The Utility of Quantum Computing for Chemistry with Jamie Garcia

Download MP3
Speaker 2:

Welcome back to another episode of the podcast. This is the 3rd episode that I recorded when I was at Rensselaer Polytechnic Institute in Troy, New York. And in this episode, we had a conversation with Jamie Garcia from IBM Research. She is the technical program director for algorithms and scientific partnerships and, a quantum chemist by background and training. So it's a really interesting conversation.

Speaker 2:

It's another in our series of investigating sort of the near term applications for quantum computing, where there's an expectation that chemistry, in particular, is one of the areas where, there will be applications that have significant value and a and a quantum advantage in the near future. And Jamie talks about the exploration that IBM is doing right now with a lot of its research partners into what they call quantum utility, which is really the frontiers of what you can do scientifically with quantum computing simulations, small scale chemistry, ground state calculations, and some dynamics simulations as well. It's really interesting. We also talk about the fact that, the Rensselaer community now has an IBM System 1 on premises on on the in their data center, and what that may mean for the research and educational opportunities to the RPI community. So that's really interesting.

Speaker 2:

And I wanna thank again RPI for providing this great podcast recording studio, and for, giving me the opportunity to talk all these amazing people during the launch event. It was really enjoyable, and I think you'll enjoy this conversation with Jamie as well. We're joined today by a special guest, someone I've worked with before at IBM, somebody who's still at IBM Research, Jamie Garcia. Welcome, Jamie. Thanks for joining.

Speaker 3:

Thank you, Sebastian, for having me.

Speaker 2:

Very happy to have you here. We're at RPI. We're at the the launch event for the IBM Quantum computer that's being deployed on campus, first on any university campus in the world, which is super exciting.

Speaker 3:

So exciting. Yes.

Speaker 2:

And you actually had a huge hand in organizing the entire event, so congratulations.

Speaker 3:

Oh, thank you. I've been so excited to see how it's been going, like, truly.

Speaker 2:

Yeah. It's been terrific so far. We're about halfway through and it's really it's I'm super excited to see the ribbon cutting tomorrow. So so, yeah, why don't we start just with the basics if you wanna just introduce yourself really quickly and a little bit about how you got to where you are in Quantum?

Speaker 3:

Sure. Yeah. So I'm Jamie Garcia. I am the technical program director for algorithms and scientific partnerships group with IBM Quantum. My background is in chemistry, so I hold a PhD in chemistry where I focused a lot on catalysis.

Speaker 3:

And then now I've been working in quantum computing for about 8 years, which is crazy how time flies.

Speaker 4:

I know. It's crazy. I felt like

Speaker 3:

such a newbie in the beginning. But, yeah, like, I basically, through my chemistry research, worked with a lot of computational scientists as a part of that journey. And there were a couple moments during research that we were trying to look at a mechanism for something or understand the theory behind what we were observing in the lab, and we couldn't explain it at all. And then when we turned to the computational, chemists, like, we had to use a lot of approximations, right,

Speaker 4:

to

Speaker 3:

be able to understand the systems. So I think, like, a little bit of my, journey into quantum was based off of that experience because I I recognized at that point that there were some severe computational, limitations Right. As advanced and as the science is. Don't get me wrong. That needed to be addressed.

Speaker 3:

So, you know, the first time I saw a quantum poster on chemistry being presented, in Yorktown Heights lab, it was Abhinav's paper in 2017. And, that just like, I was like, oh my gosh. This is a new tool.

Speaker 2:

Right.

Speaker 3:

Like, we've gotta try this and just see how it goes.

Speaker 2:

Yeah. So you were with IBM Research already, but as a computational chemist or in that that kind of

Speaker 3:

I come from an experimental background

Speaker 2:

Okay. Actually.

Speaker 3:

Actually. So but worked very closely with computational scientists because oftentimes, we would make an observation in the lab Right. And we couldn't explain it and or we needed to know how to control what we were doing.

Speaker 2:

Right. Right.

Speaker 3:

So we

Speaker 2:

Always always good to do.

