are
not
that
that
impressive,
of
course
again,
the
narrow,
photons
blob
loss
also
more
useful
for
us,
but
the
rate
is
limited.
C
Our
hope
is
to
bring
them
to
megahertz,
but
because
we're
using
atoms
there's
usually
a
hard
limitation
at
some
point,
because
it
takes
the
atoms
a
while
to
to
pump
and
then
bring
things
back,
and
so
so
the
timelines
that
the
line
beats
of
the
atoms
are
always
going
to
be
the
fundamental
limitations
of
how
fast
these
sources
can
work.
So
our
entanglement
sources
are
unfortunately
right
now
very,
very
preliminary.
C
C
The
entanglement
source
nor
the
quantum
memories
it's
the
time,
if,
if
we
are
not
multiplex,
which
these
basic
first
versions
are
not
multiplex,
is
the
time
that
we
have
to
wait
for
a
successful
swapping
to
happen
and
the
time
that
it
takes
for
us
to
know
so
the
repeater
that
I
was
showing
what
is
limiting
it
the
most
and
that's
why
my
y-axis
was
starting
from
a
rate
of
10
to
the
power
of
negative
two.
So
even
the
rate
was
not
you're,
not
even
going
to
get
one
photon
per.
C
Second,
if
you
use
these
protocols
without
really
multiplexing
is
because
what
is
limiting
me
is
the
time
that
I
need
to
wait
for
my
photons
to
travel.
Let's
say
50
miles
or
50
kilometers
get
the
successful
swapping,
and
then
something
has
to
tell
me
back
again
50
kilometers
of
travel
that
that
successfully
happened.
So
I
can
release
the
photons
from
the
the
quantum
memory,
so
the
initial
rates
of
and
not
multiplex
networks
is
going
to
be
very,
very
low.
A
Thanks
a
lot
yeah
well
bruno
can
confirm
and
chad
or
not
patrick,
asked
a
question
but
mal
answered
in
the
chat.
So
for
the
interest
of
time
I'm
going
to
skip
that
one
there's
a
question
from
answering
the
questions.
I
really
appreciate
it:
hey
chung.
Would
you
like
to
ask
your
question?
Would
you
like
to
ask
a
question?
C
C
One
thing:
that's
going
to
make
them
a
little
bit
expensive
is
that
you
need
four
single
proton
counters
if
you
are
in
a
very,
very
basic
level
of
these
experiments
as
we
are,
for
example,
because
the
very
first
field
tests
are
going
to
be
having
very,
very
low
rates,
you
cannot
use
conventional
photo
encounters,
so
you
have
to
use
nanowires
and
stuff
like
this,
and
that
will
make
it
very
expensive
because
those
nanowire
detectors
are
like
under
each
range
of
fifty
two
hundred
thousand
dollars.
C
So
you're
talking
about
at
least
two
thousand
two
hundred
thousand
dollars
worth
of
detectors
for
for
those
and
the
the
device
itself
is
very,
very
simple
yeah,
it's
just
literally
it
splits
the
photons
in
for
pass.
So
that's
that's,
not
gonna
cause
damage,
so
the
device
itself
might
be
like
10k,
but
this
detectors
eventually,
of
course,
when
the
rate
increases,
we
can
get
to
a
point
that
we
can
use
more
conventional
spc
amps
that
are
more
in
the
range
of
5k
or
so
each.
C
So
so
it's
really
that's
going
to
really
define
how
expensive
they
are,
how
successful
they
are
are
defined
by
so
many
different
things
like
a
lot
of
the
recent
papers
that
are
coming
out
from
caltech,
fermi,
labs
or
other
researchers,
they
do
have
a
relatively
high
success
rate.
Of
course,
they
naturally
fail
fifty
percent
of
the
times,
because
the
the
real
estate
swapping
has
to
map
your
map
you
into
two
only
two
of
the
states
that
gets
mapped
out
of
the
four
are
the
ones
that
can
give
you
the
information.
