G92. Core energies. Summary.
Dear netters,
Some time ago, I posted
a question about
"how to extract from G92
the core energies H_nn,
and, perhaps, the Coulomb
integrals J_nm and exchange
integrals K_nm?"
The only answer came from:
ALBERTO GOBBI,
gobbi.,at,.sg1501.chemie.uni-marburg.de,
Bantzerstr. 9, 137.248.151.1,
D-35039 Marburg,
Tel. 8.00-18.00: 0049 6421 287001,
Fax: 0049 6421 285547.
Many many thanks to him !
I first summarize about core
energies H_nm (which is successful)
and then about J_nm and K_nm
which was not successful.
IMPORTANT:
I am reporting ONLY what
I tried myself. No one else
but me is responsible for my failure
with 2-e integrals.
The Gaussian92 Programmer's Guide
is abbreviated here as G92PG.
0) General advices.
^^^^^^^^^^^^^^^^^^^
--This sort of job requires
some extent of Non-Standard Routes
and Options. Look at section 4.4
of G92PG starting from page 63.
--It is worth to look at *.log
file of your job to see what links
were executed. Then it is VERY useful
to read in the source codes of those
links the IOp() parameters meaning
and the meaning of the links themselves
- you'll learn a lot.
1) Core energies H_nm.
^^^^^^^^^^^^^^^^^^^^^^
To get them from G92
to your *.log file
put the following option
into the input:
IOp(3/33=1)
along with your other input.
This causes the matrices of
kinetic energy, potential energy and
of core energies to be printed.
These 3 matrices come together always.
One can get rid of multipoles
matrices and Fermi contact interaction
matrices (which also appear together
with those 3 matrices) adding
to the above option
the following one:
IOp(3/36=-1)
For description of these options
look at page 568 of G92PG.
2) Coulomb integrals J_nm
^^^^^^^^^^^^^^^^^^^^^^^^^^
and exchange integrals K_nm.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
This was not successful.
After many trials I cannot get
them yet.
The idea was to put into the input
extralinks=l316
This option is supposed to give
2-e integrals because this link is
exactly for "dumping 2-e integrals"
as it follows from comments to its
source FORTRAN code.
For description of extralinks
look at page 64 of G92PG.
After putting this extralink,
I got many-many numbers which
I failed to exactly identify
(may be someone can do this for us?!).
And then error message appeared and G92 terminated
abnormally.
What did NOT help to improve
the situation (including various
combinations of options below):
-- IOp(3/33=4)
-- IOp(3/33=5)
-- IOp(3/33=6),
[look at page 568 of G92PG for meaning
of these IOp's]
-- NoSymm [in *.log file this
only changed some wording]
-- NoRaff [the same as about NoSymm]
-- IOp(3/34=2)
-- IOp(3/34=3)
[page 568 of G92PG]
-- extralinks=(311,316)
[look at the source codes of
those links]
That's all I can report about
what I myself tried along the advice
given to me.
So, the core energies are okay
and 2-e integrals are NOT.
Would very much appreciate
further development of this discussion
and hints where I've made errors.
Very interesting advice came
from help.,at,.gaussian.com:
"Since I don't know how big your problems are
I suggested l804 which has the integrals you
requested produced without disk storage and
you can start by modifying the output routines
to identify the ones you need
and discard the rest.
Doug Fox."
I copied the L804 source code and looked at it.
Considering my time schedule,
it does not seem feasible for me
to revise this code in the nearest future.
May be someone else could do it
quicker and more professionally.
Sincerely,
--------------------------------------------------------
--------------
Aiaz BAKASSOV (English: Ayaz BAKASOV) | Phone: +41 1
632 79 18
Laboratorium fuer Phys. Chemie | FAX: +41 1
632 10 21
ETH-Zuerich (Zentrum) | E-mail:
aiba.,at,.debye.vmsmail.ethz.ch
CH-8092 Zurich, Switzerland |
aiba.,at,.ir.lpc.ethz.ch
--------------------------------------------------------
--------------