New method hits a theoretical limit for quantum algorithm sampling cost

Researchers have developed a new method, Mirror Subspace Diagonalization (MSD), that reduces the computational cost of simulating molecules on quantum computers by a factor of 10 to 10,000 compared to conventional algorithms. The work centers on a finite-difference formula expressing the Hamiltonian, enabling efficient estimation within a Krylov subspace and approaching a theoretical limit for sampling cost. MSD proves particularly effective for high-accuracy molecular simulations involving strong electronic correlations, where this advance offers a path toward using near-term quantum devices for complex materials science and drug discovery. Employing a finite-difference formula expressing the Hamiltonian as a combination of time-evolution unitaries with symmetrically shifted timesteps, Mirror Subspace Diagonalization (MSD) enables efficient estimation…