Supporting files for Schubert et al. (2023)

Description

This repository contains input files, output files and scripts for reproducing the results of the article "Testing Koopmans spectral functionals on the analytically-solvable Hooke's atom".

Contents

The three folders "omega_one_tenth", "omega_one_half", and "omega_ten" contain the corresponding files for the three different values of omega presented in the article. They follow a very similar structure:

  • The folder "Convergence_Analysis" contains all necessary files to run the convergence analysis. It also contains a Readme with instructions needed to reproduce these results
  • The folder "Analysis/Analysis_*" contains all necessary files to run the actual computations. It also contains a Readme with instructions needed to reproduce these results and a script to plot all potentials and extract the eigenenergies as well as the total energies.
  • Finally, a plot containing the densities for all three systems can be generated by running "plot.py" in "plotting".

Reproducing the results

For our calculations we used the following versions of Koopmans and Quantum ESPRESSO:

  1. Quantum ESPRESSO version 4.1 (hereafter referred to as old_q-e)
  2. Quantum ESPRESSO version 6.8 (hereafter referred to as q-e)
  3. Commit 5548809 of the Koopmans repository (https://github.com/epfl-theos/koopmans/tree/55488090) (hereafter referred to as koopmans)

Please provide in "omega_*/paths.txt" your installation paths:

  • old_pp : absolute path to the source code of pp.x of old_q-e
  • new_pp : absolute path to the source code of pp.x of q-e
  • new_pw : absolute path to the source code of pw.x of q-e
  • koopmans_kcp : absolute path to the source code of kcp.x of koopmans

Further you might need to change some settings in the source code of the different QE versions used to run the calculations to get exactly the same results (see below)

Modifications to the source codes

in q-e:

in Modules/read_pseudo.f90

  • approx. in line 65 in readpp() set the parameter rcut from 10.0 to 12.0 for $\omega=0.5$ and to 30.0 for $\omega=0.1$, leave unchanged for $\omega=10$

in XClib/xc_input_params_mod.f90

  • approx. in line 65 and 69 in MODULE dft_par_mod set the parameter rho_threshold_gga from E-6 to E-10 and grho_threshold_gga from E-10 to E-14

in koopmans:

in quantum_espresso/cp_koopmans/Modules/functionals.f90:

  • approx. in line 842 in subroutine gcxc() set the parameter small from E-10 to E-14
  • approx. in line 936 in subroutine gcx_spin() set the parameter small from E-10 to E-14
  • approx. in line 1086 in subroutine gcc_spin() set the parameter small from E-10 to E-14 and epsr from E-6 to E-10

in quantum_espresso/cp_koopmans/CPV/pseudo_base.f90

  • approx. in line 223 in subroutine formfn() change <10.0d0 to <12.0d0 for $\omega=0.5$ and to <30.0d0 for $\omega=0.1$, leave unchanged for $\omega=10$

in quantum_espresso/cp_koopmans/CPV/exch_corr.f90

  • approx. in line 727 in subroutine exch_corr_wrapper() set the parameter epsr from E-6 to E-10 and epsg from E-10 to E-14

in quantum_espresso/cp_koopmans/CPV/nksiclib.f90

  • approx. in line 487 in subroutine before if (draw_pot) then add a line call write_pot_sic ( vsic(:, 1) ). Uncomment this line for the KIPZ_LUMO-calculation

in quantum_espresso/cp_koopmans/Modules/xml_io_base.f90

  • approx. in line 1516 change CALL iotk_write_begin( potunit, "EFFECTIVE-POTENTIAL" ) to CALL iotk_write_begin( potunit, "CHARGE-DENSITY" )
  • approx. in line 1964 change CALL iotk_write_begin( potunit, "EFFECTIVE-POTENTIAL" ) to CALL iotk_write_end( potunit, "CHARGE-DENSITY" )

in koopmans/workflows/_koopmans_dscf.py:

  • introducing mixing can improve the convergence of the self-consistent computation of the screening parameters for KIPZ. To introduce mixing change

      alpha = alpha_guess * (dE - lambda_0) / (lambda_a - lambda_0)

    approx. in line 1015 to

      mixing = 0.9
  alpha_new = alpha_guess * (dE - lambda_0) / (lambda_a - lambda_0)
  alpha = alpha_guess * mixing + (1-mixing)*alpha_new"

in old_q-e:

in espresso-4.1/PW/read_pseudo.f90

  • approx. in line 34 in readpp() set the parameter rcut from 10.0 to 12.0 for $\omega=0.5$ and to $30.0$ for $\omega=0.1$, leave unchanged for $\omega=10$

in espresso-4.1/PW/gradcorr.f90

  • approx. in line 51 in subroutine exch_corr_wrapper() set the parameter epsr from E-6 to E-10 and epsg from E-10 to E-14

in espresso-4.1/Modules/functionals.f90:

  • approx. in line 844 in subroutine gcxc() set the parameter small from E-10 to E-14
  • approx. in line 939 in subroutine gcx_spin() set the parameter small from E-10 to E-14
  • approx. in line 1089 in subroutine gcc_spin() set the parameter small from E-10 to E-14 and epsr from E-6 to E-10