ABINIT Discussion Forums

Hello to all,
I am trying to include a new feature in ABINIT (LOTF: learn on the fly, another method to compute MD steps).
This method was developed some time ago for an old version of ABINIT (5.4) but never committed.
Now I have to put all the existent code in a new version (7.0) and I have some doubts.

In version 5.4, the essential of the LOTF code consisted to put a new method in src/21drive/moldyn.F90 when the variable ionmov was equal to 14.
As in the recent versions the subroutine moldyn.F90 has disappeared, I am asking to myself if the subroutine src/95drive/move.F90 is the descent of
moldyn.F90 or not.
Is it correct?

What I can understand from the code is that I have to create a new subroutine src/45_geomoptim/pred_loft.F90 containing the new method and then
call it in src/95drive/move.F90 when ionmov has the corresponding value (I think to set it to 31, following the present list).
Is it all correct?

thanks.
Best regards,
Marcp

Hi Marco,

I hope you got some answers in the meanwhile or at the devel conference. You are correct about the pred_XXX routines - they are now all factored out with a (semi) standard interface of positions in and out.

ciao!

Matthieu

mverstra wrote:Hi Marco,

I hope you got some answers in the meanwhile or at the devel conference. You are correct about the pred_XXX routines - they are now all factored out with a (semi) standard interface of positions in and out.

ciao!

Matthieu

Hello Matthieu,
Yes thanks, actually I wrote directly to the developer of this part (G.A.F.) which gave me a precise description of the move.F90 subroutine and the pred_xxx method.

ciao
Marco