Skip to content

SCFControl

This page gives hints on how to control the SCF cycle with the ABINIT package.

Introduction

The numerical precision of the calculations depends on many settings, among which the precision in solving the Kohn-Sham self-consistent equation.

Several parameters govern the SCF loop. The maximum number of cycles is given by nstep, but the iterative procedure might be stopped earlier, as soon as the criterion chosen by the user is fulfilled. The user is asked to give a tolerance on some measure of the convergence. The user must choose among toldfe, toldff, tolrff, tolvrs and tolwfr.

  • The most theoretically justified for the density/potential self-consistency is tolvrs.
  • tolwfr is interesting for non-self-consistent calculations.
  • For molecular dynamics (which rely on the accuracy of forces), one might prefer tolrff.

Note that tolwfr (which controls the precision of wavefunctions), can be coupled with the other tolerances (which control the precision of physical quantities and/or the SCF loop).

Some input variables relate to the solution of the Schrodinger equation. However, usually the related iterative techniques are well-tuned, so that these input variables (nline, tolrde or tolwfr_diago) are usually used only by experts. However, in cases where the convergence is difficult, it might be interesting to test improving them, as well as modifying nnsclo.

The accuracy variable enables one to tune the accuracy of a calculation by setting automatically up to seventeen variables.

{{ related_variables }}

Selected Input Files

{{ selected_input_files }}

Tutorials

  • The tutorial 2 deals again with the H2 molecule: convergence studies, LDA versus GGA
  • The tutorial 3 deals with crystalline silicon (an insulator): the definition of a k-point grid, the smearing of the cut-off energy, the computation of a band structure, and again, convergence studies …