[Pw_forum] Is the "cell_dofree" functionality fully implemented in Espresso 4.0.x?

[Serge Nakhmanson]

Hi Lorenzo,

thank you very much for your comment, I do appreciate it and it does help!

Fixing this is not a problem. I already did it VASP-style, which requires
compiling a separate binary for every particular cell-relaxation constraint.
Granted, this is much less elegant than using the "cell_dofree" switch
but it gets the job done.

BTW, I have a tiny and very conservative example that I used to study the
"cell_dofree" functionality, which converges in around 20 damped-dynamics
steps (and then sits around doing nothing until the limit of ionic steps
is reached). I can send it to you privately if you are interested and willing
to play with it.

Cheers,

Serge

Lorenzo Paulatto wrote:
> Serge Nakhmanson ha scritto:
>> Dear All,
>>
>> Is "cell_dofree" feature in &CELL completely implemented
>> in Espresso 4.0.x?
> Dear Serge,
> I can't give you a definitive answer, as I haven't really used this
> feature. What I can say is that it is not implemented with bfgs
> algorithm and even with damp I'm not really sure it is working. I tried
> to find a quick fix some time ago, but had to give up due to other work.
> 
>> condition in vcsmd.f90 is never reached. Or am I missing
>> something?
>>   
> I think you are right, but it is easy to fix, change
>   ABS( sigma(i,j) ) * uakbar
> to
>   ABS( sigma(i,j) ) * iforceh(i,j) * uakbar
> in both places (be careful with the indexes), and it *should* work.
> 
> I say "should" because I have done a few tries, but couldn't make it
> converge; it is probably due to my inexperience with damped dynamics.
> 
>> Would appreciate any sage comments on this.
>>   
> I''m not really a sage, I hope it helped
> 
> bye
> 


-- 
*********************************************************
  Serge M. Nakhmanson               phone: (630) 252-5205
  Assistant Scientist                 fax: (630) 252-4798
  MSD-212, Rm. C-224
  Argonne National Laboratory
  9700 S. Cass Ave.
  Argonne, IL 60439
*********************************************************