✨ Scalable Clifford (sub-)circuit optimizer

[jannikpflieger]

What's the problem this feature will solve?

Currently, a depth-optimal Clifford synthesis is implemented. Usually, depth-optimal approaches scale poorly and are thus limited by the number of qubits.

A first heuristic is already implemented. However this implementation is not suited to optimize Clifford sub-circuits in a non-Clifford circuit (Issue: #365).

Describe the solution you'd like

To tackle the scaling problem, we can make use of scalable heuristics, such as peephole optimization.

We can do this by identifying blocks of Clifford gates and optimizing each block within a non-Clifford Circuit.
While doing this, we can implement different strategies for performing the (peephole) optimization. Such as horizontal circuit splitting (Issue: #305).

[vishnunathasuresh]

Hi, I’m a quantum enthusiast interested in quantum compilation and would like to work on this. My rough plan would be to identify maximal Clifford windows inside mixed Clifford/non-Clifford circuits, add a bounded peephole or horizontal-splitting strategy for each window, and validate gate-count/depth improvements with equivalence checks and existing synthesis heuristics. I’d keep the optimizer strategy configurable so the current depth-optimal path remains available. Would you be open to assigning this to me?

[denialhaag]

Hi, @vishnunathasuresh! You know the drill by now: Feel free to open a PR and request a review from me. I would suggest you choose either this or #670 for the beginning and not work on both at the same time. 🙂