00:00hi everyone welcome to the a 6nc podcast
00:02I am sonal and we're here today to talk
00:03more about quantum computing and for
00:05those of you that want more of like a
00:06primer on what it is and how it works
00:08definitely listen to our other podcast
00:10but you don't have to listen to that
00:11other podcast for this one the goal
00:13today is really talk about what it means
00:14to actually build something that's so
00:16cutting edge like that's the buzzword
00:17that we throw around so lightly and what
00:19we'd like to do in this podcast is
00:20actually really like break that down and
00:21joining us to have that conversation we
00:23have jeff cordova who's the head of
00:25software engineering at Righetti and
00:26then we also have a cnc general partner
00:28Vijay Pandey who's on the board of
00:30Akkadian has a long history actually in
00:32the world of high performance competing
00:33because you used to do fold at home you
00:35know there's been a long history of
00:37advances in computer architecture you
00:39know the computers that we learned as
00:41kids were very straightforward but then
00:43with high performance massively parallel
00:44machines like folding at home we
00:46couldn't just take our algorithms and
00:48convert it you'd have to really rethink
00:50the problem when you say highly parallel
00:53machines you literally mean like
00:54thousands and thousands of computers
00:56running in parallel next to each other
00:57or not necessarily physically close to
00:59each other I mean in fact in your case
01:00it was just tribute across other
01:01people's downtime on their laptops it
01:03was like steady yeah I think I said he
01:05came out basically 6 months before we
01:07did so this was first as October of 2004
01:10the projects now been running for almost
01:1120 years so instead of finding alien
01:12worlds you guys are focusing on protein
01:15folding yeah exactly and understanding
01:17especially the intersection of what
01:18compute could do in biology in that case
01:20they're doing calculations for
01:21understanding aspects of biology of
01:23protein and the interesting thing about
01:24that at least my recollection of it is
01:26that no one thought that algorithm would
01:29work it didn't look anything like the
01:30previous algorithms except that it was
01:33also doing some kind of chemistry that
01:34was interesting and then they deployed
01:36it and they got it working and they
01:38continued to make it work and now it's
01:39like the primary way that you can fold
01:42proteins in some ways they're right
01:43those impossible was impossible to take
01:45existing algorithms and just like shove
01:47it down to a very different architecture
01:49you basically had to rethink the problem
01:51and we went through this again when GPUs
01:53came out we actually were some of the
01:54first applications on GPUs even before
01:57programming languages exist on ji-hae
02:00GPUs we mean graphical processing units
02:02like the kind of Nvidia makes and other
02:03companies make that were originally used
02:05for the gaming industry but they're now
02:06being used widely deployed in machine
02:07learning exactly yeah GPUs have a great
02:10flowing point performance useful for
02:12again we have to rethink the algorithm
02:14now for massively parallel GPUs and I
02:17think what we're seeing now with cloud
02:18computing is yet again our rethinking of
02:20the problem like how we're gonna take
02:22something that's so powerful yet so
02:24different and and try to do something
02:26really grand with it one of the the key
02:28things to understand here is what
02:30quantum computing is and how it compares
02:31to classical computing Vijay was talking
02:34about in the early days of Vasiliy
02:36parallel machines there was kind of an
02:37expression amongst the engineers writing
02:39software for those which is it's like
02:41trying to herd chickens you're trying to
02:43get all of these independent processes
02:46to run and cooperate with each other to
02:48produce an answer and do so in a way
02:50that was faster than just running on a
02:51really fast CPU why does it matter by
02:54the way that it was faster than running
02:55on a really fast CPU but if you can get
02:57good enough results on the alternative
02:58why would you even bother because it
02:59because you can scale the problem up
03:01because in theory you can then add more
03:03processors and even scale up further and
03:05that was the whole promise of parallel
03:06computing which started you know several
03:08decades ago and really honestly with
03:10with systems like CUDA what's Gouda CUDA
03:13is the invidious language for doing
03:16parallel processing ok and there's no
03:18language yet like CUDA for GPUs to your
03:20point well there actually is quantum
03:22Universal instruction language I love
03:24that as quill I love the play on that
03:25word it kind of brings to mind pen in
03:27hand and you're doing stuff yeah I know
03:29the way that it is similar to the CUDA
