FOCK matrix summary.
Hello.
Answers to the questions (shown below) that I have recently posed on the CCL
are as follows:
Concerning Question 1.
The code that I wrote to print out the lower triangular portion of the overlap,
and fock matrices (alpha and beta spins) is incorrect in the sense that in G94
the modulus of the matrix element must be greater than a threshold value that
is set to Thresh = GFloat(10)**(Key-6) where Key is in most cases 0 (See LTOut
in utilam.F [line 47349]). The corrected code is as follows:
---------------------------------------------------
NTT = Nbasis*(NBasis+1)/2
Zero = GFloat(0)
Thresh = GFloat(10)**(-6)
Call FileIO(2,-IRwSao,NTT,V,0)
do 211 i=1,NTT
If(Abs(V(i)).lt.Thresh) V(i) = Zero
211 write (IOvlp,2010) V(i)
C
Call FileIO(2,-IRwFA,NTT,V,0)
do 212 i=1,NTT
If(Abs(V(i)).lt.Thresh) V(i) = Zero
212 write (IUnitFA,2010) V(i)
C
Call FileIO(2,-IRwFB,NTT,V,0)
do 213 i=1,NTT
If(Abs(V(i)).lt.Thresh) V(i) = Zero
213 write (IUnitFB,2010) V(i)
--------------------------------------------------
I thank Professor Vincenzo Baronem who pointed this out.
As to Question 2)
I simply calculated:
FC = eSC
using the output generated by my FORTRAN link in G94. The left and right hand
sides of the above equation were equivalent.
Prof. Fink and Prof. Stuchebrukhov of UC Davis both pointed out this simple
test.
Thanks for all those that responded to my queries.
Iraj.
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Question 1)
---
Hello.
I used the following to print out the FOCK (alpha, beta) matrix from Gaussian
94:
NTT = Nbasis*(NBasis+1)/2
Call FileIO(2,-IRwFA,NTT,V,0)
do 212 i=1,NTT
212 write (IUnitFA,2010) V(i)
C
Call FileIO(2,-IRwFB,NTT,V,0)
do 213 i=1,NTT
213 write (IUnitFB,2010) V(i)
Interestingly, when I compared the values generated from this method to the
standard Gaussian output; viz., IOp(5/33=5); I found that my numbers were
slightly different from theirs.
Input Deck:
-------------------------------------------------------------------------------
# HF/sto-2g 6d 10f field=z+1000000
geom=coord pop=regular IOp(5/33=5)
currents: H2 1+ ion , uhf
1 2
H 0.0 0.0 -4.0
He 0.0 0.0 0.0
H 0.0 0.0 4.0
-------------------------------------------------------------------------------
My output: Gaussian output:
-0.528103473677385300000000000000E+00 -0.528103D+00
-0.207077182228009010000000000000E-03 -0.207077D-03
-0.957752692231662150000000000000E+00 -0.957753D+00
0.312943635332713380000000000000E-06 0.000000D+00
-0.257198121834262120000000000000E-03 -0.257198D-03
-0.446838441550770900000000000000E+00 -0.446838D+00
------------------------------------------------------------------------------
The problem lies in the (1,3) (transposed) element. The number 10^-6 should
have been printed out in the standard Gaussian output file; it wasn't.
Any information would be greatly appreciated. Since our calculations rely
almost completely on these small numbers, this may cause severe problems in our
calculations.
Thank you in advance for any aid into this matter.
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Question 2)
Hello.
Is the FOCK matrix that is printed out from Gaussian 94
F'C'=eC'
or is it:
FC = eSC.
The molecular orbital coefficients must be C with eigenvalues e. We need F
with C and S. We can do this ourselves but we need to know what 'kind' is the
FOCK matrix: viz., F or F'.
Thank you.
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
--
Iraj Daizadeh
Department of Chemistry
University of California
One Shields Ave.
Davis, CA 95616-5295
Phone: 530.754.8695
Fax: 530.752.8995
email: daizadeh (- at -) kappa.ucdavis.edu