[CPMD-list] orthorhombic space groups !!

[Alessandro Curioni]

 Imad,

   this is a  piece of the code that is  not used a lot and therefore not 
well documented -
   the symmetry module require a  precise  reference systems   e.g. choice 
of the principal axis and  relative convention that in many cases do not 
follow
   what Acelrys program give as default (that at the same time  in many 
occasion are not obvious) -   moreover  they are handled with separate 
modules  for
   crystal  and   isolated systems.    The code doesn't reorient    the 
system trying to find its symmetry group  ( e.g. with point group auto) - 
   but simply try to assign it following the  internal convention -  the 
same it is true if  you use symmetrize coordinates etc. -
   In other words if the code expect to have your molecule oriented in a 
way the C2 axis is  along  the z axis and with the origin in zero  -  you 
have to provide the initial structure following the same convention .... 
if not you will have a lot of problems.....

 We  have used  coordinates and forces symmetryzation many many times with 
success ( both for molecules and solid)  - 
 
 but I completely agree the  choice of reference system  is not documented 
 and indeed  it should  it be  to make life easier  ( time permitting we 
will do it)
  in the meantime  please have a look at molsym.F and symmetry.F-


Alessandro



 






Łukasz Walewski <ljw at icm.edu.pl> 
Sent by: cpmd-list-bounces at cpmd.org
01/24/2008 10:31 AM

To
Belabbas Imad <belabbas_imad at yahoo.fr>
cc
cpmd-list at cpmd.org
Subject
Re: [CPMD-list] orthorhombic space groups !!






Dear Imad,

I can give you a brief view based on my experience with isolated organic
molecules. They have some symmetry elements in the stationary points of
the energy surface. In the electronic ground state I did not use
symmetry at all, because optimizations converged very well. When I went
to excited states it turned out that the gradients in the vicinity of
the minimum are very small and the optimization was stopping before
reaching the real minimum (confirmed by vibrational analysis). I tried
to impose loose symmetry constraints to converge the geometry but the
symmetry module did not work for me at all. Optimization with CS
symmetry went to a geometry obviously lacking this symmetry element not
speaking of the higher symmetries like C2H which ended in floating
exceptions. The 'SYMMETRIZE COORDINATES' did not work for me either. I
wrote a simple program to symmetrize the coordinates myself and after
restarting optimization the geometry converged very well.

Axel, could you advise where to look for the symmetry routines in the
code ? Maybe I would have a look at that in my Free Time.

Regards,
Lukasz


Dnia 23-01-2008, śro o godzinie 20:22 +0100, Belabbas Imad pisze:
> Dear CPMD users and symmetry experts,
> 
> I am trying to treat a face centered orthorhombic molecular crystal
> (Beta-Acetonitrile). I have constructed a conventional unit cell by
> using the symmetry elements of the corresponding space group (Cmc21)
> taken from material studio. I have checked that the point group
> corresponding to my crystal is mm2 (C2v). 
> 
> I have made 2 tests:
> 
> * In the first test I have asked CPMD to determine automatically the
> point group.
> 
> * In the second test I have fixed the point group to C2v.
> 
> 
> In the first case the code detects that there is 4 symmetry elements
> and the unit cell is not primitive (which looks fine !!). But it finds
> a point group: C2h !!!!! which, as far as I know is characteristic of
> a monoclinic system !!!. Moreover, in the supercell section it tells
> that the symmetry is Orthorhombic and the point group is C2h !!???. I
> have checked in the international tables of Crystallography, there is
> no Orthorhombic crystal that has a C2h point group !!!. Otherwise, in
> the KPTS_GENERATION file, CPMD uses the C2h point group !!
> 
> In the second case, where I forced the point group to be C2v, in the
> supercell section, the point group is mentioned as mm2 (C2v). However,
> in the KPTS_GENERATION file, CPMD usese the C2h point group !!
> 
> One more thing, CPMD recognizes an Orthorhombic unit cell if the cell
> parameters have the form [a b c 90 90 90] (absolute, degree). this
> constrains the user, in the case of  a face-centered Orthorhombic, to
> use a conventional unit cell. Indeed, the  primitive unit cell has
> lattice parameters like: [a a c alpha 90 90] (this looks like a
> monoclinic unit cell).
> I am just wondering if this can explains why I am getting a monoclinic
> point group for my system !!!
> 
> Can someone comment on that !!???
> 
> Ps. I am attaching the input an output files of the two tests that I
> have performed + the CIF file of the crystal that I am treating !! 
> 
> 
> 
> 
> ______________________________________________________________________
> Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers
> Yahoo! Mail 
> _______________________________________________
> CPMD-list mailing list
> CPMD-list at cpmd.org
> http://cpmd.org/mailman/listinfo/cpmd-list

_______________________________________________
CPMD-list mailing list
CPMD-list at cpmd.org
http://cpmd.org/mailman/listinfo/cpmd-list

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://cpmd.org/pipermail/cpmd-list/attachments/20080124/b81a526a/attachment.html