PennyLane Catalyst
JIT compile PennyLane hybrid quantum-classical programs for significant performance gains. Supports advanced control flow, differentiation, and integrates seamlessly.
Catalyst is an experimental package for just-in-time (JIT) compilation of hybrid quantum-classical programs within PennyLane. It allows you to compile the entire workflow, including optimization loops, directly from Python using the @qjit
decorator. This integration provides significant performance improvements over standard execution. Key features include:
- Compilation of the full quantum-classical workflow.
- Seamless integration with PennyLane via
@qjit
. - Support for advanced control flow combining quantum and classical instructions.
- Infrastructure for compiling both quantum and classical components.
- Built for end-to-end differentiability.
- Support for high-performance simulators and devices.
The core compiler is built on MLIR, lowering to LLVM + QIR. A C++ runtime enables execution on various devices. The Python frontend provides the @qjit
decorator and functions for control flow, gradients, and mid-circuit measurements.
Categories:
Similar to PennyLane Catalyst:


