PM3 broken for Tl/Pb/Bi; also new SE parameters available

[radsci]

[crosspost from berlios]

Bug #12871, was updated on 2008-Jan-07 04:21
Here is a current snapshot of the bug.

Project: CP2K development
Category: None
Status: Open
Resolution: None
Bug Group: None
Priority: 5
Submitted by: radsci
Assigned to : jgh
Summary: PM3 broken for Tl/Pb/Bi; also new SE parameters available

Details: this was for 20070928 source; I didn't see mention of this in
the changelog.

Possible clues:  In PC-GAMESS, these elements also have problems for
PM3, but SCF energies work;
 gradients do not; the error is reported as PQN=6 not working with
STO-6G (principal quantum
 number??.)  mopac7 (via ghemical in knoppix5.1.1) doesn't have any
problem with these elements.

Also, there are new parameters for lanthanides in the mopac 7.1 source
(elements 57-71 as
 trivalent sparkles for AM1, Titanium in PM3, see openmopac.org) and
also RM1, a reparameterization of AM1.
  These look like they would be useful additions.  See:

lanthanides:
http://www.sparkle.pro.br/

RM1
http://www.rm1.sparkle.pro.br/

RM1 parameters:
http://www.mrw.interscience.wiley.com/suppmat/0192-8651/suppmat/2006/jws-jcc.20425.pdf

Thanks.


==================================

Follow-Ups:

===
Date: 2008-Jan-07 04:35
By: tlaino

Comment:
Hi Radsci (??),
two comments:

(1) please provide always an input file (that you think it may not
work)
(2) I would highly suggest to post your problem on the google mailing
list
(http://groups.google.com/group/cp2k). There's much more activity over
there than on berlios. Use
 that group also for possible bug reports. you may get a faster reply!

Regarding your specific problem is possible that for STO is not
available the expansion for n=6
 (Juerg can comment more on that).
Regarding lanthanides keep in mind that the semiempirical model in
cp2k allows only the use of s
 and p orbitals (no d or f at the moment!).
The model RM1 is already present in the latest "versions".

Hope this helps
Teo
-------------------------------------------------------
===




Date: 2008-Jan-08 00:00
By: radsci

Comment:
Hi Teo-

The lanthanide parameters are actually AM1/PM3 sparkles, all for
trivalent states, and don't use
 d- or f-orbitals.

Input for TlH2Me and output error message, also input for InH2Me which
works without problem
 follow.

I shall crosspost this to groups.google.

Thanks,
radsci.
================================
&GLOBAL
  PROGRAM Quickstep
  PROJECT TlH2Me0m1PM3gc10cs4
  RUN_TYPE GEO_OPT
&END GLOBAL

&FORCE_EVAL
  &DFT
    &QS
      METHOD PM3
    &END QS
    &SCF
      SCF_GUESS ATOMIC
    &END SCF
  &END DFT
  &SUBSYS
    &CELL
      ABC 20.0 20.0 20.0
      UNIT ANGSTROM
      PERIODIC NONE
    &END CELL
    &COORD
 Tl        0.31988        0.62776       -0.15078
  C       -0.26122       -0.63165       -0.19710
  H        1.56724        1.00197        0.30565
  H       -0.27708        1.86432       -0.01383
  H       -0.34193       -0.37415        0.96466
  H       -1.28841       -0.87660       -0.59022
  H        0.28152       -1.61164       -0.31837
    &END COORD
  &END SUBSYS
&END FORCE_EVAL


================================

 *****************************************************
 *** ERROR in FUNCTION exp_radius (MODULE qs_util) ***
 *****************************************************

 *** The Gaussian function exponent is zero ***


 ===== Routine Calling Stack =====

            3 qs_init_subsys
            2 quickstep_create_force_env
            1 CP2K

 CP2K| Stopped by process
number                                               0
 CP2K| Abnormal program termination


================================

&GLOBAL
  PROGRAM Quickstep
  PROJECT InH2Me0m1PM3gc10cs4
  RUN_TYPE GEO_OPT
&END GLOBAL

&FORCE_EVAL
  &DFT
    &QS
      METHOD PM3
    &END QS
    &SCF
      SCF_GUESS ATOMIC
    &END SCF
  &END DFT
  &SUBSYS
    &CELL
      ABC 20.0 20.0 20.0
      UNIT ANGSTROM
      PERIODIC NONE
    &END CELL
    &COORD
 In        0.31988        0.62776       -0.15078
  C       -0.26122       -0.63165       -0.19710
  H        1.56724        1.00197        0.30565
  H       -0.27708        1.86432       -0.01383
  H       -0.34193       -0.37415        0.96466
  H       -1.28841       -0.87660       -0.59022
  H        0.28152       -1.61164       -0.31837
    &END COORD
  &END SUBSYS
&END FORCE_EVAL




-------------------------------------------------------