Speaker 3:

Always good to do.

Speaker 4:

Right? You know? Maybe not half the you know, explode that day. Anyway, so

Speaker 3:

we we leveraged it a lot. And I think, yeah, a lot of times, these processes, it would take months, years Right. Of back and forth between lab and, you know, talking to the theorists to try to figure that out and then get it to a point where you actually could control it. So, anyway, so that's my that's my background.

Speaker 2:

Cool. And, I I mean, that's that actually gives some framing for that article you wrote in Scientific American. I think it was 2019 or 2020, something like that. Yeah. I really like that because it it put the the it sort of framed quantum computing as a tool to bridge that theory to experimentalist kind of of, Yeah.

Speaker 2:

Space in chemistry. Right?

Speaker 3:

Yeah. I think for chemistry materials and the way that we leverage computation, like, that's sort of the holy grail is, like, reducing the amount of time that it takes for that iterative process. So from, you know, I just said months or years. I mean, sometimes it can be even longer, right, especially if you're looking at industrial processes where it has to be so precise. Right.

Speaker 3:

If you can reduce that down to really as small as possible and even, you know, the vision of it being, like, in a someday world that you would have calculations that were so accurate

Speaker 4:

that

Speaker 3:

you don't even need to go into the lab first. You can do it in silico. You can do it on a computer. Amazing. And you know that the answer is gonna be, like, if not perfect, pretty close.

Speaker 2:

Right.

Speaker 3:

And so then that that just kind of flips the way that we do chemistry on on its head. Yeah. So, yeah, that that's the idea. The other thing that I think is is really important to recognize about the impact on on chemistry too is that you have new opportunities, right, with we talk about utility to actually leverage, and this is the first time I've actually thought about it this way, the quantum computer as the experiment.

Speaker 4:

Mhmm.

Speaker 3:

So in other words, you do the theoretical, you know, tensor network calculations or whatever, and then run something on the quantum computer, and that's your experimental validation of the theory. Mhmm. So that's a

Speaker 4:

new Yeah.

Speaker 3:

Paradigm. Right. And I think that that can be extended out, you know, beyond condensed matter at some point. Right? We expect it to be applicable and to see utility

Speaker 2:

Right. Right.

Speaker 3:

Style, experiments and examples for chemistry materials.

Speaker 2:

And before I mean, I wanna get to where you're going right now, but I wanna just loop back for a second. Like, the I think people don't really understand like computational chemistry classically done. You said approximations, it means really heavily on approximations, like a lot of data is just thrown out so you can get an answer. Right? And you're just, you're trying to guess essentially what what data you can do without in your account.

Speaker 2:

Your computation is still get a usable result. Is that is that sort of right?

Speaker 3:

It's yeah. You're what you can kinda get away with. Yeah. Right?

Speaker 2:

Yeah.

Speaker 3:

So yeah. So anything so Schrodinger's equation, like, anything beyond h 2, you cannot exactly Right. Like, simulate.

Speaker 2:

Right.

Speaker 3:

You have to invoke approximations. Right. It gets too complex. And so that, I think, in and of itself, is a really good driver for a lot of the, you know, thinking about how you could potentially use quantum to approach some of these these similar problems for, like, ground state energies, excited state energies, properties, and things of that nature. But, yeah, like, if you look at something like density functional theory, which is, again, like, probably one of the most heavily utilized

Speaker 4:

Right.

Speaker 3:

You know, methods in computational chemistry. That evokes approximations where you essentially, you're having to

Speaker 4:

set up your electron in sort of a,

Speaker 3:

sort of sort of as a fuzzy approximation for it. And it works pretty well. But, again, I mean, it really does lend itself to the the type of process that I was mentioning earlier where it just it's a lot of back and forth, a lot of experimental validation in the Flask, not on the quantum computer yet, to to be able to prove these these out. And a lot of times, the first answer you get is just so very far off.

Speaker 2:

Right.

Speaker 4:

Yeah.

