First-hand information for everyone
Researchers describe a way to compute charge-changing nuclear strength functions more efficiently. The method uses a symmetric Lanczos projection to approximate the QRPA (quasiparticle random-phase approximation) strength function across a wide energy range from a single Krylov run. That contrasts with standard finite-amplitude-method (FAM) calculations that must solve a separate linear system at each frequency point.
Strength functions tell physicists how a nucleus responds to a weak external probe. They matter for predicting beta decay, double-beta decay, collective excitations, and weak interaction rates in medium and heavy nuclei. The QRPA is a standard microscopic tool for describing those excitations, but building and diagonalizing the full QRPA matrix quickly becomes impractical for large or deformed nuclei. The finite-amplitude method (FAM) avoids forming the full matrix by solving linear-response equations, but it still proceeds one frequency at a time.
Starting from the FAM linear-response equations, the authors derive equivalent spectral formulas and, for the common real-matrix case, a reduced eigenvalue problem built from the products MK and KM, with M = A + B and K = A − B. Here A and B are the usual QRPA matrices. This reduced form lets them apply a symmetric Lanczos process (a matrix-free iterative projection) to approximate a Lorentzian-smoothed strength function over a broad energy interval from a single Krylov run instead of many separate frequency solves.
They tested the approach on two nuclei, 112Sn and 150Nd. As an intermediate step they show that GMRES (a common iterative solver) reproduces the converged FAM strength profiles while using fewer iterations than earlier iterative approaches. Using GMRES solutions as a reference, the Lanczos approximation gives the same strength profiles but with reduced overall computational cost. The authors conclude that symmetric Lanczos projection is an efficient and accurate option when one needs spectral information across an extended frequency range.
There are important conditions and caveats. The reduced MK/KM formulation assumes real QRPA matrices with certain symmetry and positivity properties (for example A† = A and B^T = B and positive-definiteness), so the dimension reduction and some proofs apply in that case. Strengths are computed with a small imaginary shift (a smearing width γ > 0) to avoid singular behavior when a frequency lies near an eigenvalue. The Lanczos result is an approximation whose accuracy depends on the Krylov run size and the chosen smearing. The numerical evidence in the excerpt comes from calculations for two nuclei; further tests would be needed to judge performance across the full chart of nuclides.