[CPMD-list] problem with geometry optimisation

Axel Kohlmeyer akohlmey at cmm.chem.upenn.edu
Mon Apr 23 17:05:49 CEST 2007


On Mon, 23 Apr 2007, Emmanuel Baribefe Naziga wrote:

EBN> Yes, I did look with VMD. I have attached the file.
EBN> Thanks

so essentially you are saying, you trust the 'wild' guess
that CPMD does to generate an initial hessian or a 
graphical visualization tool to determine bonding information 
more than your own knowledge of chemistry?

please consider how this 'bonding' information is generated.
both programs have tables of atom radii (and the list in VMD
is _very_ much gearde towards biomolecules and their naming
standards and assumes 1.0 everywhere else. unless you have 
a _very_ recent version, your platinum in VMD probably
has been detected as a phosphorus), and then when the sum of
those is smaller than the distance, a bond is assumed. 
a very rough measure.

a better way to validate your calculation is to compare
the bondlenghts and angles with published data (cis-platinum
is not exactly a new compound) and/or reference calculations
with a different software using equivalent methods (i.e.
DFT with the same functional).

you should also be aware that a wavefunction cutoff of
70 rydberg is at the _very_ lower end of accuracy for
the pseudopotentials that you are using. this is frequently
considered sufficient for MD since you never want absolute
accurate postitions but only a sampling of the relevant 
part of phase space (and the error due to limited sampling
is pretty big). with a geometry optimization, you usually
want more accuracy for that. i'm attaching my last test
input for your reference (which is very much on the 
higher accuracy side)

cheers,
  axel.

EBN> 
EBN> On 4/23/07, Ari P Seitsonen <ari.p.seitsonen at iki.fi> wrote:
EBN> >
EBN> > Dear Emmanuel,
EBN> >
EBN> >    Did you have a visual look on the geometry (GEOMETRY.xyz), does it look
EBN> > unphysical?
EBN> >
EBN> >    In case of doubt you can send it to us as well.
EBN> >
EBN> >      Greetings,
EBN> >
EBN> >         apsi
EBN> >
EBN> > -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
EBN> >    Ari P Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/
EBN> >    IMPMC, CNRS & Université Pierre et Marie Curie
EBN> >    Tel: +33-1-4427 7542, Fax: +33-1-4427 3785, GSM: +33-6-6736 3820
EBN> >
EBN> > On Mon, 23 Apr 2007, Emmanuel Baribefe Naziga wrote:
EBN> >
EBN> > > Thanks Axel,
EBN> > >
EBN> > > I got the following output from CPMD
EBN> > >
EBN> > >                           <<<<< ASSUMED BONDS >>>>>
EBN> > >    9 <-->  1    9 <-->  2    9 <-->  3   10 <-->  4   10 <-->  5
EBN> > >   10 <-->  6   11 <-->  7   11 <-->  8   11 <-->  9   11 <--> 10
EBN> > >
EBN> > >                           <<<<< HYDROGEN BONDS >>>>>
EBN> > >    7 <-->  5    8 <-->  1   11 <-->  1   11 <-->  2   11 <-->  3
EBN> > >   11 <-->  4   11 <-->  5   11 <-->  6
EBN> > > TOTAL NUMBER OF MOLECULAR STRUCTURES:  2
EBN> > >                             <<<<< MOLECULES >>>>>
EBN> > > MOLECULE:  1 <>       1    2    3
EBN> > > MOLECULE:  2 <>       4    5    6    7    8    9   10   11
EBN> > >
EBN> > > and I made the neccesary changes with LMAX
EBN> > >
EBN> > > Emmanuel
EBN> > >
EBN> > >
EBN> > > On 4/20/07, Axel Kohlmeyer <akohlmey at cmm.chem.upenn.edu> wrote:
EBN> > >> On Fri, 20 Apr 2007, Emmanuel Baribefe Naziga wrote:
EBN> > >>
EBN> > >> EBN> Thanks Apsi,
EBN> > >> EBN>
EBN> > >> EBN> I did like you suggested but the problem still persists .
EBN> > >>
EBN> > >> please tell us how you decided that there are two molecules.
EBN> > >>
EBN> > >> i just ran your input (with the changes ari mentioned, as
EBN> > >> they are needed) on a test machine and it looks ok to me.
EBN> > >> also the result is consistent with goedecker pseudopotentials
EBN> > >> (where the selection of local potential is encoded in the file).
EBN> > >>
EBN> > >> finally, version 3.12 is not a released version of CPMD,
EBN> > >> but the version number that is used for the current development
EBN> > >> tree. please run with an official version.
EBN> > >>
EBN> > >>
EBN> > >> greetings from sunny (finally!!) philadelphia,
EBN> > >>
EBN> > >>    axel.
EBN> > >>
EBN> > >> EBN>
EBN> > >> EBN> Emmanuel
EBN> > >> EBN>
EBN> > >>
EBN> > >> --
EBN> > >> =======================================================================
EBN> > >> Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
EBN> > >>    Center for Molecular Modeling   --   University of Pennsylvania
EBN> > >> Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
EBN> > >> tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
EBN> > >> =======================================================================
EBN> > >> If you make something idiot-proof, the universe creates a better idiot.
EBN> > >>
EBN> > >>
EBN> > > _______________________________________________
EBN> > > CPMD-list mailing list
EBN> > > CPMD-list at cpmd.org
EBN> > > http://cpmd.org/mailman/listinfo/cpmd-list
EBN> > >
EBN> 

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

-------------- next part --------------
&INFO
Gas phase: geometry optimisation of cisplatin
&END

&CPMD
  OPTIMIZE GEOMETRY XYZ
  RESTART WAVEFUNCTION COORDINATES LATEST

   PCG MINIMIZE

  REAL SPACE WFN KEEP

   CONVERGENCE INITIAL
    1.0e-4
   CONVERGENCE GEOMETRY
     1.0E-4
   CONVERGENCE ORBITAL
     1.0E-6
   CONVERGENCE ADAPT
     0.02
   CONVERGENCE ENERGY
     0.05
   LBFGS NREM
     40
   MAXSTEP
     2500
   COMPRESS WRITE32
   PRINT LSCAL ON
&END

&SYSTEM
  ANGSTROM
  SYMMETRY
  0
  CELL
  15. 1. 1. 0. 0. 0.
  CUTOFF
  120.0
&END

&ATOMS

*H_MT_BLYP
   LMAX=S
   6
   2.354418516159      0.019093690440     -0.064243108034
   1.411824822426     -1.303389430046     -0.234962448478
   1.443768382072     -0.502316892147      1.187707781792
  -1.643410801888     -1.234962463379     -0.349684238434
  -2.232093095779     -0.069135926664      0.630641579628
  -0.998527586460     -1.013527393341      1.133388400078

*Cl_MT_BLYP.psp KLEINMAN-BYLANDER
   LMAX=D
   2
  -1.733747005463      2.059042453766     -1.119829893112
   1.383087396622      2.222492933273     -1.282287359238

*N_MT_BLYP.psp KLEINMAN-BYLANDER
   LMAX=P
   2
   1.485479116440     -0.403016984463      0.193374127150
  -1.406602025032     -0.555594444275      0.344115614891

*PT_MT_GIA_BLYP KLEINMAN-BYLANDER
   LMAX=D LOC=S
   1
   -0.064197830856      0.781314134598     -0.438224017620
&END

&DFT
 NEWCODE
 FUNCTIONAL BLYP
 GC-CUTOFF
 0.1E-05
&END


More information about the CPMD-list mailing list