Quantum Circuit Superoptimization
Engineering
Quantum Computing
Quantum computing has the potential to be more efficient than classical computing in many areas. However, as quantum computers are currently expensive and difficult to run, the success of operating a quantum computer is limited by the complexity of the circuit that it is running. Quantum circuit optimization is therefore an important tool for making quantum computers more successful. In this paper, we design a new method of optimizing quantum circuits. Our method supports different definitions of optimizations for different use cases, specified by the user. We use an approach inspired by stochastic superoptimization, where a program is broken up into pieces, and each individual piece is separately optimized. Each circuit piece is optimized by modeling the "cost" of the circuit as a function, and stochastically searching for the global minima of that function. New in our algorithm is a form of stochastic synthesis inspired by evolutionary programming, where a population of circuits are simultaneously evolved, and only the most fit survive each round of mutation. In our experiments we found that we can achieve about a 20% decrease in gate count for some circuits, which outperforms or is comparable to other mainstream quantum circuit optimizers.
