The ABINIT Users Mailing List ( CLOSED )

[Alexey Dick <dick@fhi-berlin.mpg.de>]

Dear Don,

I thank you very much indeed for your efforts solving the problem. It 
was very kind of you. In fact, the way you were chasing the error-source 
is a spectacular example how one should try to localize problems. I have 
learnt quite something from our communication. Thanks again,

With best,
Alexey

> Dear Alexey & all,
>
> As speculated in my last posting in this string, setting optnlxccc = 2 
> does resolve the problem.  The attached summary shows that this 
> reduces the discrepancy between the Murnaghan-fit acell and the value 
> obtained with built-in optimization to 0.04%. The discrepancy is 1.6% 
> with the default value optnlxccc = 1.  The 3rd-derivative core-charge 
> continuity apparently present in Alexey's FHI pseudopotential is 
> clearly adequate to take advantage of this option.  The option became 
> available in 4.4.*.
>
> Don Hamann