Speaker 2:

Right. Right. Okay. So so getting back to you used the word utility. It's a word that that IBM Quantum sort of adopted in the last year or so because, I guess, you guys feel like the machines you're producing and the algorithms are right at the edge of sort of doing things that are not classically possible.

Speaker 2:

Mhmm. So you're starting to in a way, it feels like it's starting to fulfill that promise of of, you know, that Feynman made it in Endicott of, like, you know, if you wanna simulate a quantum system, you need a quantum quantum device. And Mhmm. So you're starting to do quantum, or simulate quantum systems that are not easily simulatable classically.

Speaker 3:

Some of the response by the field was immediately to, you know, publish their own papers on the classical, you know, methods. Right? And I think that's something that was really interesting in that is, you know, and maybe you've seen this plot. But if you, like, overlay it with, you know, what we got on the the quantum experiments, the actual classical, methods deviate by around 20% from each other. And it's because you still there's, again, they're great methods.

Speaker 3:

They're very sophisticated. Right.

Speaker 4:

But at the

Speaker 3:

end of the day, there's still some sort of approximation that you have to evoke. Right. And so that is, I think, what we're really thinking about and and starting to expand. Okay. So if you go beyond the icing model, what are the other things that you could potentially look at

Speaker 2:

Right.

Speaker 3:

That have a similar construct? Mhmm. And it's always gonna be, like, the classical fields, you know, and classical approaches are always going to be advancing at the same time. Right? And so I think there's a lot to be learned just by doing the experiment.

Speaker 2:

Right.

Speaker 3:

Like, the fields can learn from each other.

Speaker 2:

The the first, the the experiment that you guys used as an example that the sort of use of the utility word, I was where I learned the phrase kicked icing bottle.

Speaker 4:

Yeah. Yeah.

Speaker 2:

I love that. I love that term of phrase. It's hilarious. Yep. But yeah.

Speaker 2:

So so okay. So so this this sort of, you're on the verge of sort of doing things that are not classically easy or or the classical, as you said, the classical result has variations, their their approximations and you're you're getting something that's higher fidelity essentially with the simulation. So coming back to chemistry, do you think, like, are we at the point where we're gonna start seeing this as a as a practical tool in the computational chemistry, experimental chemist sort of, you know, dialogue back and forth?

Speaker 3:

Yeah. I mean, certainly, I do. I think it's going to be a challenge, to think about different, examples that are going to be sort of the utility example, if you will, for chemistry. A lot of the problems are really large. Yeah.

Speaker 3:

You know? So chemistry, one of the the biggest challenges with it with when you start looking at the quantum circuits necessary for it is that there's a really large depth requirement.

Speaker 2:

Right.

Speaker 3:

And so, you know, you could have, like, a 5 cubit example and a depth of, you know, easily, like, 5 to 10 x Right. That number. Right. And so having to figure out, what to do with that, I think that's where things like circuit knitting, circuit cutting Yeah. Comes in.

Speaker 2:

And error mitigation, potentially. Right?

Speaker 3:

Oh, yes. Error mitigation, like, we use systematically, basically, in any of these Right. Examples.

Speaker 2:

Right.

Speaker 3:

So it's gonna be a combination probably of the techniques, and it's on you know, dependent on the problem. But, yes, I do foresee that it's gonna start becoming a tool. I mean, the most interesting innovations, I think, are what comes out of when you actually start looking at the hardware for things and trying things out.

Speaker 2:

Right.

Speaker 3:

Really understanding, what error mitigation needs to be used. We've we've seen for different problems that you need different and different you need to tweak the algorithm

Speaker 4:

Right.

Speaker 3:

But you also need to, like, change which error mitigation methods you need to use for for that call.

Speaker 2:

Gonna say, we interviewed Martin Savage from you Oh, yeah. Washington just a few weeks ago, and he told us about the experiment he did over over Christmas, basically. Yeah. And the nature of his problem allowed him to do some pretty impressive error mitigation that got, like, I mean, I think his his depth was something like 15,000 gates or something. It was like

Speaker 3:

Yeah.

