Proceedure for construction of the demo package.  

Below are the specifics for these examples (only the changes are shown).

copy the file demo1.parx to the working directory.  Reset the input
parameters using lblf found on the BASIS directory:

skeeve>../../lblf demo1.x
  truncation values  [lemx,lomx,kmx,mmx,ipnmx] :   2   0   0   1   1
New value(s) > 6 5 6 2 2 

Create a basis set that is complete:

skeeve>../../eprbf demo1.x

Prune that basis set:

skeeve>../../prune demo1.x


  ######################################################################


                            PROGRAM : PRUNE
                             --------------


  cut-off tolerance for pruning :    .0000E+00
New value(s) > .01 

  old dimension =   66     new dimension =     8


There you see a significant reduction in the matrix size from 66 to 8.
Check out the example to see that this only effects the spectrum slightly.

Modify demo1.run to initially use the full basis set for the first
calculation:  

let lemx,lomx,kmx,mmx,ipnmx=6,5,6,2,2   
basis   
and remove the data file reference

The output file is simu1.fit when the data file is not specified and is
constructed with the full basis set.

Next use Matlab (pad.m) to reformat the simu1.fit file so it is equivalent
to a data file and may be read by the fitting program.

We use the output of the above fit as a data file to be fit with the pruned
basis set for testing the completeness of the pruned data set.  Matlab
function pad.m changes the dimension of the fit file and adds noise, making
it a suitable data file.

>> pad simu1.fit

skeeve>mv simu1.spd demo1.spc

Now modify the run file for using the pruned basis and the data file created
above:

basis demo1.indx
data demo1 noshift nostore

Fit the data:

NLSPR> read demo1.run
basis demo1.indx   
 Dimension of matrix: ndimo =    8 ndimd =   10
 MTS lemx,lomx,kmx,mmx,ipnmx =   2, 0, 0, 1, 1
NLSPR> fit
          Single calculation : RMS deviation =  5.595438    

Graphical comparison of demo1.spc and demo1.fit shows the quality of the 
fit is excellent.

***********

Now repeat the above sequence for demo2 where slower motion is selected.  

copy the file demo2.parx to the working directory.  Reset the input
parameters:

skeeve>../../lblf demo2.x

  diff. tensor [dx,dy,dz] (1/sec) = 4.6415900E+07 4.6415900E+07 4.6415900E+08

truncation values [lemx,lomx,kmx,mmx,ipnmx] :  4 0 2 2 2
New value(s) > 6 5 6 2 2

Note the slower motion.  

skeeve>../../eprbf demo2.x

skeeve>../../prune demo2.x


  ######################################################################


                            PROGRAM : PRUNE
                             --------------


  cut-off tolerance for pruning :    .0000E+00
New value(s) > .01

  old dimension =   66     new dimension =    16


  ----------------------------------------------------------------------
   error       Le        Lo        K         M         pI     LKMdff
            min  max  min  max  min  max  min  max  min  max  max

  .100E-01    0    4    0    0    0    2    0    2    0    2    0

  updating files : <id>.parx, <id>.indx


Note the larger basis set required for the slower motion.

Modify demo2.run to use the full basis set for the first calculation as
above:

let lemx,lomx,kmx,mmx,ipnmx=6,5,6,2,2
basis
and remove the data file reference

Next use Matlab to make a file suitable for reading by the fitting program.
We use the output of the above fit as a data file to be fit with the pruned
basis set for testing the completeness of the pruned data set.  Matlab
function pad.m changes the dimension of the fit file and ads noise, making
it a suitable data file.

>> pad simu1.fit (note simu1.fit is the default output when no data is
present)

skeeve>mv simu1.spd demo2.spc

Now modify the run file for using the pruned basis and the data file:
basis demo2.indx
data demo2 noshift nostore

Fit the data:
NLSPR> read demo2.run
basis demo2.indx   
 Dimension of matrix: ndimo =   16 ndimd =   22
 MTS lemx,lomx,kmx,mmx,ipnmx =   4, 0, 2, 2, 2
NLSPR> fit
          Single calculation : RMS deviation =  4.641262    

Graphical comparison of demo2.spc and demo2.fit shows the quality of the
fit is excellent.

This testing of the pruning choices must be done regularly to insure correct
fits. 

Note, in demo2.run the dead times in t1 and t2 are set to zero.  This is not
a physically realizable experiment at this time but was done to show the
spectra before significant weakening of the peakes occurred.  

