QuteFuzz: Fuzzing quantum compilers using randomly generated circuits with control(…)
summary
QuteFuzz generates fully random quantum programs with nested subroutines and control flow, runs them through Qiskit, Pytket and Cirq stacks, and does differential testing via state-vector comparison and KS tests for shot‑based outputs. The tool finds 17 bugs (12 compiler, 5 simulator), discovers issues like miscompiled unitaries and crashes when adjacent subcircuits appear, and scales to ~500,000 test circuits to produce small, minimized reproductions; QuteFuzz is implemented and effective.
AI-written summary · about
chapters (12)
- 0:00QuteFuzz introduction
- 1:28Quantum compiler testing
- 2:29QuteFuzz contributions
- 3:38Random circuit generation
- 4:21Live circuit demo
- 5:26Differential testing
- 7:54Automated bug detection
- 8:35Bugs found and impact
- 9:17Notable bugs
- 10:42Lessons learned
- 11:49Future work
- 12:55Q&A
related talks
22mSatisfiability Modulo Fuzzing: A Synergistic Combination of SMT Solving and Fuzzing
53mLocknote: How Badly Do We Want Correct Compilers?
17mQGAT: A Generate-and-Test Paradigm for Quantum Circuits
17mQVM: Quantum Gate Virtualization Machine
26mGenerating Compilers for Qubit Mapping and Routing
39mBuilding a Compiler for Quantum Computers
01h 10mThe Rise & Collapse of a Quantum State
19mEquality Saturation for Quantum Circuit Optimization