03:31language for GPUs is it sews together
03:33the way you interface quantum computers
03:36with classical computers and without
03:38that it might be difficult to actually
03:40use near-term quantum computers and the
03:42reason for that is just because unlike
03:44classical computers which you can kind
03:45of run for days and weeks or in perhaps
03:47years at a time quantum computers kind
03:50of run in bursts of a hundred
03:51microseconds they're not quite stable
03:52yet like you don't have the full control
03:54but they're getting they're getting so
03:56we can run them for longer periods of
03:58time in that period of time you can do
03:59incredibly interesting and complex
04:01calculations that you actually can't be
04:03done at least theoretically on classical
04:06computers but you need a place to store
04:08the results and to interrogate the
04:10results and to do other kinds of
04:11classical post-processing on the data
04:13that you produce out of the quantum
04:14computer so there needs to be a way to
04:16interface the two that's where cool
04:17comes in is is hooking together the
04:19classical and in quantum machine we call
04:22that classical quantum hybrid computing
04:24this is now becoming a common pair
04:26at first there was just a CPU then CPU
04:28plus GPU CPU plus TPU right so central
04:32processor unit geographical processing
04:33unit - tensor processing units and
04:35processing unit each one of these things
04:37are specialized hardware for a
04:38particular task and that can do things
04:40that the other ones really just can't
04:41but you know it's interesting because
04:42you painted that as a continuum like CPU
04:44- cheap - GPU to GPU and there is some
04:46sort of continuum like effect but it
04:48feels like when you move into quantum
04:50computing that it's actually more
04:51discrete than continuous like you're
04:52actually doing something very
04:54qualitatively different yeah it's a
04:56completely different type of computation
04:58GPUs were intended for graphics but
05:00powerful for many things and now
05:01optimize a bit for machine learning GPUs
05:04have been designed from the ground up
05:06for machine learning and there's
05:07interesting pros and cons of each
05:08approach calling computing is actually
05:10different still we've done a little bit
05:11more for me about how I mean first off
05:14understanding the hardware is probably
05:16going to be important part of things I
05:17mean like to prob code a CPU okay no to
05:21code a GPU you actually have to
05:22understand memory access and things like
05:24that reasonably well to have a high
05:25performing a loretta
05:27fact that we have to rethink our
05:29algorithms is really nothing new the way
05:31that people would understand protein
05:33folding is that they would run one very
05:35long trajectory and then sort of watch
05:37the movie of what happened and all of
05:40this is inherently stochastic and
05:41statistical anyways and so once you
05:43realize what you really want to be doing
05:44is statistical inference you can do
05:46sucessful inference with many shorter
05:48trajectories anji standards can became a
05:51complete sort of different way to think
05:52about the problems and in the end in
05:54hindsight we really care about the the
05:56statistical model not the movie so many
05:58problems have been done because the way
05:59they didn't done because we've had only
06:01access to classical computers okay so
06:02but how does this apply then to quantum
06:04computing so you do have to think about
06:05the hardware but this is maybe on maybe
06:07a notch above that where you have to
06:09think about that nature the noise models
06:11and other aspects that are more unique
06:13to the hardware the appealing thing
06:14about con computing is that it allows
06:16you to take advantage of these mixed
06:18States to be able to take to the cue
06:20classical operations and do it in
06:22essentially in one step yeah one way
06:24that we think about it is that nature is
06:27inherently quantum mechanical you look
06:29outside and you see the light bouncing
06:31off the trees and that's a quantum
06:32mechanical process and it turns out it's
06:33very difficult to do lots of those
06:35simulations on a classical computer
06:37because a classical computer isn't
06:39quantum-mechanical thing it's
06:40deterministic boolean logic you can kind
06:42of think of quantum mechanics as
06:44probabilistic in nature so one of the
06:46rethink of the algorithms that you have
06:48to do is that we have to rethink how one
06:50constructs algorithms so that the
06:51outcome of them is in probabilistic
06:53sense what would the answer we're
06:55seeking no that's a very different way
06:57of thinking about about anything frankly
07:00more used to more cause and effect in
07:02our life and we see this macroscopic
07:04cause an effect that's deterministic it
07:06happens the same way all the time
07:07but these quantum processors actually
