Backpropagation for Pauli Propagation

A backpropagation algorithm for Pauli-propagation simulation that targets quantum-circuit parameter gradients with lower memory than conventional reverse-mode AD and fewer function evaluations than finite differences.

TL;DR — The abstract presents a backpropagation algorithm for evaluating parameter gradients in quantum circuits using Pauli propagation simulation. It claims complexity comparable to standard sparse Pauli simulation, gradient accuracy of the same order as observable expectation values, O(n_param) lower memory than conventional reverse-mode automatic differentiation, and O(n_param) fewer function evaluations than finite differences. Demonstrations include low-energy state preparation for transverse-field Ising and Heisenberg models and compression of two-dimensional time-evolution circuits.

Core contribution

The abstract’s central claim is that the authors develop “a backpropagation algorithm for evaluating parameter gradients in quantum circuits using Pauli propagation simulation.” This is a gradient-evaluation method for classical optimization workflows involving quantum circuits.

The contribution should be cited narrowly. The abstract does not claim a new optimizer, a hardware measurement protocol, or a general quantum advantage result. It claims a way to compute gradients within Pauli propagation simulation with favorable comparisons to specific baselines: standard sparse Pauli simulation techniques, conventional reverse-mode automatic differentiation, and finite difference methods.

Full analysis, extracted claims, numerical results, entity graph, FAQ, related work, applications, and BibTeX are available via x402 micropayment.