Applying the Liouville-Lanczos Method of Time-Dependent Density-Functional Theory to Warm Dense Matter

TL;DR

This paper introduces and validates the Liouville-Lanczos method as an efficient alternative to standard LR-TDDFT for modeling dynamic structure factors in warm dense matter, enabling broader and more scalable simulations.

Contribution

The paper presents the Liouville-Lanczos approach as a new, computationally advantageous method for simulating DSFs in WDM, avoiding the need for empty states and extending the accessible energy range.

Findings

Liouville-Lanczos method accurately reproduces LR-TDDFT results

The approach effectively models large momentum transfer and broad frequency ranges

Validation against PIMC benchmarks confirms its reliability

Abstract

Ab initio modeling of dynamic structure factors (DSF) and related density response properties in the warm dense matter (WDM) regime is a challenging computational task. The DSF, convolved with a probing X-ray beam and instrument function, is measured in X-ray Thomson scattering (XRTS) experiments, which allows for the study of electronic structure properties at the microscopic level. Among the various ab initio methods, linear response time-dependent density functional theory (LR-TDDFT) is a key framework for simulating the DSF. The standard approach in LR-TDDFT for computing the DSF relies on the orbital representation. A significant drawback of this method is the unfavorable scaling of the number of required empty bands as the wavenumber increases, making LR-TDDFT impractical for modeling XRTS measurements over large energy scales, such as in backward scattering geometry. We consider and test an alternative approach that employs the Liouville-Lanczos (LL) method for simulating the DSF. This approach does not require empty states and allows the DSF at large momentum transfer values and over a broad frequency range to be accessed. We compare the results obtained from the LL method with those from the standard LR-TDDFT within the projector augmented-wave formalism for isochorically heated aluminum and warm dense hydrogen. Additionally, we utilize exact path integral Monte Carlo (PIMC) results for the imaginary-time density-density correlation function (ITCF) of warm dense hydrogen to rigorously benchmark the LL approach. We discuss the application of the LL method for calculating DSFs and ITCFs at different wavenumbers, the effects of pseudopotentials, and the role of Lorentzian smearing. The successful validation of the LL method under WDM conditions makes it a valuable addition to the ab initio simulation landscape, supporting experimental efforts and advancing WDM theory.