AI and the World's Hard Problems · Quantum physics & quantum computing · arXiv · 2026-10-22
Variational quantum-neural hybrid imaginary time evolution
Variational imaginary time evolution (VITE) for NISQ devices mimics non-unitary dynamics with parameterized quantum circuits, but conventional variational quantum algorithms struggle to reach high accuracy because of their strong dependence on the choice of ansatz circuit. The proposed VQNHITE combines VITE with a neural-network-based non-unitary operator to enhance expressive power, and introduces an initial-parameter optimization procedure at a small imaginary-time step to stabilize the evolution against random initialization. Numerical simulations on Heisenberg spin chains under both nearest-neighbor and all-to-all circuit connectivities show that VQNHITE consistently achieves higher fidelity with the exact ITE state than VITE.