[Pw_forum] Using wfc with scf after vc-relax

Axel Kohlmeyer akohlmey at cmm.chem.upenn.edu
Mon Jan 19 17:05:35 CET 2009


On Mon, 19 Jan 2009, Marino Vetuschi Zuccolini wrote:

MVZ> Hello to all users,

hello marino,

MVZ> I'm using QE 4.0.4 with mpi on my macosx.
 
MVZ> I relaxed successfully a structure with variable-cell method obtaining  
MVZ> a minimum with a diagonalized matrix of stresses.
MVZ> 
MVZ> Interested in the postprocessing of wfc I used the files written at  
MVZ> the end of the relaxation with projwfc or pp, but it seems that the  
MVZ> wfc are not readable by those tools with the well known error messages  
MVZ> found in the FAQ when cutoff of mesh nodes are not of the same size  
MVZ> (dimensions do not match ..).

that is most likely due to mistakes in your input files.
hard to tell without seeing them, though.

MVZ> Then, computing a single scf point with the equilibium (!) geometries  
MVZ> of the converged vc-relax job, this geometries is not recognized as  
MVZ> minimum, with a mean of the trace of stress eclose to 87 kbars,  
MVZ> appearing mechanically unstable.

again, this is most likely due to a mistake in the input.
have you checked for convergence of the stress tensor
wrt. cutoffs in your vc-relax calculation. please note, 
that when the box size changes this also changes the 
"effective" cutoff, so you have to be _well_ converged
or you get significant pulay stress. using vc-relax is
convenient, but has one serious drawback: you run at
a constant number of plane waves with a fixed grid 
and not at a constant cutoff (there is a method available
to work around it to some degree). but in general, the
requirements for convergence are much higher than for
regular relaxation or single point calculations.

also, when you do a single point calculation after vc-relax 
and your cell has changed, so that a different FFT grid
would fit into the cell, it will be used by default and 
if you are not well converged wrt stree, it will show
as a significant difference. 

cheers,
   axel.

MVZ> Then the question is why vc-relax generate an odd wfc useless to other  
MVZ> tools and the geometries of vc-relax are not computed as equilibrium  
MVZ> geometries by scf run? ..or why I'm in this strange situation...
MVZ> 
MVZ> I apologize if the questions hides lack of knowledge of mine.
MVZ> 
MVZ> Thanks to all in advance
MVZ> 
MVZ> m.
MVZ> 
MVZ> 
MVZ> *******************************************************
MVZ> Marino Vetuschi Zuccolini
MVZ> zucco at dipteris.unige.it
MVZ> Researcher / Geochemist
MVZ> Laboratory of Geochemistry
MVZ> 
MVZ> DIPartimento per lo studio della TErra e delle sue RISorse -  
MVZ> Università  di Genova
MVZ> Tel. ++39 010 3538136 Fax. ++39 010 352169
MVZ> Corso Europa 26, 16132 - Genova - Italy
MVZ> 
MVZ> 
MVZ> _______________________________________________
MVZ> Pw_forum mailing list
MVZ> Pw_forum at pwscf.org
MVZ> http://www.democritos.it/mailman/listinfo/pw_forum
MVZ> 

-- 
=======================================================================
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.


More information about the users mailing list