[Pw_forum] convergence of pw.x

[Konzern]

Hi everyone,

Merry Christmas and Happy New Year to you.

But I have a problem. I am using pwscf (version 2.1.5) to calculate the
phonon of a slab.
In the scf calculations, I set the parameter of conv_thr as 1.0d-12, and so
in the nscf where
calculation='phonon'. The scf converged successfully, while the nscf just
could not converge,
there was always some " xx eigenvalues not converged." So I have to increase
the conv_thr
of nscf run to 1.0d-10 or so, and this time, it could converge finally,
although there was still
some "yy eigenvalues not converged" ( yy=1 or 2) duing the calculation.

I don't know what is wrong with this. In my expectation, the nscf should be
able to converge
easily based on the previous scf run, at least, to the save accuracy level.
Could you tell to
what extent will this affect the resultant frequencies? i.e., can I still
believe in the resultant
frequencies? The threshhold I used in ph.x is 1.0d-14; I tried to use
conv_thr=1.0d-10 in both
scf and nscf while tr2_ph=1.0d-12 in ph.x, the resultant frequencies were
"wrong" ( too many
imaginary ones).

Thanks.

Konzern
-------------- next part --------------
An HTML attachment was scrubbed...
URL: /pipermail/attachments/20051223/9985e9ef/attachment.htm