07:09give you different answers every time
07:10you run them and engineering the reality
07:12of quantum computing how does that play
07:14out in practice well what one thing one
07:16thing that you end up having to do when
07:17you write quantum algorithms is you have
07:19to run them a lot of times and then take
07:21statistics on the answer to find out but
07:23the answer that nature would give is and
07:25that's a very different way of thinking
07:26about computation it's it turns out in
07:29at least in computer science over the
07:30last couple of decades that these
07:31probabilistic algorithms and the like
07:34have become very important in in solving
07:37large-scale problems you can kind of
07:39sample a large enough space of answers
07:42to get a pretty good answer even though
07:44you haven't looked through the whole
07:45space and it turns out that quantum
07:47computers can actually search
07:48combinatorially a huge space for certain
07:51kinds of problems and actually find the
07:54real thing that nature would do and
07:56that's just a fascinating concept to
07:57think of what you could do with that
07:59that's just fascinating to me especially
08:00because I think about the history of
08:01just of Statistics and the whole science
08:03was built on this idea of having limited
08:06sample size and sample sets and then you
08:08kind of move to this world where it
08:09changed entire fields like when I think
08:11of the early days of natural language
08:12processing versus now where you have
08:13huge data sets to actually be able to
08:15learn on versus having to be
08:16parsimonious about your calculations and
08:18how you go about it and now what you're
08:20saying which i think is completely
08:20mind-boggling is you don't have to go at
08:22the sample set you can go to the reality
08:24of the actual population capital it's
08:26pretty neat yeah and so here's an
08:28interesting thing is that for a
08:29classical computer its power goes like 2
08:31to the N for a quantum computer goes
08:32like 2 to the Q which itself is 2 to the
08:35N and with the new technologies using
08:37silicon qubits the number of qubits
08:39follows Moore's law and so in this case
08:41the number the number of qubits is more
08:43akin to like the number of transistors
08:45and so here the size of the quantum
08:46computer would roughly double every year
08:48so it's like 2 to the 2 to the end and
08:51is going to catch a lot of people by
08:53surprise which is it what will happen is
08:55a qualm computer at first will seem like
08:56it won't be all that useful it'll be
08:59below the number of qubits that you need
09:00maybe you need a hundred qubits to solve
09:02the problem and the existing machine
09:04only has 64 and so a classic computer
09:06would easily transfer but then the next
09:08year the quantum computer has 128 qubits
09:10and suddenly it handily beats any
09:12classic computer that ever existed for
09:14that problem you know what that is it's
09:15actually just basically something that's
09:16extremely difficult for human beings to
09:18process mentally in our own computer
09:20sloshing around in our heads which is
09:21exponential thinking in general when
09:23this is hyper exponential which makes it
09:24even harder and I think what that means
09:26is that for different applications
09:28there'll be a different number of qubits
09:30that will be that boundary from where
09:32the classical machine loses and because
09:34you have Moore's law kicked in here it
09:37will just be this very sharp change
09:38people will think the quantum computer
09:40won't be useful and all of a sudden it
09:41will dominate so it's like the classic
09:42example where it happens very fast very
09:44suddenly like it's accelerating
09:46basically I would say the couple decades
09:48I've spent Selleck invalids I've seen
09:50I've seen the movie a couple times of
09:53and only a couple times because there
09:54isn't happened very often and one of the
09:56lessons I've learned from that is that
09:58it takes it takes a lot of people to
09:59build a new market it's it's not
10:01possible for one company to build a new
10:03market and you get innovations from all
10:05over the place and at some point you
10:07reach that accelerating point then and
10:08then the whole ecosystem benefits but
10:10there's some real fundamental building
10:11blocks when you say that you've seen
10:12that movie before but a few times there
10:14are some things that have to happen in
10:16order for that to become a reality
10:18and when you think about the history of
10:19computing and I actually think we should
10:21be careful about this too because we
10:22can't necessarily extrapolate from the
10:24history of classical computing but
10:25that's all we have to go on so that said
10:27how do we think it's gonna play out
10:28given what we observed before and where
10:30we're going next one thing to not forget
10:31is that people have been working on
10:33quantum computing for a couple decades
10:35and the kind of the remarkable thing is