Speaker 2:

It was a lot. Yeah. It was a 102 cubits or something involved. It's like really impressive.

Speaker 3:

Right. Yeah. And I think yeah. So those all of those methods are gonna be really again, and I think also to another important, piece here is that your intuition goes a long way. And so the people who are able to get, like, the best results on Quantum Hardware are folks that have been, you know, trying their hand at it for a while, and they get an intuition of, like, okay.

Speaker 3:

If I'm gonna look at this type of problem, these are the kinds of things that I need to do. And so your your starting point is a little bit closer.

Speaker 2:

Yeah. Yeah. And I think Martin has been working on IBM systems for quite a long time. I was surprised how long we've been doing it. So Yep.

Speaker 2:

Yeah. It's it's a very good argument for, industrial r and d scientists to get involved in this stuff as early as possible because

Speaker 3:

Oh, yeah.

Speaker 4:

Yeah.

Speaker 3:

For sure. And you talk you know, we we've talked a lot about the sort of continuum, if you will. And, like, one of the things that I think is, was a point that stuck with me when we were talking about our road map that we put out at the summit is that there's not gonna be a fundamental difference in the API between, like, the error mitigation devices available in the next, like, you know, 4 years to then when we switch over to now being in a in a regime where we have access to logical cubits. And so in principle, if you're learning now, you're not gonna necessarily notice the change because it's just what's going on in

Speaker 2:

the

Speaker 3:

hood that's changing. So even more impetus, I think, to start as early as possible.

Speaker 2:

Right. Yeah. It makes sense. I mean, the like Andrew started talking that last talk about magic state distillation and the the gates implementing gates on logical cubits is gonna be so complex that you're gonna need to abstract it for the end users so that the API doesn't really change at all. You're still you're just thinking in terms of the existing gate set.

Speaker 2:

And under the covers, it's doing all kinds of crazy stuff. But Yeah.

Speaker 3:

Yeah. Exactly. Yeah. And, like, you know, a lot of the people that will be the ultimate users of this are not, you know, experts in quantum. Right?

Speaker 3:

They're not quantum physicists.

Speaker 2:

Right.

Speaker 3:

There are folks that I think we heard someone earlier, you know, mention, like, that they they just wanna be able to use their program that they're familiar with. Right? And so, like, how can I, use, you know, whatever, you know, computational chemistry package of choice is, right, Gaussian or whatever, and then just have it, like, interface

Speaker 2:

Right?

Speaker 3:

With the quantum computer. Right. And even, you know, thinking about workloads and how you divvy up across, like, HPC and Quantum and where you introduce each of these things. Like, you don't necessarily, like, as a user, wanna be thinking about that. You just want that to be automatic.

Speaker 2:

Right. Right. Right. The the one thing that another thing Martin said, actually, which I've been thinking about ever since that conversation was that he was saying that he thinks that quantum information science, some amount of it is is becoming sort of a prerequisite for anybody in the physical sciences. Like, they Interesting.

Speaker 2:

You're gonna need to that's that's gonna be almost the assembly language of physical sciences in a way. It's like everything is gonna be relying on quantum computing in some way, shape, or form in that sort of toolkit kinda way that you described. And therefore, you're gonna have to have some kind of familiar of thinking of problems in that in those terms. Do you think that's true, or do you think we we do get to that abstraction where you're you're just a chemist and you're using Gaussian or whatever where you're Yeah. With and it's quantum under the cover?

Speaker 3:

It's it's a really good question because I think if you're thinking about it from, like, a university standpoint or an education standpoint, you're thinking about, like, okay. What skills do the future does the future workforce need? Right. You know? And I think that a lot of the richness of what we can do with quantum and what we can do in quantum computational science is at the intersections of different fields.

Speaker 3:

So I've kind of just, you know, from talking to folks, I've seen a couple of different approaches to that. Like, some have, like, full blown quantum programs that they're building out with

Speaker 4:

Right.

Speaker 3:

Core curriculum that is, you know, just based on quantum information science and

Speaker 2:

Right.

