[CPMD-list] problems with point group recognition and symmetrisation of coordinates

[Christian Tuma]

Dear CPMD developers,

using CPMD 3.7.2 (Linux-PC, Redhat 9.0, ifort 8.0, ATLAS 3.6.0) I encountered
problems with the recognition of molecular point groups and subsequent
symmetrisation of coordinates.
For the case of the methane molecule (CH4, symmetry point group "Td") I used
two different orientations of the molecule in a cartesian coordinate system.
Depending on which orientation I have chosen, I got quite different results
concerning the recognition of a number of subgroups of "Td" and "Td" itself: 

A) one C2-axis corresponds to the z-axis, and the x-axis is orthogonal to
   a mirror plane

   ...
   &ATOMS
   *TM/C_TM_PBE_pade_110 KLEINMAN-BYLANDER
    LMAX=P
    1
     0.0000    0.0000    0.0000
   *TM/H_TM_PBE_pade_070 KLEINMAN-BYLANDER
    LMAX=S
    4
     0.0035   -0.8952    0.6336
    -0.8988   -0.0005   -0.6284
     0.0031    0.8958    0.6327
     0.8920    0.0000   -0.6379
   &END
   ...

   C2, C2v : recognition and symmetrisation ok
   S4      : recognition ok, symmetrisation problem "ATOMS ARE VERY CLOSE"
   Td      : recognition ok, symmetrisation problems
                            DELTA=0.01 : bonds far too short
                            DELTA=0.1  : loss of 4 symmetry operations
   D2d, D2 : not recognised


B) all C2-axes correspond to cartesian axes

   ...
   &ATOMS
   *TM/C_TM_PBE_pade_110 KLEINMAN-BYLANDER
    LMAX=P
    1
     0.0000    0.0000    0.0000
   *TM/H_TM_PBE_pade_070 KLEINMAN-BYLANDER
    LMAX=S
    4
     0.6355   -0.6304    0.6336
    -0.6352   -0.6359   -0.6286
    -0.6312    0.6356    0.6327
     0.6307    0.6308   -0.6379
   &END
   ...

   C2, D2  : recognition and symmetrisation ok
   C2v, Td : not recognised
   S4      : recognition ok, symmetrisation problem "ATOMS ARE VERY CLOSE"
   D2d     : recognition ok, symmetrisation problems (bonds far too short)


The rest of the input was as follows (with different Schoenfliess symbols):

   ! CH4
   &CPMD
    OPTIMIZE GEOMETRY
    MAXSTEP
     1
    STRUCTURE BONDS ANGLES
   &END
   &SYSTEM
    ANGSTROM
    POINT GROUP MOLECULE DELTA=0.01
    DND  2
    sYMMETRIZE COORDINATES
    SYMMETRY
     0
    CUTOFF
     50
    CELL
     5.0 1.0 1.0 0.0 0.0 0.0
   &END
   &ATOMS
   ...
   &END
   &DFT
    NEWCODE
    FUNCTIONAL PBE
   &END


These observations for (A) and (B) are somehow complementary, especially
concerning the "D" point subgroups. That's why I think there is no way to
put the CH4-molecule such that everything goes fine with all these symmetry
point subgroups.
So what's wrong? I took a look into the source code, but at some point I
was unable to follow the programmers' ideas. My only vague idea is that
the generation of the symmetry operators might not be sufficiently flexible,
i.e. some of them could be excluded from each other. Do the CPMD developers
know where the problem is hidden and how it can be fixed?

Thanks a lot,

Christian Tuma.

PS: Another example for the failure in symmetry recognition is the case
    of the staggered ethane molecule C2H6. Although being aligned along
    the z-axis, CPMD doesn't recognise it as "D3d".

-- 
Christian Tuma             Humboldt-Universitaet Berlin
ct at chemie.hu-berlin.de     Arbeitsgruppe Quantenchemie (Prof. Sauer)
phone: +49-30-20937140     Brook-Taylor-Str. 2, 12489 Berlin, GERMANY
fax:   +49-30-20937136     http://www.chemie.hu-berlin.de/ag_sauer