Constant-Per-Layer-Depth MPS-Pretrained Ansatz for Noisy Distributed Quantum Processors
The study compared three circuit designs—ladder, mixed-canonical, and brick-wall—for training quantum algorithms using matrix-product states (MPS) under noise. While all designs performed well ideally, the constant-depth brick-wall layout consistently yielded the lowest error in noisy simulations. Extending this to distributed processors showed that hardware structure and communication costs significantly impact variational algorithm performance.