                         NOTES ON DATAFILES

 (1) Scope of test-data.

 (a) The test-data in TESTDATA.DAT are intended to demonstrate the
     capabilities of MOPAC. 
 (b) The test-data files are not designed for checking MOPAC for
     errors.
 (c) Only the main paths through MOPAC are demonstrated. 

 (2) The test data are intended to be run as follows:
 
 (a) Run first the single test-data file MNRSD1.DAT, the one described in
     the Manual. Check that the results are acceptable. Small differences
     in calculated quantities are to be expected due to different definitions
     of double precision numbers, and to minor changes made to MOPAC as a
     result of suggestions made by early users of MOPAC.
 (b) Run, in turn, the first four files in TESTDATA.DAT, these show the
     four main electronic pathways: the RHF closed-shell, the UHF open-shell,
     the RHF triplet open shell (does not use C.I.), and the RHF biradical
     (which does use C.I.). If any of these fail to run successfully, no
     associated geometric path will work. 
 (c) The next four data files in TESTDATA demonstrate the geometric pathways.
     These are, in order, the FORCE calculation, which calculates force
     constants, vibrational frequencies, and thermodynamic quantities;
     the FORCE calculation being used with an ill-defined geometry (FORCE
     should only be used with systems at their stationary points, if it
     is not at a stationary point, Bartell's method will be used to optimise
     the geometry to the stationary point; this is normally less efficient 
     than using the D.F.P. method or SIGMA method); the SIGMA method for
     locating a stationary point (here the MINDO/3 geometry for formaldehyde
     is used to start a MNDO optimisation); and a Bartell's method
     optimisation.
       Bartell's and the SIGMA method are mainly used to locate transition
     states. Bartell's method will always work, but can be extremely slow
     at the end, and can locate a point of inflection if the initial geometry
     was very poor. SIGMA is slow to start, since it constructs a full Hessian,
     and will normally fail if the geometry is not near to a stationary point,
     but when it does work is very fast and accurate. Sometimes the results
     of a SIGMA optimization that failed are sufficiently good to act as
     data for a second SIGMA optimization that will work.
 (d) The next two data files demonstrate the reaction options. The first
     is the "reaction" of one methyl group rotating relative to the other
     fixed methyl group in ethane. The other is the location of the transition
     state in classical ethyl cation hydrogen migrating from one carbon atom
     to the other.
 (e) Finally, there is a demonstration of the location of a transition state,
     and its characterisation as a transition state via force constants.

 (3) Initially, it is suggested that the testdata are used as "templates"
     for generating data. Users are recommended to experiment with 
     various combinations of key-words to their test understanding of MOPAC.
     To save time, small compounds, such as H2O and CO2 etc. are advised
     as model systems.
