
Video

Table of contents

Video
 Description
 Transcript
 Discussion
About the talk
Brian Neyenhuis, Senior R&D Manager of Honeywell Quantum Solutions, Wojciech Burkott, Cofounder/Chief Physics of BEIT, and Jules Tilly, Research Scientist atRahko, present to attendees on December 8, 2020 at Q2B20  Practical Quantum Computing, an annual conference hosted by QC Ware.
Recorded sessions and conference details can be found at: https://q2b.qcware.com/
More information on QC Ware can be found at: https://qcware.com/
Get in touch with the QC Ware team at: https://qcware.com/contact
To learn about how to get started with Forge, QC Ware's data science platform running on quantum computers, visit https://forge.qcware.com/
About speakers
I'm excited to be here and talk about something exciting demonstrations that have been done on on the Honeywell trapped. I am, computer today. I'm joined by state and jewels Chili's from Rocco and we'll be talking about some of the different demonstrations that have been run on the honeymoon machine. So the hangar, computers based on trap stomach, ions is stored into hyper find State Theater, be in 171 and Zion and these cubits are these ions or levitate above a
microfiber cated trapped that's manufactured in design, inhouse in Honeywell, so we can rearrange the apps that well and was called that you see, see the architecture and having mobile Cubix that can be rearranged or several very nice advantages. The first is that we naturally get all talk conductivity if you want to do a gate between a cubit on one end of the chain. Rather than having to do a series of information from one side to the other, whenever Jade is done between two, two bits. Actually. Those are the only two
cubits in in the well, where the gate is done. And so this means that our gates are also very scalable. If you had more and more Cubist, the device, the interaction that gives us start YouTube at gate remains exactly the same. It's always just those two cubits in a single. Well, while we do our to Cuba State, even as we have, the means that we have sort of a native ability to get circuit Cubit measurement, and chew it reuse. We want to measure a single Cubit. We can just move all the other Cupids outoftheway measure, that Cubit reinitialize, it and continue on with the
circuit. Without destroying the quantum information, stored in the other cubits on the device. And we can actually use the results that measurements do a conditional, Gates later on. In circuit is absolutely essential elements. If you're going to do any sort of Connemara, correction, so everyone needs to develop eventually about something that we started natively have in this qcc d'architecture where we can move are cubits around. We're going to hear about the several examples and that use all of these features and the first little bit of time talking primarily about a cubit measurement and
reuse right now. Also to start, I'll give an example with the Bernstein visor on a house over there. Some Quantum circuit shown here on the screen. We're marking is Oracle in this case is 0 1, 1 0 1. But I found this in a very arbitrary way, if we look at just the very top, she was listed there in the circuit do realize that although we initialize the stupid and did, its first single to get rotation, severe begin, the circuit that was entirely unnecessary, that could have
waited all the way until right before the soup, with the Ancilla, which is our bottoms. You can look at the second cubed in the circuit. You'll notice it because it's marking a zero on that unit. It doesn't do any to Cuba gates at all. It's just the initial eyes has two single Cubit, rotations, men is measured that we can perform that really at any time. During the entire circuit in this manner. You can see that I've written this in a very suggestive way. That shows it actually in the sprinting does Ronnie algorithm. Each one of these Cubans acts very independently, but only ever
interact with the end. So I can stretch the circuit. I like this. And you can see that if I can measure a cubit and then reinitialize it. I can condense the circuit down to run just on suits. You it, until we've done this, we run this on the Honeywell system. Heromachine, how about the standard version using six cubits available and the stupid version and we see we get the same, we get the same out of it because we can do now any arbitrary lines just to show off a little bit. We went all the way out to a hundred bit Oracle here. I thought I would normally take a hundred once
you with the run but we've done it. Just the two because of cubic measurements and reuse. Now, it's the person visor on the algorithm in this manner. Gives you a little bit of heartburn. I don't blame you. It's a Nora Cloud for them and it's really not clear what's going on inside that black box of the Oracle when I'm doing repeated measurements initial ization or anything, but I think it's a very nice way to show how many Circuit one cubit is done being used before, another Cube has even started. If you have mistook a measurement and reuse, there's a lot of
algorithms. You can run with us. She would send you might not have the option where you smoke Tampa. Let's look at the quantum matter. Adding two different numbers. In this case has the number three in the number one. We're on this stops unregister. I program in the number to a Quantum Fourier transform and then do a series of controlled rotations based on the info. In this case has the number one, but of course before I'm doing that Quantum Fourier, transform in Reading at my answer I could add in other numbers. If I still wanted to share. I'm at in the number one in repeatedly that
circuit that this takes more and more Cube. It says you have more than one but just like, in this prancing Paws Ronnie, I'll go to my showed you. You'll notice that that first input register sort of done doing all this job before the second and put register starts to do anything interesting in all so, stupid measurement reissue. Can condense this down to run on just the six cubits that were available at the time for Honeywell System. Model. H0a is our friends at Accenture helpdesk run this algorithm and they showed that by raising these two. But you can avoid having multiple inputs
registers, but they also showed that you could, if you were to run this algorithm in a Where you at? The two numbers for a Transformer? Use that answer is the input. You could run an arbitrary number of additions will just be secure but a require many extra to cubic as your undoing and redoing this pain for a chance from over and over again. Here's here's a different samples from our friends and in tropical ABS related, some Quantum machine learning and machine. Are you have some circuits with some variable parameters to
give you some out from you? Compare that outcome, some classical distribution that you want and then optimize those parameters in your circuits to have the machine. Learn how the distribution that you're after. In this case. There were looking at this threebythree grid for Bars & Stripes. The temperature looking to identify all the states or sell, vertical or horizontal line, represent the machine learning a measuring, some of the four layers of entanglement, in the circuit shown below. They are able to do this machine learning problem. And on the right is the data I run on our
system. So, and orange, you have the simulation This is what you get. If you had a perfect on a computer where you get 60% overlap with the desired States and the flu is the actual data for your performance of machine and perhaps in the onset used to get higher Fidelity on the simulation Rising. This is a very appealing demonstration. It does more than just moving around our boxes on our lines. Looking forward to showing that you can do the machine learning algorithm with only having part of the entanglement travel from one half the circuit to the other. I'll
give you one. Last example from Justin Erica and eats the bocce at University of Texas and Los Alamos National Labs respectively, where they showed how to do Cubit efficient entanglement spectroscopy. So, I suggest we usually what you're trying to do. If your measuring the trace of powers of your reduce density Matrix, and one way to do this, with what's called a to copy test for you, make two copies of your Quantum State, you entangled to get that additional information. So that equals to give me four copies of your Quantum State as any + 3, you would now need six copies of your phone.
Since I was just staring at this diagram that Sumptuous you're done with your job before others even start. That is indeed the case and they were able to show that with just six cubits. You can go to arbitrary. Honda circuit to get all the information, you want about your entanglement trip spectroscopy. There's three different lines, for three different values of fade us. A different amounts of entanglement to the to Cuba State, their probing here, and you see that our experimental data it with the dash line. They're the ideal and the
experimental effect. They're all to go out all the way to any. That's an equal 7 with typically, take 28 cubits to do, but you're just on the six cubits are able to measure this. Entanglement. I don't have any time to go into it. But there is a talk on Thursday, for my phosphate. Also, from Honeywell showing how you can also stimulate large Quantum systems, using small numbers of pi again using the truck and measurement and she would really want to turn some time over to voice text from Paige, is going to be talked about some of these currents.