10:37all of that work has reached a point
10:40where it's moved from research into
10:42engineering in terms of building the
10:43machine and that's why we believe that
10:45we can build anti BM and Microsoft and
10:48Google and the other players in this
10:51ecosystem can also believe that has to
10:53do with the fact that they use the
10:54technology that was been perfected in
10:56Silicon Valley in other places over many
10:58many decades so we know how to we know
11:00how to make these lots of them if we get
11:02it the first one working when you say we
11:04that no how it's a it's the know-how of
11:06semiconductor manufacturing the
11:08superconducting circuits we made are
11:10using standard semiconductor
11:12manufacturing technologies what's
11:14complicated about them is writing the
11:16software to figure out how to how to
11:17make them do what they're supposed to do
11:18and that that's there's a lot of really
11:21interesting physics and mathematics and
11:23computer science that goes into that
11:24yeah with that said though I think this
11:25reminds me of the early days of
11:27computing we're just getting your hands
11:29on the device is sort of getting a
11:32ticket to sort of something that is
11:34really a part of the future but most
11:36kids won't be able to have a qualm
11:38computer in their in their house they're
11:41probably a couple million dollars and
11:42like that like with cloud efforts or
11:44something where this could be so much
11:45more broadly available what do you mean
11:47by the economics won't necessarily have
11:49the Moore's law property of becoming
11:50cheaper necessarily it will it's just
11:52that these are more akin to maybe the
11:54early days of IBM mainframe and so there
11:56will be a few of them in the world at
11:57first but the differences in those early
12:00days IBM mainframes you know a kid in
12:03Asia or in the Midwest but not people
12:05have access to one no but what it's
12:07funny is that the analogy works in
12:08another way though that is similar which
12:10is that was the original cloud and the
12:11sense of physically located because
12:13frankly we don't really care what the
12:14cloud is located right and today we just
12:16care about sharing and time sharing that
12:18those resources and in those days people
12:20did actually have check-in and check-out
12:21sheets to go use the mainframe for
12:23whatever application there was but I
12:24think the point that you're making even
12:25more valuable is that when it does go
12:27into so many hands to that kid in Asia
12:29the kid in the Midwest somewhere else
12:31we are completely surprised by the
12:33applications people come up with because
12:35we the inventors have never been good at
12:37inventing predicting what their tools
12:40will lead to of what people can do when
12:42you put that ingenuity in people's hands
12:43yeah building cloud access into how you
12:46get at a quantum computer will quicken
12:48the pace at which the killer apps are
12:49found instead of there being one like
12:51there wasn't the early days of the PC
12:52and electronic spread we might see a
12:54half a dozen of them pop up all of a
12:56sudden I mean it just feels like very
12:57premature to be talking about cloud
12:59computing before quantum computing when
13:01you think about the history of classical
13:02computing and how long it actually took
13:04us to get went to an AWS like state so a
13:07where are we and be like what's your
13:09view on where we should go to get that
13:11that's really that's really an excellent
13:13question so it turns out that we
13:15actually have quantum computers today
13:18it's just that their software their
13:20software simulators we call em and
13:21wantin virtual machine you can run a
13:23quantum virtual machine in software up
13:25to about 30 or so qubits and what that
13:29means is that people can access and
13:32practice quantum programming on a
13:34quantum simulator before the hardware's
13:36here and get ready that's mind boggling
13:38I'm glad you brought that up because we
13:40need to hear that the number one
13:41question I usually get asked is well gee
13:43is quantum computing real or science
13:45fiction when's it going to be here and
13:47the answer is it's here now but it's in
13:48the form of software and that will
13:50eventually be surpassed with a real
13:52piece of hardware but you can actually
13:53do a real quantum programming on the
13:54software that's important because I
13:56think that wasn't true in the previous
13:58world of computing right and we had no
14:00cloud we couldn't write access to a
14:02piece of software to everybody providing
14:03cloud access to quantum computing in the
14:06quantum simulator in the early days has
14:07kind of two main values and and that
14:09falls into two camps there's customers
14:11who are interested in how do you use
14:13quantum computing or quantum algorithms
14:15to solve a problem that I have and then
14:16on the other side we have this entire