Speaker 3:

Beyond. And then some programs are sort of thinking about it more of, like, how do I have something that is, like, along the lines of a certification program where maybe you are a chemist, but then you have this additional, like, you know

Speaker 2:

Right.

Speaker 3:

Quantum certification on top of it.

Speaker 2:

Yeah. In fact, University of Washington has a quantum certification program Right. For 1. Yeah.

Speaker 3:

Right. Yeah. Exactly. And so how you design that is probably that's an interesting question. Obviously, I'm not in, in the university, so I can't answer that

Speaker 2:

or project that.

Speaker 4:

More

Speaker 2:

than one answer too.

Speaker 3:

I bet there is. Yeah. Like, not a one size fits all kind of thing because the, you know, you know, the students are gonna have their own kind of slants on that. But do I think, you know, I I think that what in these early days, especially, I think you're always gonna need some of that.

Speaker 2:

Mhmm.

Speaker 3:

I think that people, in the classical realm have have gone very you know, I I've known folks that are computational chemists that don't know how to code. Right? Right. And that's not like any it's just that they have the software.

Speaker 2:

Yeah. Right.

Speaker 3:

They don't need to know. Right. And so maybe, you know, if we just, like, project out

Speaker 5:

Right.

Speaker 2:

Right. Right.

Speaker 3:

Then that's maybe the future as well. But I think in the early days, yeah, you're gonna have to have some kind of understanding that covers a couple of

Speaker 4:

different bases.

Speaker 2:

Abstraction doesn't come overnight.

Speaker 3:

Yeah. Yeah. I like that. Abstraction doesn't come overnight. Yeah.

Speaker 3:

It's totally true.

Speaker 2:

Yeah. Yep. Yeah. That's interesting. And so like, you know, again, sort of coming back to RPI, there's now, IBM System 1 in the building just behind us there.

Speaker 2:

What do you think the university or a like, the academic world has to benefit from from an on premise device? What do you think they're they're gonna you know, what are you excited about seeing coming out of the RPI faculty and student body?

Speaker 3:

Yeah. Well, I think, like, one of the things that it definitely does is it inspires, the imagination of, like, what do you what can you do? What can you actually try with your own two hands? And having something, like, physically present, you know, there's a lot of motivation, I think, to do that, because it's you're you don't have to wait in a queue. Right.

Speaker 3:

Yeah. You just hop on. You can use it as long as you want. You know? Maybe not forever.

Speaker 2:

But the freedom are definitely Exactly. Not to be undervalued.

Speaker 3:

Exactly. You know, and there's sort of, like, some added additional, like, things that I think will be very interesting to see. You know, having it be colocated with an actual HPC, like, center

Speaker 2:

Yeah.

Speaker 3:

Is very will be very interesting.

Speaker 2:

Yeah.

Speaker 3:

So there's just, I think, a lot of different potential avenues to explore. The other thing is too, like, you know, it's not they're the 1st university to to have a a quantum, you know, computer on-site like this. And it's you know, I've been chatting with people, and I've heard a little bit of, like, anecdotal, like, stories from folks saying, like, oh, yeah. Like, we knew of, you know, the student who told us that they turned down other schools Wow. To come here.

Speaker 2:

It's a recruiting tool.

Speaker 3:

It's a huge recruiting tool. Yeah. Because, I mean, the students are really I think most universities, even if they don't have a quantum computer on-site, in fact, they won't yet anyway, have, like Is

Speaker 2:

there an 800 number that I can call Yeah. Order 1?

Speaker 3:

Yes. Just call us up.

Speaker 2:

1800 cubits.

Speaker 4:

Yeah. There you go.

Speaker 3:

But, yeah, I think that they yeah. It's funny. I think that they have, you know, student clubs, though, that Yeah. Like, a lot of them very large student clubs.

Speaker 2:

Yeah.

Speaker 3:

And so you you have to take that as an indicator. If I'm someone, you know, at a university that, like, there's a lot of interest, that there's going to be a lot of interest in using something and having access to something that they wouldn't otherwise.