C
C
So
I
love
the
ones
that
you
see
in
research,
labs
or
the
ones
that
they
use
their
own
fiber
pool
and
stuff
like
this
is
relatively
highly
efficient,
but
that's
mainly
because
it's
much
much
easier
to
time
everything
precisely
if
you're
in
a
lab
or
to
use
a
fiber-
that's
for
example,
polarization
maintaining.
So
you
don't
have
the
issues
with
these
things.
I
am
pretty
sure
the
more
we
do
these
networks
out
there.
This
the
swapping
rate,
is
not
gonna
that
easily
be
anything
better
than
10
percent
in
the
near
future.
C
A
Thanks
since
we're
kind
of
just
a
bit
beyond
time,
there's
one
more
question
from
keon
and
we'll
we'll
have
this
question
and
wrap
up.
I
would
just
before
I
ask
it.
I
would
just
like
to
remind
everybody
to
fill
in
the
blue
sheets
at
the
bottom
of
the
link
that
I
posted,
where
I'm
also
keeping
the
q
a
notes.
B
So
we
have
21
names
on
the
blue
sheets
and
we
peaked
at
about
70
people
here
in
the
in
the
room.
So.
A
D
No,
I
can,
I
can
go
for
it
as
well
yeah,
because
you
mentioned
entangled
photo
pair
generation
rates
and
I
was
at
first
I
was
wondering
how
much
filtering
you
would
have
to
do
after
or
if
that
number
was
already
included
in
those
rate
numbers
then
bruno
answered
that
or
sorry.
My
l
sorry
answered
that
these
rates
are
after
filtering
process.
C
D
C
Oh
okay,
yeah
yeah,
so
so
this
these
rates
are
always
going
to
be
very,
very
affected
by
several
things.
Honestly,
there
yeah.
So
that's
the
I
don't
want
to
go
like
50
slides
back,
but
this
is
recorded.
Fortunately,
it's
the
oh
actually.
You
can
skip
here
to
show
this
yeah,
so
the
expected
rates
are
very,
very
low.
C
So
if
we
started
like
10k
or
100k
or
so
the
expected
rate
after
one
successful
node,
which
is
like
100
in
140
kilometers
long,
the
rate
is
going
to
come
from
around
10,
starting
from
10k
to
one
photon.
Every
100
seconds,
yeah
10
to
the
power
of
negative
two
you're
gonna
lose
a
lot
because
you
still
have
a
lot
of
fiber
loss,
you're
still
using
like
70
kilometers
or
so
the
fiber
loss
itself
and
the
memory
loss
and
the
successful
swapping
loss
and
all
those
things.
C
So
your
rate
is
very,
very
low
at
the
beginning.
If
you
do
not
do
any
multiplexing
and
it's
a
lot
also
limits
you.
So
it's
not
about
how
fast
the
memory
is
or
how
fast
the
entanglement
generation
itself
is.
It's
all
about
how
long
it
takes
for
the
photos
to
arrive,
and
you
do
the
synchronization
and
make
sure
everything
happened
successfully,
so
without
multiplexing
or
or
a
very
proper
or
to
optimize
protocol.
The
original
rates
are
going
to
be
very,
very
low
after
you
swap.
C
Yeah
I
mean
yeah
yeah.
You
might
wonder
why
we
do
this
if
the
rates
are
gonna
be
so
low,
but
that's
pretty
much
how
the
very
initial
versions
of
the
current
internet
was
also
yeah.
All
that
matters
for
us
to
show
that
these
give
us
at
this
stage
right
now
that
these
devices
really
work
together
and
when
they
work
together,
they
can
beat
the
loss
in
the
fiber,
because
that's
pretty
much
the
rate.
You
also
expect,
no
matter
how
amazing
your
entanglement
source
is,
if
you
just
directly,
send
them
to
fibers
you're,
barely
gonna.
