Ocean for Analog Hamiltonian Simulation

[christianbmadsen]

Modify https://github.com/aws/amazon-braket-ocean-plugin-python/ in a new branch such that you can embed a QUBO problem onto QuEra's Aquila machine via Braket.

Rewrite and demonstrate that existing examples can indeed run against Aquila with reasonable modifications (may need to for instance reduce the number of variables).

[JordanAWS]

If you want to work on this issue, please feel free to ping me to request AWS credits to cover the cost of running on the QuEra Aquila device :)

[Aaron-Robertson]

@JordanAWS I'm giving this one a shot! Could you send some AWS credits? I couldn't find your username on the discord to reach out (I'm A-Aron)

[JordanAWS]

Great! Can you send me an email at [email protected] with your AWS account ID? If you haven't yet created an account, please do so first :)

[Aaron-Robertson]

Apologies @JordanAWS, I got wrapped up in other issues for now! My goal is to come back to this one later today or tomorrow, but probably best to consider the "playing field" open for other hackers at the moment!

[JordanAWS]

@Aaron-Robertson Got it. It looks like you're currently the only one working on this issue, so it's all good if you'd like to come back to it later :)

[golanor]

I'd be happy to work on this with @Aaron-Robertson if he wants, or to work on it alone.

I want to make sure I understand the scope of the work here:

1. Create a new class BraketAquilaSampler in the plugin.
2. Modify the examples to work on Aquila.
3. Test on the device.

Is that correct @JordanAWS ? Any other things to take into consideration? (Other than asking for credits for testing the examples)

[Aaron-Robertson]

@golanor A collab sounds great to me! I'm not so sure we'll get by so easily, but if my hunch/brainstorming turns out correct, it will be a lot of fun!

First off, I think we'll need some metadata mapping, as is done in the BraketSolverMetadata.

Second, I'm looking in to a slightly different approach, though still just brainstorming for now. Ocean, as far as I can tell, does BQM based sampling under the hood, almost exclusively. Based on the resources on Aquila[1][2], it is similarly singularly focused, but on MIS problems. So, I'm wondering if what we need is really a topological embedding from QUBO to MIS, such that the ocean-based samplers directly (or as close as possible) interact with the embedded QUBO, which map to an Aquila MIS problem under the hood. Let me know what you think!

[Aaron-Robertson]

@golanor Not sure if you're still interested, but I reached out to Quera yesterday about a broken link in the optimization section of their site, and turns out it's the exact solution to the problem at hand 🤣! Classic, anyways I'm reading through it[1] now, to incorporate properly with the above stubs.

If you are still interested in collaborating, we may want to hop on a call and discuss the path forward and which pieces to each take on. Feel free to message me on the UF discord if that's easy, I'm A-Aron.

[JordanAWS]

I'd be happy to work on this with @Aaron-Robertson if he wants, or to work on it alone.

I want to make sure I understand the scope of the work here:

1. Create a new class BraketAquilaSampler in the plugin.
2. Modify the examples to work on Aquila.
3. Test on the device.

Is that correct @JordanAWS ? Any other things to take into consideration? (Other than asking for credits for testing the examples)

Yep!

[JordanAWS]

@Aaron-Robertson It looks like there's still some work remaining on your PR. Do you think you'll be able to finish it by the end of UnitaryHack today?

[JordanAWS]

@Aaron-Robertson It looks like the PR above doesn't quite cover the acceptance criteria for being eligible for the UnitaryHACK bounty, but if you are still interested in working on this issue, we can offer AWS credits as an alternative :)