[Pw_forum] vc_relax convergence criteria

Paolo Giannozzi giannozz at democritos.it
Wed Jan 7 09:44:24 CET 2009


Timothy Mason wrote:

> does anyone see value in having consistent convergence 
> VC-relax criteria for future releases?

of course everybody sees value in it. The relevant question is however
a different one: "is anybody 1) interested in implementing consistent 
criteria for convergence in VC-relax, and 2) able to do it?"
Note that the intersection of set 1) and set 2) might be the empty set.

> the documentation has definitions for what force_conv_thresh 

"forc_conv_thr"

> and etot_conv_thr mean. Do these definitions not apply to bfgs?

they should, since both are passed to the bfgs routine
(see Modules/bfgs_module.f90). You can check this quite easily:
locate the lines after this one
       ! ... convergence is checked here
It seems to me that they are used for damped dynamics as well
(routine PW/vcsmd.f90; locate line
   ! ... check if convergence for structural minimization is achieved
). Note that there are other "damped dynamics" in the code

Paolo
-- 
Paolo Giannozzi, Democritos and University of Udine, Italy



More information about the users mailing list