As the scale of quantum circuits grow, so too does the bottleneck of circuit compilation. In light of the anticipated scale of quantum circuits for useful operation, methods of reducing compilation time are needed. However, not only are few such methods available, our ability to even test them is highly constrained by the availability of very few large-scale circuits in benchmarking suites, a situation exacerbated by the inability of random circuit generators to produce results whose properties match those of the same suites. This work describes a random circuit generator which enable control of circuit depth, width and density. This is used to enable 8,000 test cases to evaluate a novel compiler parallelisation approach that divides the circuit into sub-circuits which are compiled in parallel. With Qiskit, this approach achieves a peak speedup of 15.56 with overheads below 1%.
Robin Adams Chalmers University of Technology, Gothenburg University, Sweden, Jean-Philippe Bernardy Chalmers University of Technology, Gothenburg University, Sweden, Lorenzo Perticone Chalmers University of Technology, Gothenburg University, Sweden, Jeremy Pope Chalmers University of Technology, Gothenburg University, Sweden