Variational Quantum Algorithms via Measurement-Induced Passive Steering

Variational quantum algorithms (VQA) combine the advantages of classical and near-term quantum computation for solving problems on today's noisy quantum devices. Variational Quantum Eigensolver (VQE) is one of the widely used VQAs, which aims to find the approximate ground state energy of a given Hamiltonian. While traditional VQE implementation is promising for ansatz-based trial state preparation, it requires an initial fiducial state or a reference state, which can be infeasible for large quantum systems. In this paper, we propose a novel approach for trial state preparation in VQE algorithms. This method leverages passive steering, a circuit-based approach with repeated measurements, eliminating the need for an initial fidu-cial state or a reference state. Experimental results demonstrate that passive steering-based state preparation provides improved accuracy and scalability of VQE compared to traditional ansatz-based solutions. Our proposed solution can also be effectively combined with the existing ansatz-based methods, where passive steering prepares the reference state while ansatz prepares the trial state, facilitating a robust and scalable state preparation for variational quantum algorithms.