New method estimates quantum circuit transitions using bidirectional paths

A new algorithm uses random walks from both ends to sample paths in large quantum circuits.

Researchers proposed an approximate classical algorithm. It estimates transition amplitudes ⟨Ψf∣U^∣Ψ0⟩\langle \Psi_f | \hat{U} | \Psi_0 \rangle for quantum circuits.

The method uses Monte Carlo sums over bidirectionally sampled Feynman paths. Feynman paths are sequences of intermediate computational basis states.

The algorithm combines information from random walks. These walks start at the input state Ψ0\Psi_0 and the output state Ψf\Psi_f.

It uses Multiple Importance Sampling to reduce variance. This technique avoids variance peaks from low sampling probabilities of hard to reach paths.

This bidirectional approach shows a superior convergence rate compared to forward-only sampling. It finds and weights paths with non-zero amplitude in sparse path space.

The computational cost depends on the variance of the estimator. This method works on very large quantum circuits with up to 4096 qubits.

Key facts

  • The algorithm uses bidirectional sampling of Feynman paths.
  • It employs Multiple Importance Sampling to avoid variance peaks.
  • The method was tested on circuits with up to 4096 qubits.

All sections

Edition