basis set limitations

Posted by Lori Burns on April 18, 2005 at 20:12:10:

Hi,
I am running Molcas 5.4 on a 64-bit SGI workstation and am encountering a confounding basis set limitation. When performing a CASSCF optimization on a moderate-sized (15 atoms, 9 heavy) molecule with C_2v symmetry, the calculation runs the seward and scf modules and writes a non-zero INPORB file. Upon entering the rasscf module, some introductory information is printed, then the program fails with:

The MO-coefficients are taken from the file INPORB
title: SCF orbitals
TEST IN ORTHO 87
0.000000 0.000000 0.000000 0.000000 0.000000 0.000000
*** omitting many zeros - total of 174 zeros ***
0.000000 0.000000 0.000000 0.000000 0.000000 0.000000
Error in ORTHO2. Norm= 0.0000000000000000E+00
--- Stop Module: rasscf at Fri Apr 15 21:03:23 EDT 2005 /rc= 16 ---

Occurrences of this error are strongly correlated to the basis set employed. Built-in bases (sto-3g, cc-pvtz) uniformly run properly and do not give this error. Molcas 6.0 bases downloaded and installed in the molcas54/basis_library directory run for smaller systems (aug-cc-pvtz on formaldehyde) yet fail for the one of interest (aug-cc-pvdz on tropolone). Bases downloaded from the EMSL library and specified inline fail if larger than 6-311++G**. This would all seem to point to a limitation in the number of allowed basis functions, but then why does 6-311++G** (369 primitive / 240 contracted) work, aug-cc-pvDz (381 / 261) fail, and cc-pvTz (474/354) work? Any illumination on this matter would be greatly appreciated.

Thanks in advance,
Lori Burns

Follow Ups:

Re: basis set limitations Björn Roos 09:03:35 4/19/05 (0)
Re: basis set limitations Valera Veryazov 21:14:27 4/18/05 (1)
- Re: basis set limitations Lori Burns 08:41:38 4/25/05 (0)

Post a Followup

Name:
E-Mail:

Subject:

if B is 1s22s22p1, what is Li?

Passfield:

Comments:
: Hi, : I am running Molcas 5.4 on a 64-bit SGI workstation and am encountering a confounding basis set limitation. When performing a CASSCF optimization on a moderate-sized (15 atoms, 9 heavy) molecule with C_2v symmetry, the calculation runs the seward and scf modules and writes a non-zero INPORB file. Upon entering the rasscf module, some introductory information is printed, then the program fails with: : The MO-coefficients are taken from the file INPORB : title: SCF orbitals : TEST IN ORTHO 87 : 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 : *** omitting many zeros - total of 174 zeros *** : 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 : Error in ORTHO2. Norm= 0.0000000000000000E+00 : --- Stop Module: rasscf at Fri Apr 15 21:03:23 EDT 2005 /rc= 16 --- : : Occurrences of this error are strongly correlated to the basis set employed. Built-in bases (sto-3g, cc-pvtz) uniformly run properly and do not give this error. Molcas 6.0 bases downloaded and installed in the molcas54/basis_library directory run for smaller systems (aug-cc-pvtz on formaldehyde) yet fail for the one of interest (aug-cc-pvdz on tropolone). Bases downloaded from the EMSL library and specified inline fail if larger than 6-311++G**. This would all seem to point to a limitation in the number of allowed basis functions, but then why does 6-311++G** (369 primitive / 240 contracted) work, aug-cc-pvDz (381 / 261) fail, and cc-pvTz (474/354) work? Any illumination on this matter would be greatly appreciated. : Thanks in advance, : Lori Burns

[ Follow Ups ] [ Post Followup ] [ Molcas user's WWWBoard ]