Speaker 4:

Yeah. Yeah.

Speaker 3:

So I do think it's a it's a huge recruiting tool and also, will serve as an educational aid.

Speaker 2:

Yeah.

Speaker 3:

In those kind of courses that we're talking about or certification programs or whatever, because you can just you can literally use it how you want.

Speaker 2:

Yeah. Well, in particular, you mentioned the integration or colocation with an existing HPC Yeah. Center. Do you think I mean, that's really interesting and exciting to me because I wonder, you know, are there scenarios that you can imagine where, there may be some, you know, role of the quantum computer can play in otherwise a classical solution for chemistry simulation or something.

Speaker 4:

Mhmm. And

Speaker 2:

it's acting as an accelerator for one specific, you know, subcomponent of the overall solution that that you can do because you're connected to the same Yeah. InfiniBand or some kind of high bandwidth connect interconnect.

Speaker 3:

Right. Yeah. I think that's the idea. And then you, like, avoid any latency or anything like that, issues that would come up. So in you could potentially, like, explore that, a little bit, easier with higher fidelity, and and start looking at, like, pieces of the puzzle.

Speaker 3:

And, again, it goes back to that sort of, like, workflow and how do you divvy it up and, like, what do you wanna do where, and which of the different processes are are better to do on the quantum device versus, you know Right. On a GPU or a CPU or what have you.

Speaker 2:

I mean, going back to Martin again, he actually did a variational preprocessing step classically before he, like, actually loaded the circuit up on the on the machine. So Yep. It feels like sort of what you're saying, like, there isn't a one size fits all. We don't really know Right. What the right approach is for any particular any number of different configurations might be might be sort of opening new doors, so to speak.

Speaker 3:

Yep. I mean, in a lot of the chemistry research we've done, we've done sort of, like, to put on the quantum computer, and options are usually classical for that. So, like, you might use DFT, for example, to down select for the orbitals that you really that's really important for a chemistry problem, then you can try, you know, the chemistry part on the quantum computer, to get get the ground state energy or what have you. So I think, like, yeah, having having that capability, I think, will be really interesting, open up, new doors. And, you know, I you know, I think for many will serve as a starting point to you because they're, again, the barrier to entry is probably a bit lower.

Speaker 2:

Yeah. Absolutely. Yeah. You've done a ton of really interesting projects. So you just mentioned, you know, sort of in the the projects you've done for customers and and, just internally at IBM.

Speaker 2:

Are there any projects you've done in the past that you could you're sort of waiting to revisit? Like, is there a 127 or or, you know, like, is there something sort of this

Speaker 3:

Yeah.

Speaker 2:

As soon as we have this capability, I wanna go back and improve on that result.

Speaker 3:

Yeah. I think, like, you know, we've done some, like, small molecule examples of, like, certain I think, like, the the one that I really loved was the sulfonium, study that we did, which is like a that we did with JSR.

Speaker 2:

Right.

Speaker 3:

So it was basically trying to mimic a photo acid generator for photoresist. Right? So we did Sulfonium, which is just h three s plus. There was a lot of, like, algorithmic challenges that came with that because you have, like, a positively charged molecule. You're trying to look at a a dissociation where now you have radicals involved, potentially.

Speaker 3:

And so there was a a lot and you wanna look at, energy levels and excited states, which are a little bit more complex than just ground states to look at. Right?

Speaker 4:

So, like, all of those things

Speaker 3:

taken together made it you know, like, we had to figure all of this out, figure out the error mitigation method. So it was it was research project in and of itself. However, it would be even better if, right, we could actually get to the system that is of industrial interest, you know, to JSR.

Speaker 4:

So something where now you

Speaker 3:

don't have the protons, but you're looking at phenyl rings. So you have, like, you know, 6 carbon atoms and 5 hydrogens. It's a lot bigger.

Speaker 2:

Right. Okay.