14:18community of enthusiasts the people who
14:20are going to find those killer apps
14:21I would actually even maybe very
14:23simplified simply simplify it to people
14:26who have needs and people who have once
14:27and essentially just kind of getting
14:29right in the middle of that yeah and
14:30having a centralized place to both to
14:33express the need and put the solutions
14:35there for those as well the big point
14:37you were making is that we're trying to
14:39get to this place where we can use cloud
14:41deployments including cloud simulations
14:43as a way to get there what happens next
14:46after that happens I think I think that
14:48once the cloud is cloud access to a
14:50quantum computation that's available
14:53then we can trying to reinvent these
14:55algorithms as quantum programs so an
14:57example that a lot of people are
15:00thinking about is how do you make
15:01machine learning go faster machine
15:03learning has an inner loop or an inner
15:04part of the algorithm that's an
15:06optimization step an optimization step
15:08has to literally look through all
15:09combinations of things to find the right
15:11answer and so the cloud will help
15:13facilitate the discovery of such
15:15algorithms and then secondly it'll help
15:17you couple to the classical computer so
15:19that you can trade-off between the
15:21classical and where you can you can use
15:23cloud computing resources for your HPC
15:26or high-performance computing and you
15:28can couple up to a quantum computer so
15:29ok so what I'm basically hearing is
15:31cloud while in classical computing took
15:32a while to get there cloud is actually
15:34now sort of like a pseudo infrastructure
15:37it's sort of a way basically for us to
15:38stitch together the reality where we
15:40want to be yeah it's it's a very natural
15:42delivery mechanism the natural delivery
15:43mechanism so why does that matter so
15:45that's because that's actually more
15:46about the consumption model and more
15:47about the go-to-market I mean SAS is
15:50really unique capabilities in terms of
15:53the fact that you ship one thing you
15:55don't have to support all these on-prem
15:56and and that's a whole separate sort of
15:59discussion I think here cloud alone is
16:01just very much empowerment and the
16:03ability to get this in many people's
16:05hands but all those other aspects will
16:07layer on top once we get to the point
16:09that there are these killer algorithms
16:10so beyond the delivery there's also this
16:12component though of stitching together
16:13this world of classic and quantum
16:15computing because it seems like the only
16:16way you'd be able to do that is by
16:18having cloud as a connective tissue
16:20which in those two worlds well it's not
16:21the only thing you need software and in
16:23particular you need some way of handing
16:26off the computation from the quantum
16:28piece to the classical piece and well
16:30can facilitate this transfer of
16:32information from the quantum computer to
16:34the classical computer and back like you
16:36can't actually store data on these
16:38quantum computers they're just literally
16:39at right now just compute engines just
16:41the amazing compute engine almost like
16:42Urkel process almost like a coprocessor
16:44right so you have to have the data
16:45someplace else and so you need another
16:48place for the quantum computer to
16:50interface with the gift that data into
16:52store it's I think it's also important
16:55to understand that the other reason for
16:56clauses in days gone by we would build
16:58these complicated machines and we would
17:00install them on customer sites and
17:02premise well a quantum computer requires
17:04a cryogenic cooling system okay and it
17:06requires a special thermal and vibration
17:10stabilized platforms the quantum
17:11processors like the size of a quarter
17:13the rest of its the size of two or three
17:14refrigerators to house it and keep it
17:16cool at like a barely above zero degrees
17:19Kelvin way more complicated all that
17:24you lose cooling in the refrigerator
17:25maybe you damage the whole thing right
17:27so you never really at least in the
17:29early days want to ever put those things
17:30on a customer site you want to put them
17:32in a secure facility someplace and
17:33provide cloud access or remote access
17:35you know so that's the other reason that
17:37what's the right delivery mechanism for
17:39this technology right when I should go
17:41back to this idea of the hybrid it's
17:42just so fascinating to me because one
17:43debate that plays out when you think of
17:45adopting next platform you had these
17:47early adopters who are risk takers are
17:48gonna get ahead they're gonna try to
17:50adopt the new thing before it so they
17:51get a competitive edge and then you have
17:53people who are shy laggers and they sort
17:54of follow after everyone else has done
17:55it and yet people in the middle which I
17:57suspect as a reality of the fortune 500
