Hello.
I noticed Bloqade-circuit already supports PyQrack, Stim, and TSIM as simulator backends. I'd be interested in contributing a backend using Maestro, an open-source tensor network simulator with MPS support (CPU + GPU).
The main thing Maestro would add is MPS-based circuit simulation that scales beyond what statevector backends can handle (~25+ qubits), while still supporting full gate-level simulation, sampling, and expectation values.
Before putting together a PR, I wanted to check:
- Would this be welcome as a contribution?
- Is there a preferred pattern for adding new simulator backends? I've been looking at the PyQrack integration as a reference, but wondering if there's newer guidance given the StackMemorySimulator / DynamicMemorySimulator refactor.
- Any particular features/requirements that would make this most useful to Bloqade users?
Happy to discuss further, just wanted to gauge interest before diving in.
Hello.
I noticed Bloqade-circuit already supports PyQrack, Stim, and TSIM as simulator backends. I'd be interested in contributing a backend using Maestro, an open-source tensor network simulator with MPS support (CPU + GPU).
The main thing Maestro would add is MPS-based circuit simulation that scales beyond what statevector backends can handle (~25+ qubits), while still supporting full gate-level simulation, sampling, and expectation values.
Before putting together a PR, I wanted to check:
Happy to discuss further, just wanted to gauge interest before diving in.