Robust ab initio calculation of condensed matter: transparent convergence through semicardinal multiresolution analysis

TL;DR

This paper introduces a wavelet-based all-electron density-functional method that achieves systematic and predictable convergence with chemical accuracy, enabling precise exploration of materials and development of accurate density functionals.

Contribution

It presents the first wavelet-based all-electron density-functional calculations with gradient corrections in solids, providing transparent convergence with error prediction.

Findings

Achieves chemical accuracy (millihartree level) in condensed matter calculations.

Provides systematic convergence with a priori error estimates.

Enables reliable exploration of materials under novel conditions.

Abstract

We present the first wavelet-based all-electron density-functional calculations to include gradient corrections and the first in a solid. Direct comparison shows this approach to be unique in providing systematic ``transparent'' convergence, convergence with a priori prediction of errors, to beyond chemical (millihartree) accuracy. The method is ideal for exploration of materials under novel conditions where there is little experience with how traditional methods perform and for the development and use of chemically accurate density functionals, which demand reliable access to such precision.