Exploring Non-Linear Programming Formulations in QuantumCircuitOpt for Optimal Circuit Design

Given the limitations of current hardware, the theoretical gains promised by quantum computing remain unrealized across practical applications. But the gap between theory and hardware is closing, assisted by developments in quantum algorithmic modeling. One such recent development is QuantumCircuitOpt (QCOpt), an open-source software framework that leverages state-of-the-art optimization-based solvers to find provably optimal compact circuit decompositions, which are exact up to global phase and machine precision. The quantum circuit design problem can be modeled using non-linear, non-convex constraints. However, QCOpt reformulates these non-linear constraints using well-known linearization techniques such that the resulting design problem is solved as a Mixed-Integer Linear Programming (MILP) model. In this work, we instead explore whether the QCOpt could also be effective with a continuous Non-Linear Programming (NLP) model obtained via relaxation of the integer variables in the non-linear constraints. We are able to present not only multiple significant enhancements to QCOpt, with up to 11.3x speed-up in run times on average, but also opportunities for more generally exploring the behavior of gradient-based NLP solvers.