Physically Interpretting Quantum Architecture Search
Quantum Architecture Search (QAS) refers to the process of automatically generating a parameterized quantum circuit (ansatz) which can minimize the energy of a family of Hamiltonians, typically for a variational quantum algorithm or a quantum machine learning objective. Many state of the art QAS approaches apply classical machine learning methods to generate quantum circuits, without considering the inherent quantum properties of the search problem. This project aims to provide a physical interpretation of QAS by modeling the search process with mixed states, and glean insights from this interpretation to improve the state of the art.
Swagat Kumar, Jan-Nico Zaech, Colin M. Wilmott, Luc Van Gool
A sampling-free differentiable QAS algorithm that models the search process as the evolution of a mixed state, enabling the incorporation of generic quantum noise channels and requiring significantly fewer quantum simulations for convergence when compared to other methods.