17:59global 2000 where they're really want to
18:01get ahead but there's this tension
18:03between adopting something new and
18:04sticking to what you know in the old and
18:06one of the things I think is fascinating
18:08is that people a lot of people have done
18:11hybrid cloud in classical computing as a
18:14way to sort of straddle both worlds and
18:16maybe not the ideal because you actually
18:18want to leapfrog and go to cloud versus
18:20doing this intermediary step however in
18:23the case of quantum computing it's a
18:25necessity it's the only way to currently
18:27get there is what I'm hearing you say
18:29it's it's it's the it's a it's the right
18:32path to the first ones just because if
18:34we go back to the complexity of
18:36operating these systems that are
18:38essentially physics experiments and I
18:40can the customers want the power they
18:42don't want the hassle of having to
18:43manage and operate these things now we
18:46have this beautiful thing called cloud
18:48computing and cloud access and so we
18:49actually understand how to build that
18:51infrastructure and host those systems
18:52and provide the right API calls and so
18:54on even beyond just cloud there's aspect
18:57of micro services that plays in that is
18:59very naturally because you can imagine
19:00quantum micro services that do a variety
19:02of things and that in a day where you
19:04have different servers doing different
19:05things you're just doing API calls this
19:07would be just another server doing and
19:08that unique type of API call yeah and I
19:10think it's actually interesting cuz we
19:11talked briefly about VMs and virtual
19:13machines earlier and it's it's a next
19:14phase breaking things down to that sort
19:17of micro level yeah but the other thing
19:18that I think is really fascinating is
19:19how that plays out organizationally
19:21because then you have software
19:22developers and and product managers who
19:25are essentially running their own little
19:26business units many business units for
19:28owning their own projects soup-to-nuts
19:29right because it sort of self-contained
19:31in these little content and they'll
19:32bring in things as needed well I love
19:34hearing that to you because it's
19:35actually a way for big companies to
19:36actually embrace these experiments
19:38without even having to know what's
19:39served inside the box yeah exactly
19:41because you don't have to you really
19:42don't have to that's out of the whole
19:43point of the whole cloud in the first
19:45place right there's kind of another way
19:46to think about cloud 2 which is this
19:47vertically integrated stack of
19:49Technology so at the very bottom you
19:51have you know the compute power itself
19:53and the operating systems and then I'll
19:55tear up from that you've got some
19:57intermediate programming layer
19:58and at the very top you have API
20:00services and micro services and things
20:02like that when there's gonna be a bunch
20:03of other hardware that's classical in a
20:05in a cloud like you know your network
20:07stack and your storage stack and your
20:09and your other classical compute stack
20:11on the software side though it's
20:12literally a soup to nuts ground-up
20:14effort where you have to build the
20:16operating system to run the quantum
20:18processor then up above that you need to
20:20actually build the quantum algorithms
20:22and quantum programming language so that
20:24you can program the quantum a group so
20:26once it's all in one place you can kind
20:28of provide independent access to those
20:29different tiers depending on who wants
20:31to do stuff you can basically interact
20:33with that tower of software in different
20:35ways depending on what the level of
20:36granularity you want one question I have
20:38is when I think of a vertical stack
20:40seems like that's a problem it's too big
20:42for a single startup to tackle like it's
20:44something that a Google can do and IBM
20:46can do why would I start up be able to
20:48do this vertical stacks are very
20:50difficult to build I so I agree with you
20:52about that but there's a couple of
20:54different things that make them
20:55difficult to build one is just the rate
20:57at which you can iterate on the
20:59different components to see what the
21:00vertical ization looks like you might
21:01guess incorrectly and usually do guess
21:03incorrectly then yeah these layers
21:05actually are next to each other but with
21:07rapid iteration you can find out exactly
21:09what the layers are and the closer they
21:11are the more you can iterate yeah this
21:12just the discovery of what those layers
21:14are and what the right layers are is an
21:16extremely important problem and I think
21:17a startup has a massive advantage or
21:19just because startups are essentially
21:21engines for agile engineering themselves
21:23our optimization boom I think it's it's
21:26maybe even more essential for building a
21:28quantum computer because there's so many