Speaker 3:

There's different kind of chemistry because of that. In fact, the chemistry that's, you know, they're targeting usually has to do with the aromatic nature of those rings. And so, yes, let's get to that point. Right? Like, so how do you know, once we have these ways to divvy up the circuit

Speaker 2:

Yeah.

Speaker 3:

And really kind of break it down, maybe we can start tackling things of that size, which would be amazing. Yeah.

Speaker 2:

And You're talking about circuit knitting and and entanglement forging. Right? Sort of Yeah. Like needs to sort of decompose the problem?

Speaker 3:

Yeah. Like, entanglement forging is a really good example of that. And so, you know, how can you use that? How can you use embedding methods potentially? Those are all the kinds of tools that are in that toolkit, circuit knitting, embedding, and that type of thing.

Speaker 3:

And I think that, you know, again, it's not a one size fits all. You do have to do, like, some trial and error. Like, does it actually help? It might. It might not

Speaker 2:

Right. Right.

Speaker 3:

In certain examples too.

Speaker 2:

But I guess all of those experiments, like, that's where you're gonna extract those principles. You're gonna end up building abstractions for for the end user who doesn't have, Yeah. Incredible depth in QIS in order to use these things. Yeah.

Speaker 3:

You learn something every time. And so over time, like, as you build up a body of these, then you can make some generalities about what needs to be done and what kind of things are helpful to do in certain instances and then potentially automate that.

Speaker 2:

Right. Right. And, has Circonitin been used in in any, like, whatever, published research that you guys have done at this point? Or I

Speaker 4:

know it's

Speaker 2:

been experimentally done, but is it a real technique that you guys are using?

Speaker 3:

Yeah. Cool. So we, recently applied it. We did a a study with Boeing, and we used, you know, basically entanglement forging, techniques there. I would actually say in most of our current chemistry research, there's something Okay.

Speaker 3:

Like that.

Speaker 2:

Cool.

Speaker 3:

It doesn't work for everything. Again, it's that's my caveat. But, most of the time, we're we're starting to you know, when we start from ground 0, we'll we'll see if it if entanglement forging helps.

Speaker 2:

And sorry. I think I said circuit so you're saying mostly you're using entanglement forging. Right. And just describe at some level, like, how that works.

Speaker 3:

So entanglement forging, is basically it's based off of spin pair coupling. So, essentially, like, spin up, spin down, you can separate the 2 calculations out, so then you half the number of cubits that you need. Okay. So, essentially, like, if you had a 10 cubit problem, you could just use 5 qubits. And you run circuits, sort of, simultaneously and then use a post processing method to sort of knit the the you know, what she got out of

Speaker 2:

classical processing.

Speaker 4:

Back together.

Speaker 2:

Processing.

Speaker 3:

Classical pros processing. And so that's the trade off. I think Right. You've probably heard Jay mention it earlier in his talk today. Right?

Speaker 3:

Like, is that there is some overhead Yeah. Affiliated with that piece of it. Right. But it does allow you to do kind of larger systems or systems that would be out of reach, by kind of cleverly, you know, breaking down, the method. And, again, it doesn't work for everything.

Speaker 3:

You need to have, you know, weekly entangled spins. So, like, we used the first example of it was on the water molecule because it was like a poster child, if you will Yeah. For that. Yeah. And then, again, it's like you always when we start with chemistry examples, one of the first things we do is we evaluate, like, okay.

Speaker 3:

Is this something we could use for this? Like, does a molecule kinda fit the you know, check the boxes for what what you could use entanglement forging for?

Speaker 2:

Right.

Speaker 3:

So but, definitely, like, a very helpful technique and, I think, very useful for looking at, some of the more challenging problems that we see, like, in this really in

Speaker 2:

Well, I guess I mean, there's there's always going to be some amount of classical overhead, even, like, state preparation and Mhmm. Error correction is gonna introduce some kind of classical overhead. So and as you said, it's it's not always raw performance. It's also fidelity. Like, if you get a higher accuracy Yes.

Speaker 2:

Prediction, then that's gonna help you more than necessarily just getting, you know, something faster but still inaccurate.

