Previous talks at the SCCS Colloquium

Omar Ibrahim: A Compilation and Optimization Pipeline for OPENQASM 3.0 using MLIR

SCCS Colloquium |


Combining quantum and classical computation is crucial to solving many problems in
quantum computing. OPENQASM 3.0 is a specification for a quantum programming language
that allows for just that. However, the need for a compilation pipeline to lower this language,
optimize it, transform it, and execute it is essential to make use of this specification. In this
work, we present a compilation pipeline for OPENQASM 3.0 using MLIR. We create a new
MLIR dialect for general quantum operations and a specific dialect for the restricted set of
gates of a superconducting quantum processor. Utilizing MLIR’s IR and passes, we were able
to perform quantum circuit optimizations, and lower our quantum circuit to the restricted
set of gates. To test our compilation pipeline, we created a state-vector quantum simulator
in MLIR, allowing easy lowering to LLVM intermediate representation and execution using
LLVM’s just-in-time compiler. While we were able to produce interesting results regarding
the optimization of quantum circuits, and their transformation, our quantum simulator is
still in its early stages and needs further development to simulate more complex quantum
circuits.

Master's thesis presentation. Omar is advised by Martin Knudsen.