Thank you very much Bryan. And also thank you very much for the opportunity to test our ideas, especially around the metric of measurement in the actual algorithm and implementations on the real Hardwood. So we were obsessed with a structured search algorithm as it seems to be the easiest one showing in the larger number of Cribbage actual Advantage. So we try to think about how to build a deficient play band. Also has to demonstrate that in the actual. We started with the previous implementations of
superconducting cables machines and up until recently, the best you could do to show any advantage over the classical solution was a all of Reykjavik Reykjavik Church, eight states. And you are about to find this one state to the probability better than what we could achieve a classically. Show me the machine, but it started trying to Benchmark against the superconducting qubits worry, I will results for for could be close to the classical number. What's the probability of success? For unstructured
implemented? In Honeywell allowed us to the tremendous progress, which allowed us to make sure that we are actually getting some result of measurement and spoken activity, allowed us to reduce the number of cxk, the basic pivot gate when implementing larger. Doctor games necessary, for this complication measurement, allowed us to have an implementation of the CTX gate. I wish an average year was just hoping to see Xscape again. Thanks and the reliability of a reset. This also, an important
forest and actually Brian mentioned the conditional execution of the circuit, depending on the measurement result. Yes. We had bath with fried fish, any lecture 3 results. As I said before, there are, we did experiments on the superconducting to be, so we managed to get a photo image. Search almost as good as the classical War. When is 45% of the results and 3 cubits, which was done before and we repeated those results. Here are the results. As you can see, we run
different algorithms. You can judge this by the different shapes of the wire frames, depicting the theoretical values and after repeated rounds of many series of runs without the black thoughts and for the expected result from which this time span between the best and worst of our results and middle picture shows the measurements for a used to enhance the results summarizing you get something which is truly remarkable. The five different limitations of the instructor search for five cubits, as you can see, all those algorithms better than the classical counterpart
performing, they have tracked, you search having some advantage. Classical instrumentation. 60 bit strange that you can imagine. We had the 800 so that you could do but with those results, and first thing that you can actually have even that big a search. So you can notice there is some glitch at the top and we just different. I just want to buy it and that was something which we were able to rectify. We're thankful Honeywell. We have two sessions and we rectified the glitch
of the first form of providing almost realtime feedback on vehicle measurable results and the second round of weeks so that I can assure you that especially thanks for the Smile Sessions for many years as necessary for any error correction, to also helping Getting stuff more efficient in. You can examine that in the actual Hardware, that's where our results with us to just talk about a different brand of Spades experiments of the animal mission. Thanks, love toy Chica and thanks a lot Bryan and Missy and the only World seemed
for inviting me to share the results of the golf on the 8S zero system implementing a computational system to add so who we focus on finding a new methods of using quantum, computers to sell quantum. Chemistry problems. In particular. We're looking at a message in which cumpton chemistry fails in general. One example of this is Mullica Lakeside Estates. Through classical and quantum, chemistry, usually fails at getting good accurate results. For for the states of computing requirements for its weight loss to be implemented. And as a result
of that place itself as a technology that could unlock screen rotation of excited States. And with Minnie Mouse feels like a drug Discovery or another important application is for example of batteries. Do in order to do that has developed a method which is focused on It's a message which has inspired from a, the machine learning a research and more specifically what we called. The different Works through in order to implement this method, we need three of them circuits. The first one is a
model of our molecular ground state. So you can obtain this sample to buy some solder. At the second circuit's is a similar at the moment. We don't know what it doesn't matter for now is tasked to take either of these two as input to distinguish between the two now. We can do to do if these two states top 100. That means that if we do that in which way do the circuit together and at the same time, we try to minimize the energy of the state produce. We can obtain the first excited state set, the volume to sue
is very convenient to because it runs on computers it several times. Now and what we wanted to do with Honeywell is really try to push this this message to its limit and the main benefits of my wheels of HD vs. The hsus system for us, was the Austin engaged Italy's in addition because of the food connectivity with labeled as well. As you can see, we about standing actresses on all of the results. City of you that I have more of a background in chemistry and molecules are both within 1 million, heart rate of the results, which means that we are
without receipt on the 4th with molecule. The first result is within 1 Million Lottery, while the first excited state is within Taylor of the technology. So what you can say is that this reduces of a very promising future Quantum Computing and in particular that the future of application of colon cancer in the closest, my presentation,
Buy this talk
Ticket
Interested in topic “IT & Technology”?
You might be interested in videos from this event
Similar talks
Buy this video
Conference Cast
With ConferenceCast.tv, you get access to our library of the world's best conference talks.