Speaker 3:

Yeah. Well, that's like when I talk to, you know, chemists and computational chemists, that's, like, really, at the end of the day, what they care about is that, you know, accuracy level. So they'll tell you be the first ones to tell you that a lot of these approximate meth methods that we've been talking about get you very, very close to the exact value. And so there's a certain kind of trade off there. It's like if it's good enough, then why would you Right.

Speaker 3:

You know, address it as as a challenge? Right? So I think that it's, in some cases, much harder to get those accurate results.

Speaker 4:

Mhmm.

Speaker 3:

And that's where, like, if there is something that quantum brings there, they would be very excited

Speaker 4:

about.

Speaker 3:

So, like, transition metals is, like, the perfect example because they have the highly entangled d electrons.

Speaker 4:

Mhmm.

Speaker 3:

So that's a really good, I think, again, challenge

Speaker 4:

Right.

Speaker 3:

To start looking at with quantum.

Speaker 2:

And any guess as to what sort of quantum volume you need to be able to start looking at transition miles?

Speaker 3:

Oh, well, I mean, I think, the it's gonna come well, I don't know is the is the answer. That's the

Speaker 2:

quantum answer.

Speaker 3:

So the quantum answer. It's gonna depend on the development of all these different kind of tools and methods, as well as, you know, just having everything kind of the stage set for it. But I will say that, you know, a lot of the techniques that we've been developing out, I I think, you know, should apply

Speaker 4:

Right.

Speaker 3:

To those harder examples in chemistry that we just haven't been able to tackle yet.

Speaker 2:

Awesome. That's awesome. That's great. I mean, it it really does feel like, as you said, there's there's no one size fits all, and you're you know? But at the same time, you're developing this broad set of tools that are incrementally progressing the field and, like, you know, we are getting closer to this stuff actually being super, super useful.

Speaker 3:

Yeah. And and for things that you could never be able to do classically. Right. Ever. Yeah.

Speaker 3:

So Amazing. Anyway.

Speaker 2:

Excellent. Thank you so much for joining us.

Speaker 3:

Thank you, Sebastian.

Speaker 2:

Oh, my great

Speaker 4:

god. Conversation.

Speaker 3:

Yeah. It's awesome.

Speaker 2:

Excellent. Thank you. Alright.

Speaker 5:

Okay. That's it for this episode of The New Quantum Era, a podcast by Sebastian Hassinger and Kevin Roney. Our cool theme music was composed and played by Omar Costa Hamido. Production work is done by our wonderful team over at Podfly. If you are at all like us and enjoy this rich, deep, and interesting topic, please subscribe to our podcast on whichever platform you may stream from.

Speaker 5:

And even consider, if you like what you've heard today, reviewing us on iTunes and or mentioning us on your preferred social media platforms. We're just trying to get the word out on this fascinating topic and would really appreciate your help spreading the word and building community. Thank you so much for your time.

Creators and Guests

Sebastian Hassinger🌻
Host
Sebastian Hassinger🌻
Business development #QuantumComputing @AWScloud Opinions mine, he/him.
Jeannette (Jamie) Garcia
Guest
Jeannette (Jamie) Garcia
IBM Quantum. Quantum Computing, Chemistry, Polymers and Sustainability. Thoughts are my own.
Omar Costa Hamido
Composer
Omar Costa Hamido
OCH is a performer, composer, and technologist, working primarily in multimedia and improvisation. His current research is on quantum computing and music composition, telematics, and multimedia. He is passionate about emerging technology, cinema, teaching, and performing new works. He earned his PhD in Integrated Composition, Improvisation and Technology at University of California, Irvine with his research project Adventures in Quantumland (quantumland.art). He also earned his MA in Music Theory and Composition at ESMAE-IPP Portugal with his research on the relations between music and painting. In recent years, his work has been recognized with grants and awards from MSCA, Fulbright, Fundação para a Ciência e a Tecnologia, Medici, Beall Center for Art+Technology, and IBM.
The Utility of Quantum Computing for Chemistry with Jamie Garcia
Broadcast by