de mining is sign invention
together with my colleague in a sense of the quantum technologies in singapore i've been
thinking about particular approach to a building
quantum computers
with both there is so what we do is what task on paper and hope
it is interesting to the experiments as
but i love a paper is high threshold
distributed quantum computing with three two bit nodes just quite an awful i think it
by explain what that means all of done a pretty stable job of explaining
okay
so on some computers of course of a stream of modern physics the idea of
the machine that harnesses quantum states in order to do calculations there are
effectively impossible with ordinary technology
in order to build such machine we need to have a very large number of
components each of which is stored in quantum state so that be the basic components
is that you that we need lots of them
or under good control
be distributed quantum computing approach is kind of an architectural overall scheme have to do
this
which tries to make things a scalable as possible
so the idea is don't put only a few bits into a single grand or
a monolithic structure that is one week ago but
in this t v two q i p instead you try and build small units
which michael more tools pornos each of which has only a few bits inside it
you get to the building that's and controlling it and you make lots of them
and welcome to get to make the last machine the network you expect to the
and we see that work for network
and that's actual scheme for the entire machine it's distributed in the sense that it's
kind of exploded out from a single monolithic structure but of course not distributed over
large distances in one room
okay so that but thresholding
okay so a threshold in the u refers to
a level of precision that you need to reach in order that when you trying
to a large scale computation the errors don't get out of control instead the arrows
are coming in at low enough rate that you can detect and correct and for
that basically if you're
within your threshold you can do this outside of the threshold and errors will build
up so fast that the calculation goes off track and get you know
so of course you want a threshold the bs the mystic as possible as high
as possible
what you know like what's also this let's try and figure out what is the
threshold for this distribute a quantum computing approach we're gonna have to keep numbers how
it is the network okay just a controlled inside the node and we can also
only three cue but spend a week or about that case "'cause" we put it
would be the simplest case that would give us
a good threshold
three people nodes are reasonable thing to ask for lots of experimental groups can basically
do that can basically q three q s
so what we found was
that we a partial was
ten percent noise in the network which means one time and ten
when you try and communicative network it just on bananas the you across to keep
it simple
and no point one percent noise with in the nodes themselves that means
on those rare occasions when you try and the human or something it actually correct
that q so
ten percent of the network what one percent for the
local operations in each that
but as numbers are time especially the ninety nine point nine percent precision within each
node star
but not ridiculously so there are experimental groups in a syntactic units or in
all tracks lindy sentences were pushing past sort of ninety nine percent threshold so
that could be rich but also we stress that we have with trying to improve
this point back to schemes that have
even more optimistic specials
you know the same time as the experiments is trying to improve
their stuff and
we have that soon not and you know pretty soon
we may see these numbers actually meeting and then in principle you do we are
trying to
a large-scale machine
okay so a few remarks the experts a how do we get a high threshold
with such a small number keep a node we basically a and in something which
is pretty much most of the previous role as a done which is to distill
well as within each node
a high quality and use those two power actual once again between local units
your client of data humans we don't do that instead we currently project no easy
hardy projections on the core units we do that repeated the until the party projection
becomes
effectively your and it's the resulting to o can go from a while back but
it's basically can be done
so then we just use the art objections to make some useful what we may
is the three d cost the state of which is a resource for a topological
quantum computing in the start described by prosody
so that's okay and we find we only six party projections to do it which
is basically why we're able to guess i threshold
okay what if any if any of this it is of interest reading the papers
of interest please don't hesitate to our contact me
comments criticisms