Tips, news and FAQs.
VB2000, Version 2.7,
was released in 2013, and included in GAMESS(US) in 2014. Version 2.7
is still in the GAMESS(US) 2016 release, but version 2.8 is almost ready
for release. The purpose of this page is to collect
together everything we, the authors, and the VB2000 community have
learnt about using the program. Work of course is continuing on the
next release, and we will, at times, mention this. However, the main
thrust of this page is material that will help a user with the current version
We welcome items from users. Please e-mail them to Brian Duke
Tips and tricks
The following advice may help you to run VB2000
more effectively for your systems.
- If you want to run a multi-group VB calculation, specify the VB
groups explicitly using the "dot" notation, such as
VB(8).CASVB(6,6).VB(8). You can specify the same kind of calculation by
using GPF(ng) and other controls, but the "dot" notation is much
easier to use and intuitive.
- You can do a VB calculation of a molecule at one geometry and do
another VB calculation on the same molecule at a different geometry by
reading the VB orbitals of a previous calculation. This is not only a
good way to speed up convergence of a new calculation, but also a
necessary step for generating initial VB orbitals for a bond breaking
calculation in some cases.
- If a VB group has many resonance structures and you have difficulty
to determine which ones make major contributions, you can try a CASVB
calculation first and then check the biggest contributors.
- Try to avoid running CASVB or SCVB for groups with large number of
electrons (N>8 for CASVB and N>12 for SCVB - but SC(12,m) with m say 11
or 13 is more challenging). If you want to run SCVB(n) (n slightly larger
than 12), you can try to run a VB(n) for orbital optimization with one or a
few structures, and then do a SCVB(n) as a restart calculation without
further optimization of VB orbitals (using key word CIONLY).
- The spin-density calculation can be very time consuming for group
with more than 12 electrons.
- If VBSCF for VB orbital optimization becomes unstable, you can turn
on DIIS. See Chapter 6 for more details about key word DIIS. The DIIS
option helps to maintain the symmetry of wave function during VBSCF
- If the macro iteration is unstable (total energy becomes
oscillating, you can set a damping factor for rigid rotation. See
Chapter 6 for control $DAMPROT.
VB2000 on 64bit processors.
- All tests are using only one processor. The VB2000 codes do not support
- The Gamess/VB2000 version and the stand-alone version work fine with the
gfortran and intel compilers.
We have used earlier versions VB2000 on Itanium 64 bit machine, on a SGI Irix,
and on intel 64 bit processors with the Gamess TARGET=linux64
(using the g95 and pgf77 compilers). Thanks are due to Jörg
Saßmannshausen, Institut für Chemische Technologie von Materialien. TU-Graz,
Austria (now at University College, London) for allowing access to his
machines for one of these tests.
We have tested both Gaussian03 and Gaussian09 on intel 64 bit processors
at the Australian National Facility.
Particular thanks are due to Rika Kobayashii, as Gaussian09 involves
changes in the interface with VB2000. No such changes were involved in
moving from Gaussian98 to Gaussian03. These Gaussian versions, where the
VB2000 code is a separate Gaussian mode, is no longer supported.
New VB2000/Gaussian version.
From Gaussian09D1, where a large file of integrals and other value can
be outout, there is a stand-alone VB2000 Gaussian version that reads that
file. There is no longer the problem of compiling VB2000 into Gaussian
when most uses do not have the source code.
Geometry optimization with VB2000 is possible in two ways:-
Problems arise because VB2000 does not use symmetry. Method 2 can only use
symmetry for reading in the data. It then has to switch off symmetry.
Method 1 can use symmetry as you can define contraints on the variables.
However the geometry data has to be input using the Hilderbrand internal
coordinates and these are widely recognised to be a real pain. If you have to
use Gamess/VB2000 for geometry optimization on symmetric molecules, master
Hilderbrand internals. This was not made as clear as it could have been in
earlier documentation. Indeed, we have only recently realised how restrictive
symmetry is in this respect.
- Using Gamess/VB2000 with the TRUDGE facility.
- Using Gamess/VB2000 with numerical gradients.
Effective core potentials of with VB2000.
In case this is not clear from the Gamess/VB2000 test examples, the
use of efffective core potentials is quite straight forward. Any
potential that can be run with Gamess or Gaussian directly can be used with the
Parallization of VB2000.
First parallization tests are likely to be with the Gamess/VB2000
version but it is still a daunting task.
Last updated: 10 December, 2016.
Brian Salter-Duke (Brian Duke)