National Energy Research Scientific Computing Center (NERSC) Quantum for Science Day, October 24, 2022, 4 Nov 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Advanced Quantum Testbed (AQT @ LBL)

Description

Advanced Quantum Testbed (AQT @ LBL)
Kasra Nowrouzi (LBL)

A

My name is Catherine aruzzi I am the head of hardware at the advanced Quantum test bed at Lawrence, Berkeley National, Lab So. Today we're going to give you an overview of what the aqt is, including the technical aspects and the user program.

A

um So we are one of two U.S department of energy Quantum test bets. Our sister testbed is Q Scout at Sandia, National Labs. It's a trapped ion platform, so check them out, but we are the the superconducting platform we develop and operate a full stack and also conducting Quantum Computing platform, with a growing number of quantum processors, uh cryogenic platform development in partnership with blacksimo, which is a startup here in Berkeley.

A

We work on room temperature, controlled Electronics, both in-house developed at lbl, which is open sourced and then also commercial solutions by Zurich instruments and others, and we also work on the software stack and collaborate with others at the Lawrence booking National Lab, to develop tools and deploy them to optimize circuits and mitigate errors, and uh this platform is then made available to users from across Academia National, Labs and Industry, and over the last two three years when we have had an open call for proposals, we've had uh really a mix of all of these entities that we've uh collaboratively run.

A

Experiments with, um on the industrial side, uh to particular users, we're from Quantum Benchmark and super tech which, upon demonstrating their uh their experiments and products here at the eqt, were then acquired by keysight and called quanta over the last year, or so. The 2022 call for proposals is open now and everyone can apply at hqt.lbl.gov.

A

It's official we're officially past the the deadline for the once a year, the annual call, but we do accept proposals on a rolling basis.

A

So uh this is what the full stack looks like. Roughly speaking. We have, of course, all of the cryogenics, so dilution fresh to cool down uh the processors to 10 Milli Kelvin. We have the room temperature Electronics uh here, the two versions that I spoke about and uh the the quantum processors and then, of course, multidisciplinary team here at UC, Berkeley and northbrooki National, Lab and our collaborators.

A

So it starts with the the quantum processor. We have a number of these, the really the bread and butter. uh The Workhorse that we use for running experiments right now is a transmon based Quantum processor, uh we're now in version six or seven uh of the Trailblazer chip.

A

um This is roughly what it looks like so there's there are basically eight trans months in packed in a ring geometry. The transmods here are in green they're, coupled they're all fixed frequency, qubits they're, coupled through fixed resonators in uh purple and driven individually through the lines, the control lines here in blue um and uh they are coupled to a common readout bus uh through these resonators in Red. So readout is done in reflection in in multiplexed, um and uh so this is what we've been using. We have a number of other.

A

uh Well, we have a few different types of um uh two-bit implementations under development, including uh flexonium and and more um and also we have other uh architectures for Quantum processors, also under development, the Trailblazer chip, uh so these are this- is a bit more specifications for those who are interested uh details on coherence times, uh frequencies, fidelities um and uh gate development.

A

Also, uh particularly, we were pretty much the only facility where you can find a native uh Italy gate um and uh the the standard two cubic gate that we use is is the CZ gate. We've also implemented q-treat experiments on this chip as well, so we have access to larger, larger Hilbert spaces.

A

So this is where we fall on the coherence plot. This is the coherence data averaged among all of the different qubits um and uh is representative of our devices above 100 microseconds each um and uh the plot is slightly out of date now over the last year or so, IBM has rolled out another processor that is somewhere to the top right of the plot here above 150 microseconds, but other than that is relatively uh relatively active. Relatively up to date, then, we have the cryogenic platform.

A

So, between the campus side of uh efforts at the quantum Electronics lab and the aqt, we have access to seven or eight cryostas. This is the largest one which we affectionately refer to as blizzard. uh It has 160 uh microwave drivelines uh microwave lines. 16 of them are superconducting for readout under the base, plate is an experimental stage, as you see here on the top right.

A

This allows us to compartmentalize the experimental space so that we can, uh you know, really partition and use it for multiple experiments at once.

A

So at any given point, we have uh two to three different Trailblazer chips available for experiments, but also a number of uh more novel architectures and then for the development of this cryo stage and the the cryo packaging that houses and thermalizes the chip we collaborate with black similar, which is the startup I mentioned about earlier, um and some more details and and pictures uh of of the same platform or the controlled Hardware. We have two solutions on the industrial side. uh Our industrial partner here is Rick instruments.

