A new method helps find molecular energies on quantum hardware

The CURE-QKSD method uses classical post-processing to recover accurate energy values from quantum calculations.

Researchers developed Classical Unwrapping for Resolving Eigenvalues (CURE-QKSD). This method helps compute molecular energies on early fault-tolerant hardware.

The method works with unitary quantum Krylov subspace diagonalization (UQKSD). UQKSD extracts energies from a projected real-time propagator.

At a small time step τ\tau, Krylov vectors become nearly linearly dependent. This makes the overlap matrix ill-conditioned.

A larger time step improves conditioning and reduces the Krylov dimension MM. However, energies are then known only modulo 2π/τ2\pi/\tau.

CURE-QKSD uses phase-consistency landscapes from one or two UQKSD runs. This classical post-processing recovers the missing multiples of 2π/τ2\pi/\tau.

In tests on six molecules with 12--20 qubits, errors were at most 1.1×10−71.1\times10^{-7}~Ha. With Gaussian noise where σ=10−4\sigma=10^{-4}, the median error reaches chemical accuracy.

Key facts

  • CURE-QKSD recovers missing multiples of 2π/τ2\pi/\tau using phase-consistency landscapes.
  • The method works on six molecules using 12--20 qubits.
  • Uncorrected energies were off by 0.70.7--115115~Ha under Gaussian noise with σ=10−4\sigma=10^{-4}.

All sections

Edition