Program APEX4, Version 1.02
                        ---------------------------------------
                                                                                
                         I. Mayer and A. Hamza
            Chemical Research Center, Hungarian Academy of Sciences
                    H-1515 Budapest, P.O.Box 17, Hungary
                                                                                
             e-mails: mayer@chemres.hu,  hamza@chemres.hu
                                                                                
 ------------------------------------------------------------------------
                                                                                
   This program performs *a posteriori* analysis of the single determinant
   Hartree-Fock ab initio wave wave functions produced by the widely used
   "Gaussian" system G92, G94, G98, G03. (For DFT-type wave functions
   - actulally B3LYP - the HF-like energy of the single determinant built
   up of the Kohn-Sham orbitals can be considered.)
                                                                                
  The analyses performed:
                                                                                
  A) Bond order and valence analysis (I. Mayer, Chem. Phys. Lett. 97, 270
     1983; addendum for open shells: Chem. Phys. Lett. 117, 396, 1985, etc.);
                                                                                
  B) "Chemical Energy Component Analysis" (CECA), introduced in I. Mayer,
     Chem. Phys. Lett. 332, 381, 2000. CECA permits to express the
     molecular energy  -- approximately but to a good accuracy -- as a sum of
     one- and diatomic energy components, the computation of which requires
     one and two-center integrals only; three- and four-center effects are
     compressed to one- and two-center ones by performing appropriate
     projections;
                                                                                
  C) The decomposition of the two-center energy components into terms
     of different physical origin, as defined in I. Mayer and A. Hamza
     Theor. Chem. Accounts 109, 92, 2003.
                                                                                
  D) The "Exact Energy Decomposition" as defined in I. Mayer, Chem. Phys.
     Lett. 382, 265 (2003). It differs from CECA by the absence of
     approximations and a different treatment of the kinetic energy
     and the two-center finite basis corrections.
 

 
       In fact, four different energy decomposition matrices are computed,
  two approximate and two exact. They differ in the approximate or
  exact treatment of the the 3- and 4-center integrals from one side,
  and in the treatment of kinetic energy integrals and (two-center)
  finite basis corrections from the other. One type of decompositions
  (the approximate CECA and the exact "CECA + 3- and 4-center terms")
  contain kinetic energy mainly in the atomic terms (and in the finite
  basis corrections) while the "CECA/T" and "exact" decomposition contain
  both one-and two-centyer kinetic energy contributions and there is no
  need in the finite basis correction terms. Only two of these four have
  been described in publications as yet: see CECA in A) and "exact" in D)
  above. The diatomic components printed are essentially those in C).
  (Plus the kinetic energy is printed).
                                                                               
  The "approximate" and "exact" schemes become identical for diatomics.
                                                                               
  The CECA-type quantities are good indicators of different interactions,
  but are not on the "chemical scale": there arelarge atomic promotions and
  large (in absolute value) two-center binding components. The CECA/T and
  "exact" schemes give much more "chemical" values, but should be used only
  at the equilibrium molecular geometries.
                                                                            

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

Installation of the program:
---------------------------------

The "tar" file contains all the necessary FORTRAN source files as well

as a "Makefile", which should be appropriate to produce the executable
by issuing the command "make" on a UNIX or Linux system. (If necessary,
replace the call to "f77" in the Makefile by the name of your FORTRAN
compiler.) The name of the executable generated is "apost4".

Usage of the program:
----------------------------

The only input required by the program is the "formatted checkpoint
file" (Test.FChk) which is produced by the Gaussian, provided that the
FormCheck keyword is included in the Gaussian input. In the program
APEX4 the file Test.FChk is used for input as the FORTRAN file fort.14,
and is copied to file "Tmp" for actual processing (file fort.15).
 

An alternative is to conserve the binary checkpoint file which is
produced during the Gaussian run (include %chk="filename" in the input)
and to format it by using the command "formchk" which is a part of the
Gaussian system. The formatted checkpoint file should then be renamed
(copied) to Test.Fchk (or to fort.14, if the statement
"open(14,file='Test.FChk')" has been deleted from "subroutine input".
 

A convenient way of using the program is to include it in the
script performing the Gaussian run; one should take care of ensuring
that "APEX4" should NOT start before Gaussian is finished. It is
desirable to delete or rename the formatted checkpoint file Test.FChk
after the calculation, in order to avoid its unwanted reuse.
 
The program is based on our program APOST. Note that the present
program may be significantly more demanding as for computer memory
and computation time.

 

Cite this program as:
--------------------------

I. Mayer and A. Hamza, Program "APEX4", Version 1.0
(Chemical Research Center, Hungarian Academy of Sciences),
Budapest, 2004.
 
 

Limitations:
---------------

This version of the program does not perform energy partitioning
in the case of basis orbitals of type f (or higher). In these cases
bond order and valence analysis is still performed.

 

Bugs etc.
------------

Please report bugs, problems etc. to either of the e-mail addresses
above - we will try our best to help you.