HUGR: A Quantum-Classical Intermediate Representation
summary
HUGR is a graph‑based intermediate representation that encodes mixed quantum‑classical programs with strict, static typing and linear quantum types to prevent invalid cloning; it models nodes as nested hierarchical regions so control flow, unbounded tail loops and measurement‑driven branches live in the IR. A Rust reference implementation with Python bindings and open‑source GitHub repos lowers to MLIR/LLVM (and the team’s tiket‑2 compiler), targets devices in the thousands of qubits era, and already enables pattern matching across hundreds‑of‑thousands of patterns for optimization.
AI-written summary · about
chapters (14)
- 0:00HUGR introduction
- 1:06Limitations of quantum circuits
- 2:55Post-NISQ era needs
- 4:44Why circuits are insufficient
- 7:52HUGR’s data flow design
- 9:16Linear types for safety
- 10:41Control flow in HUGR
- 12:48Functions and custom types
- 15:14Compiler pipeline with HUGR
- 17:25Rust toolchain and ecosystem
- 18:28Conclusion and future
- 18:49Q&A: Timing hierarchies
- 20:34Q&A: Type system challenges
- 21:58Q&A: Hardware limits
related talks
16mCompiling and Running High-level Quantum Programs
18mGraph IRs for Impure Higher-Order Languages: Making Aggressive Optimizations Af...
55mScalable Quantum Compilation With Jasp
19mQuantum Virtual Machines
20mQuantum Register Machine: Efficient Implementation of Quantum Recursive Programs
15mCompositional Quantum Control Flow with Efficient Compilation in Qunity
15mThe Fuzion Intermediate Representation
17mQVM: Quantum Gate Virtualization Machine