[CPMD-list] problems on atomic charge calculations

[Axel Kohlmeyer]

On Mon, 6 Aug 2007, Hui-Hsu Gavin Tsai wrote:

GT> Dear CPMD users,

dear gavin,
  
GT> We followed the CPMD tutorials to calculate the charges from the converged 
GT> wavefunction. Unfortunately, the calculation failed.

[...]

GT>  RESTART INFORMATION READ ON FILE                       ./RESTART
GT>  ***     PHFAC| SIZE OF THE PROGRAM IS   38536/ 140680 kBYTES ***
GT>  ESP CHARGES| NUMBER OF FITTING POINTS                   ********

ok, so you have an overflow in the number of fitting points.
for ESP derived charges.

GT>  MEMORY| MEMORY REQUIRED:     -4409758  WORDS
GT>  
GT> PROGRAM STOPS IN SUBROUTINE MEMORY| NEGATIVE OR TOO BIG VALUE [PROC=   0]
GT> p0_4043:  p4_error: : 999

 
GT> However, our test system is small with one water molecule only and
GT> our computer system has 4GB memory/node installed. It looks very

you may have 4GB/node, but with 32-bit integers, CPMD can maximally 
allocate 2GB at a time. however, the problem lies most likely elsewhere.

GT> unlikely that the 4GB memory is not enough to perform the charge
GT> calculation on one water molecule. More interestingly, we have no
GT> problemto calculate other properties from the wavefunction.

GT> We had tried different options, but still failed. Thereofore, we

which options?

GT> sent this email out to ask for your help.

GT> Is there anyone has done the same calculations before and has
GT> solution to fix this problem?

this topic has come up and should be documented in the mailing list
archives. for ESP derived charges, it is imperative that the
charge distribution is centered in the unit cell and that there is
enough space around it. otherwise the estimate of the number of
fitting point does not converge, or - as in your case - it ridiculously
large.

GT> Any comment is welcome. 

GT> The input files along with the error messages from the output file are 
GT> listed at the end of this email.
GT>  
GT>  
GT> Many Thanks.
GT>  
GT> Gavin Tsai
GT> Department of Chemistry
GT> National Central University
GT> Taiwan
GT>  
GT> Input Files:
GT>  
GT> &CPMD
GT> OPTIMIZE WAVEFUNCTION
GT> STRUCTURE BONDS ANGLES
GT> RHOOUT

please at a CENTER MOLECULE here

GT> &END
GT>  
GT> &DFT
GT> FUNCTIONAL PBE
GT> &END
GT>  
GT> &SYSTEM
GT> SYMMETRY
GT> 1

and/or use SYMMETRY 0 here.

GT> CELL
GT> 15.0000 1.0 1.0 0 0 0
GT> CUTOFF
GT> 70.0
GT> &END
GT>  
GT> &ATOMS
GT> *O_MT_PBE.psp KLEINMAN-BYLANDER
GT> LMAX=P
GT> 1
GT> 0.0000 -0.1267 0.0000
GT> *H_MT_PBE.psp
GT> LMAX=S
GT> 2
GT> 0.0000 1.0058 1.4524
GT> 0.0000 1.0058 -1.4524

as you can see, your molecule crosses the
box boundary and this is not conformant with
the requirement for ESP charges. other than
that, you can also turn off the calculation
of ESP derived charges (if you are not interested
in them, that is) by commenting out the corresponding
code in the source code.

cheers,
   axel.

GT> &END
GT>  
GT> ===========
GT> &INFO
GT> charge
GT> &END
GT>  
GT> &CPMD
GT> PROPERTIES
GT> RESTART WAVEFUNCTION COORDINATES
GT> &END
[...]

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