[Pw_forum] example01 errors;conjugate-gradient algorithm;changing code functionals

[Axel Kohlmeyer]

On Mon, 21 May 2007, Helen wrote:

helen,

HE> I have recently downloaded PWscf and I have several questions.
HE> 
HE> 1. I have compiled PWscf and tried to run example01. During the running of 
HE> espresso-3.2/bin/pw.x <si.scf.cg.in>si.scf.cg.out
HE> 
HE> I obtain the error message 
HE> 
HE> forrtl: severe (174): SIGSEGV, segmentation fault occurred
HE> 
HE> Stack trace terminated abnormally.
HE> 
HE> The output file seems fine but I cannot run the band program (si.bands.out):

is it really complete? 
did you compare with the reference output in the results directory?

HE> from pp_check_file : error # 1
HE> 
HE> file /home2/helen/tmp/silicon.save not found
HE> 
HE> Any ideas about what is going wrong?

it might be that your compiler miscompiled the iotk library.
what platform (os/cpu/compiler/libraries) are you running on?

HE> 2. I would like to carry out an ion relax on an Si crystal. I see
HE> from INPUT_PW that the method is a quasi-Newton algorithm, is it
HE> possible to use a conjugate-gradient algorithm? Is the quasi-Newton
HE> algorithm efficient is the initial ion positions are not so close to
HE> the equilibrium structure?

the damped dynamics should be pretty efficient in 
that situation. you can switch to BFGS later.

HE> 3. I am exploring the possibility of running PWscf with modified
HE> functionals, specifically modifiying the exchange and correlation
HE> energy functionals to include a descreened two-body interaction. Is
HE> this possible? Which parts of the code include the functional forms?
HE> Which sub routines should be modified?

there are several places related to functionals. also pseudopotentials
have to be consistent with the functionals. i would start by looking
at the file Modules/functionals.f90 (which is pretty good commented) 
and then at flib/functionals.f90, flib/lsda_functionals.f90, and 
flib/more_functionals.f90 which provide the low level implementations.

cheers,
   axel.

HE> Thank you for your help, 
HE> Dr. Helen Eisenberg

-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.