A

We have a number of these boxes, the so these are all fpga based control, Electronics uh that generate the RF pulses to control the qubits and do readout, um and so we've worked closely with work instruments to develop uh firmware features that would basically Place us at The Cutting Edge of solutions available, so that we can run uh experiments efficiently and we're now working with them on uh on more modern feedback schemes to enable more sophisticated experiments that our users are demanding, um and so that is the standard hardware for the users.

A

But in addition to that, uh we also have been working a portion of our team at the advanced technology at the accelerator Technologies and applied physics division at Berkeley lab uh we've been working with them to develop an in-house fpga based solution, which we refer to as a cubic standing for qubit control. So this allows us more flexibility in terms of uh rolling out new features. So if a particular experiment uh needs features that are not available on the commercial Solutions, then then we can use this as well.

A

This has been an active area, development for us, and we've demonstrated you know fast reset and crosstalk compensation and some other features on this uh and then the system, integration level at the the heart of that is a framework we refer to as qtroll for Quantum control, um and this really sits in between uh the hardware and the the abstract circuits that that users can submit.

A

um You can generate the circuits from any number of uh open uh Solutions, and then this gets transpiled into the native uh Native library of qtrl, and then the cultural interfaces with the different um different uh control Electronics to run the experiment.

A

um And then we've been actually also working to integrate a number of uh noise, characterization and mitigation tools into this, and also been working with uh the part of our team or our collaborators in the Computing Sciences area, at Berkeley, lab to integrate uh and take advantage of circuit, optimization tools such as big skit and Q, search to reduce the the depth of circuits so that we can bring more complicated, more complex experiments within the reach of our existing nisk Hardware.

A

So just putting it all on the map. uh This is roughly what it takes to to uh run all of this and to deploy this uh a few different divisions. Here at Berkeley lab uh the we've been, we've been working with the molecular Foundry to work on uh to improve the material processes and and increase our coherence times.

A

The accelerator division is the contribution is to the fpga controlled Hardware development, Computing Sciences area for uh software tools, and also this is where nurse is um and then here to the left would be the campus where we have our fabrication facilities and some of our measurement labs.

A

uh So a little bit on the user uh on the implementation process for the user projects, applications uh as I mentioned can be submitted. So this is the link, uh and this is again open to users from Academia industry and National Labs. The way the process works is once the user proposal advances through the LOI stage. So it's a letter of intent. It takes maybe like half an hour or an hour just to submit a simple Loi.

A

Then the next step after that is a full proposal, that's slightly more detailed and then it makes its way through the review stages, internal and external, and at that point points of contact are identified within the advanced Quantum test bed for each post project uh based on interest and expertise and availability, and the points of contact coordinates uh with the proposal pis to set up introductory meetings and uh to discuss the details of collaboration and feasibility and scope of the project.

A

And typically, what we find is that jeep collaboration is required between the two teams between us and the proposing team to really arrive at the right form of the project to be run. So we don't just simply execute circus is really deeply collaborative, which we find is really required to make things work with with existing Technologies.

A

And then at that point, uh once the the two teams agree and and uh feasibility is, is assessed. um You know, after making sure that the projects really align with the aqt mission and leverage its unique capabilities. So in other words, it's not just something that you can run on any Cloud solution. It has to be a little bit more complicated and and require the Deep expertise of uh and the resources of, the advanced ground, testbed and our Personnel.

A

um Then at that point, uh work starts, and uh it takes sometimes between six months to a year to run these projects and the two teams produce monthly updates that that are reported to the US Department of energy.

A

um So since we're here at nurse I'm talking about Quantum Computing, a little bit of an analogy with the history of classical Computing is is useful.

A

um In uh you know, roughly about a century ago, as classical Computing was getting started, it took us a while to really Master the vacuum tube uh systems and by the 40s and 50s people were packing vacuum tubes into uh big rooms and and creating large classical computers. That way and uh then decade by decade.

A

Slowly, progress was made through the invention of the transistor and packing these transistors into integrative circuits and creating microprocessors out of them, and then you know half a century later here we are, and we have all the stuff in our phones and uh for perspective. It's really good to keep in mind that uh in Quantum Computing we're really in the vacuum tube era, and we really have to care about every single uh detail of how the systems work and and every single qubit requires individual attention and every single experiment has to be crafted carefully.

A

um So I find it helpful to keep this perspective in line um and uh I. Think that's I'll stop there. I have backup, slides and other things. If people have questions happy to answer them.