Questions about the section Mixing in the XAS calculation

zh vale... at
Thu Jan 28 13:08:26 UTC 2010

1). In the manual of cp2k,  it mentions that there are a couple of
optional charge mixing methods for the XAS calculations. However, from
the source code,i.e., xas_tp_scf.F, it is found that only the
"DIRECT_P_MIXING" is indeed supported, as shown in followings
421        IF (scf_env%mixing_method.EQ.direct_mixing_nr) THEN
422           CALL scf_env_density_mixing(scf_env%p_mix_new,&
423                scf_env,scf_env%iter_delta,&
424                qs_env=qs_env,diis=diis_step, error=error)
425        ENDIF

Once the other mixing method (e.g., BROYDEN_MIXING, PULAY_MIXING) is
chosen in the self-consistent calculation of the ground state, the
self-consistent calculation for the excited state (e.g., DSCF, TP_HH
calculations) will stop in the first iteration step. This is observed
in my test calculations.  I want to confirm whether the above thing is
true or not.

2). If it is true, i.e., only the "DIRECT_P_MIXING" is indeed
supported for the self-consistent calculation of excited state for XAS
spectrum, how can we modify the source code to enable other mixing
methods for XAS calculations? Is it easy do such thing? Actually,
these charge mixing methods are already implemented in
"qs_gspace_mixing.F".  Wish someone  give  me some hints.

More information about the CP2K-user mailing list