            
     

                    Program LOCAL, Version 1.01
                   -----------------------------

   Calculating:
   ------------

   A) Conventional (Hilbert space) bond orders and valences
      (I. Mayer, Chem. Phys.Lett. 97, 270, 1983; IJQC 29, 73, 1986...)

   B) Mulliken populations

   C) "Extremely localized" set of non-orthogonal molecular orbitals, 
      their strictly localized projections as well as their 
      Lowdin-orthogonalized counterparts.
      (T. Zoboki and I. Mayer, to be published) 


   Usage:
   ------

    The program is devoted to a posteriori processing of the results obtained
in a Gaussian run. It requires the "formatted checkpoint file" generated by
Gaussian (for the basic calculations) and the Gaussian output to generate
files which permit to visualize the orbitals obtained by using Molden.
These files are only read by the program, so they remain unchanged. 

    The system consists of two Fortran programs: program "groups" which 
performs the actual calculations (its executable can be produced according to 
the "Makefile" from the different files by using command "make") and the stand 
alone program "promoldg" which "inserts" the orbitals obtained in the 
calculations in a copy of the Gaussian output file. (It should be simply 
compiled.)

    The calculations are directed by the small script "local", the use of
which is highly recommended, by issuing the command
                    "local NAME"
where NAME is chosen by the user. For the use one should prepare an
input file NAME.inp and some other files in special cases.

For standard calculations one needs only a file "NAME.inp" in which the first 
line contains the actual name of the "formatted checkpoint file" to be 
processed (with full path, if necessary) in the first line, and the name of 
the Gaussian output file (also with path, if necessary) in the second line. 
For saturated molecules with conventional electronic structure, that is the 
whole input required - for more complex systems see below. 

Program "groups" calculates the different orbitals and puts them on the
unformatted Fortran files "fort.77" to "fort.80". Program "promoldg"
always reads the data from Fortran file "fort.77" and puts them into
human readable format to the output as well as produces a copy of the
Gaussian output, in which the orbitals are inserted before the canonic
ones. For that reason, the script copies different Fortran files to
"fort.77". The orbitals produced are

A) the set of nonorthogonal localized orbitals which have maximal projections 
on the respective one- and two-center subspaces of basis orbitals (and on the 
Pi-system, if necessary, see below): files "NAME.lout" and "NAME.lorb", the 
latter being suitable for being studied with Molden;

B) the set of strictly localized projections of the above orbitals: files
"NAME.gout" and "NAME.gorb", the latter being suitable for being studied 
with Molden;

These orbitals are related to each other by Lowdin's pairing theorem.
That means that the localized orbitals obtained for *a given fragment&
and their strictly localized counterparts both form an orthonormalized
set and their "interset" overlaps differs from zero only in pairs. The
respective overlap integrals (lambda_i values) characterize, how close is 
a given orbital to its strictly localized projection.

C) the set of orthonormalized localized molecular orbitals, representing
Lowdin-orthogonalized counterparts of those in A): files "NAME.lwdout" and
"NAME.lwdorb", the latter being suitable for being studied with Molden.

If the system contains one- or more pi-subsystems, then the atoms
constituting each such subsystem should be listed on a separate card 
(format 20i2) in the file "NAME.inp" following the two lines discussed
above. (There can be maximum 20 pi subsystems with maximum 20 atoms in each.)
E.g. for benzene with carbon atoms numbered from 1 to 6 one has to give the 
line
 1 2 3 4 5 6  

If the system is charged or one wishes to define a zwitterionic configuration
as reference, then a file named "NAME.chg" should also be prepared. It
should contain the respective number of lines each with the number of
a charged atom and its charge. (Free format). E.g. for nitrone molecule
CH2-NH-O for which atoms C N and O are numbered 1 to 3, one may specify
a zwitterionic structure with positive N and negative O as
 2  1
 3 -1
(However, it is perhaps more adequate to specify a pi-system in NAME.inp
as 
 1 2 3
If one specifies both charges and pi-system, then practically only the letter
will be taken into account. (That combination may be necessary for really 
charged pi-systems.)

Normally the orbitals follow in the order: one-center (core and lone pair)
orbitals, sigma bonding orbitals and pi orbitals. As there is no special
distinction between the sigma and pi basis orbitals, and the orbitals with
largest projection are selected, in some cases a mixing over of the sigma and
pi orbitals may happen. In such cases instead of an one- or two-center orbital
which one expects to be a sigma one, an orbital with pi-symmetry is 
obtained. Although the localized orbitals obtained still span the occupied
subspace, one may wish to return to a conventional picture. That can be
achieved by skipping the orbital of undesired character, and using the
next one. In that case a file "NAME.exc" should be prepared, containing
for each orbital to be skipped two lines: on the first the atom(s)
are listed to which this procedure should be applied (one atom, two 
atoms or the pi subset) on the second the number of the orbital(s) of the
group which is to be skipped should be specified (format 10i2)
For instance, if the second orbitals of the 3d atom is to be skipped,
the file should contain
 3
 2
If the first orbital of the bond 3-4 should be skipped, use
 3 4
 1
Up to 20 exclusions may be specified.

Limitations: 30 atoms, 500 basis orbitals, 2000 primitive Gaussians.
the second orbital on the 3-d atom 

Instead of the orbital energies the overlap integrals "lambda_i" appear
in the outputs, where "lambda_i" is the overlap between the i-th localized 
orbital and its strictly localized projection. (These are related to each 
other according to Lowdin's pairing theorem.) 


    Cite this program as:
    ---------------------
    I. Mayer, Program "LOCAL", Version 1.01, Budapest, 2010.     

-------------------------------------------------------------------------------
  The program has been written by using parts of the program APOST by
  I. Mayer and A. Hamza, Budapest, 2000-2003.
      

                     
          e-mail: mayer@chemres.hu

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


