This program performs *a posteriori*

    A) Bond order and valence analysis and

    B) "Chemical Energy Component Analysis" (CECA)

of the ab initio SCF LCAO-CO wave functions produced by the CRYSTAL98
program system. It uses as input the output file (".hftxt") of CRYSTAL98.


Bond order and valence indices represent LCAO counterparts of the
respective classical chemical parameters (I. Mayer, Chem. Phys. Lett. 97,
270 1983...) while "CECA" is a new tool permitting to express the total
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 (I. Mayer, Chem. Phys. Lett. 332,381,2000);
three- and four-center effects are compressed to one- and two-center ones
by performing appropriate projections.


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

Two "tar" files can be downloaded. One contains the original version of
the program that we have used on the large IBM SP2 cluster in Heidelberg,
another a Linux version whith reduced dimensions, tested in Budapest.
(It requires 1GByte core memory. See however below!) Both "tar-s" contain
an example input (i.e. the output of CRYSTAL98) and the respective
CR-APOST output (files test.hftxt and test.out).

Due to "historical reasons" the program consists of three Fortran codes
which should be run consecutively.(A simple script "crapost" is included
doing the job.) The first two codes (manipx-ibm and dengenx)  perform data
preparation, the third (cr-apost) the actual calculation of bond orders
and energy components.

The three Fortran programs should be compiled independently. For compiling
cr-apost there is a Makefile. However, if desired, one can combine all the
constituents of cr-apost into a single Fortran file and compile them at
once. The only special point is to use for an IBM the linking statement in
the file Makefile. (The "tar" files" contain also the executables - they
may, or may be not, usable on other systems.)

Both the IBM and the Linux "tar" files contain a program BOD.f
performing only bond order and valence calculations. It is to be used
instead of cr-apost for systems with basis orbitals of f (g...) type or
which appear too large. (As the program has been adapted from a molecular
one, a number of auxiliary matrices is used. They exclusion - although
probably posssible - would require a major project.) This program can be
used by replacing in crapost the call of cr-apost by that of BOD. In this
case the core memory of 512 MB was found sufficient for the present
dimensions of the Linux version.

Remarks:

1. We do not have Linux versions of CRYSTAL98. The Linux version discussed
here uses the output of CRYSTAL98 calculation performed on IBM in
Heidelberg. However, the Linux version of CR-APOST may appear more
portable, as well as it permits independent post-processing of the results
calculated by CRYSTAL98.

2. If necessary, replace the call to "f77" in the Makefile by the name of
your FORTRAN compiler.)


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

The only input required by the program is the output (".hftxt"
file) which is produced by the CRYSTAL98. See the  "tar" file with
example. Take care that the ouutput of CRYSTAL should contain those
- and only those - quantities as the file test.hftxt in exactly
the same format!


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

I. Mayer, Program "CR-APOST", Heidelberg-Budapest 2000-2003;  based on
I. Mayer and A. Hamza, Program "APOST", (Chemical Research Center,
Hungarian Academy of Sciences), Budapest, 2000-2003.


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

This version of the program is not applicable for energy decomposition
if there are basis orbitals of type f (or higher). In these cases
program BOD is still applicable.


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