C
If
the
fiber
is
long,
you're
not
gonna,
get
anything
once
every
100
years
or
so
yeah,
and
so
all
we
care
about
is
to
show
that
these
things
can
work
together,
be
integrated
using
the
telecom
fibers
and
give
you
a
better
rate,
and
then
we
and
a
lot
of
hopefully
researchers
around
the
world
will
kick
kick
up.
Our
engineering
part
and
the
protocol
parts
to
make
them
significantly
more
optimized,
significantly
more
high
rate
and
multiplex.
So
we
can
bring
this
to
something
that
is
reasonably
meaningful
for
applications.
D
Yeah
sure
so
I
was
also
wondering
about
the
surrounding
hardware
that
you
have
to
build
to
get
this
to
work,
because
you
mentioned
you
need
you
know
all
these
all
this
extra
hardware,
also
in
terms
of
timing
and
in
terms
of
frequency,
locking,
I
guess
how
how
mature
do
you
see
this
and
and
if
and
at
what
level
is
it
and
how
much
effort
do
you
see
that
that
has
to
be
put
into
in
the
future.
C
Okay,
that's
a
very
good
question
so
so
far
for
the
stage
that
we
are,
we
are
developing
seven
devices.
Three
of
them
are
the
main
ones,
the
swapping
memory
and
source
of
four
side
devices
that
they
they
are
for
the
hard
ancillary
hardware
that
we're
using,
but
that
is
literally
the
stage
that
we
are
yeah.
So
our
understanding
of
networking
is
very,
very
limited.
C
That's
the
main
reason
we
want
to
do
this
field
test
because
the
more
we
do
this
field
test,
the
more
we
realize,
the
parts
that
we
are
missing
and
we
need
to
develop
so
these
timings
and
the
polarization
compensations
the
things
are
like
the
more
obvious
things
that
you
definitely
need
yeah.
So
that's
really
really
not
that
good.
Then
it
gets
to
classic
digital
communication
stuff.
They
are
relatively
mature.
So
our
plan
is
on
our
side
to
have
all
of
these
out.
C
Fortunately,
we
do
have
a
good
mixture
of
physicists
machine
learners
and
engineers
right
now
in
our
team,
so
we're
really
hoping
that
at
least
the
alpha
version
or
the
beta
version
of
these
devices
are
already
by
the
end
of
this
year.
We
do
have
the
the
locking
devices,
for
example,
already
the
very
first
version
of
the
polarization
stabilization
device
has
already
been
tested.
The
problem
we
did
is
that
it
does
the
job,
but
it's
all
the
matter
of
like
how
long
down
time
it
needs
like
compared
to
the
network.
C
So
right
now,
for
example,
it
works
with
like
a
ten
percent
downtime.
It
can
do
it,
but
but
we
want
something
that
is
hopefully
like
less
than
one
person
downtime
and
it's
much
much
faster
in
bringing
everything
back
to
to
the
nodes,
or
what's
this
name
the
qsync
right
now
that
the
synchronization
devices
right
now
works
in
in
a
nanosecond
scheme.
So
we
still
like
a
little
bit
different
from
from
this
game.
That
brings
us
to
sub
nanosecond.
A
All
right,
great
thanks,
maddie,
so
with
that,
let's
wrap
up.
Thank
you
very
much
mandy
for
this
talk.
It
was
great.
I
really
enjoyed
it
and
I
hope
that
all
the
participants
actually
learned
a
lot.
So
if
I
don't
know
if
maggie
and
mal
or
for
people
from
connector
or
other
prgmail.
B
A
B
A
Join
and
for
those
who
are
actually
managed
to
find
their
way
into
this
call
and
are
not
on
the
mailing
list.
I
also
encourage
you
to
join
the
qrg
mailing
list
with
that.
Okay,
thank
you
very
much
and
I
hope
to
see
you
on
the
mailing
ietf
meeting
awesome.