Optimized binning for response function reconstruction via Chebyshev expansions
Immo C. Reis, Joanna E. Sobczyk, Sonia Bacca
TL;DR
This work tackles the challenge of reconstructing nuclear response functions from integral transforms without performing ill-posed inversions. It introduces a stochastic regularized density-of-states approach to adaptively bin spectra in equal-area bins, guided by Chebyshev moments to rebuild $R(q,\omega)$ with transparent uncertainty bounds. Demonstrated on the deuteron using a chiral NN interaction in a harmonic-oscillator basis, the method achieves excellent agreement with exact theory and experimental data for both electric dipole and longitudinal responses, and provides a general framework for ab initio lepton-nucleus cross sections in larger nuclei. The approach promises robust, uncertainty-quantified predictions for neutrino-nucleus and electron-nucleus scattering across a range of many-body systems.
Abstract
We propose an optimized histogram binning strategy to reconstruct nuclear response functions via the Chebyshev expansion bound-state method. Our approach employs a stochastic regularization of the density of states to define adaptive, equal-area bins. Using the deuteron solved in a harmonic-oscillator basis with a chiral interaction, we benchmark on dipole and longitudinal responses, obtaining excellent agreement with exact theory and experiment. This general framework readily extends to other many-body systems and opens the door to new ab initio calculations of lepton-nucleus cross sections in medium-mass nuclei.