21:29different pieces that need to be quickly
21:31iterated on to figure out how they fit
21:33together and this is the difference
21:35between engineering and science the hard
21:37science problems have been solved but
21:38how do you build a superconducting qubit
21:40now do you send it a radio frequency
21:42data to program it the hard part now is
21:45is going through all the different ways
21:47that these things can be hooked together
21:48to build a reliable machine there's an
21:51engineering problem now but it does
21:52require like rapid iteration and fast
21:54agile engineering teams in order to
21:56solve that problem okay so for people
21:58who are developers or engineers thinking
22:00about getting into quantum computing
22:02what does that mean for the people who
22:03are actually working on this stuff and
22:04who are trying to adopt this stuff
22:06wherever they are in that you know cycle
22:08of wanting to get ahead or catch up
22:10well I'll use an analogy from there
22:12of Intel when they invented the
22:13microprocessor there was no there was no
22:15engineer that was called a
22:16microprocessor engineer it was a
22:18combination of electrical engineering
22:19and fabrication technology and maybe
22:22even some computer science and so
22:23intimated investment to build a whole
22:27team of microprocessor engineers to
22:29figure out how to build the first
22:30microprocessor and actually build the
22:32successive generations and why does that
22:33matter it matters because the first
22:37thing you build isn't the thing that is
22:39usually the thing that dominates the
22:40market you have to do many iterations of
22:42it to finally get there you need a
22:43combination of skills in order to be the
22:46Jew the quantum engineer that's gonna
22:47build these machines you know the four
22:49thousand four microprocessor the very
22:51first microprocessor was built as I
22:53understand it for a calculator but it
22:55had that early application which could
22:57get to market and I think WOM chemistry
22:59could be that application that's
23:00something now that one can do but
23:02there's limits in the accuracy due to
23:05the expense and time so the four n atoms
23:09of the most expensive algorithms scale
23:11like n factorial now there are more
23:13efficient algorithms that go like and Q
23:16door n to the sixth but those types of
23:18algorithms are often not sufficiently
23:20accurate to go after areas where real
23:22chemistry happens where bonds break or
23:24you look at excited state electrons
23:26we're talking about nature is going to
23:27computing it's like what nature really
23:28does yeah and especially a lot of
23:30interesting applications from a
23:31commercial point of view are enzymes all
23:34the chemistry you know so so this is
23:36something where a quantum computer could
23:37be able to do calculations that are
23:39either much bigger and much higher
23:41accuracy or where I could would be
23:43essentially much bigger at higher as
23:44much add much higher accuracy and and
23:47and with a state of a machine that
23:49doesn't have to be this hundred thousand
23:51Cupid machine to do something
23:52interesting it could be much smaller and
23:53isn't there also actually stuff you
23:55cannot even do in classical computers
23:56today with for computational chemistry
23:58it depends on the number of atoms so
24:01like you can do this full calculation
24:03for like tens maybe a hundred atoms but
24:06you probably couldn't it'd be very
24:07difficult I think I do it for a thousand
24:08atoms or 1000 chemistry is everywhere
24:11and that's you think about energy when
24:15you drive a car it's going through
24:16chemical reactions when you your
24:18proteins in your body are working those
24:19are chemical reactions I mean your
24:22plants are going and using fertilizer
24:24and all this stuff it's all chemistry
24:25there could be one that gets into market
24:27and that we start seeing use and uptake
24:29and that's where it starts getting
24:30interesting because once there is an
24:32application and these things become
24:34cheaper and more ubiquitous I think
24:36we're going to see the second or third
24:37or the fourth or fifth and it's going to
24:38roll from there one of the surprising
24:40aspects of quantum computing is that
24:43scientists and mathematicians and
24:46computer and computer engineers don't
24:48really know all of the problems that can
24:51be solved by a quantum computer we just
24:53know some of them and that's not how
24:55classical computing work it was always
24:57the case that if you could solve a
24:58classical computer problem a classical a
25:00small classical computer then it would
25:01just automatically work better on a big
25:03classical computer but we don't actually
25:05know that class of problems that can be
25:07solved on quantum computers yet and that
25:08itself is a pretty interesting mystery
25:10well that's great thank you guys for
25:12joining the a 6nz podcast thank you