✨ MLIR - Machine Learning Compiler

[flowerthrower]

Long-term goal

The long-term goal of this issue is to develop an ML-assisted compiler that can automatically select suitable circuit transformation passes and compile a quantum circuit for given hardware constraints (similar to the MQT Predictor).

This main issue will serve as a tracker for multiple sub-issues required to achieve this goal.

Initial sub-tasks

• Add more MLIR passes — we need a sufficient variety of passes to enable meaningful selection by the ML model.
• Set up a compiler pipeline — configure and execute the passes in sequence.
• Integrate an ML framework — enable training and inference of a model that can select and order MLIR passes based on circuit and hardware characteristics.

These tasks vary in complexity and effort. Each will be split into dedicated sub-issues, where we will further refine scope, requirements, and technical details.