This file briefly explains the steps to add a new distribution type to 
the LHS UNIX Library/Standalone version of Latin Hypercube Sampling. 

1.  In the LHS/mods directory, change the Distnm.f90 file to add
distribution types.  Specifically, change LEND (increment it by the number 
of distributions you are adding).  Change DIST and allow for enough 
characters in the names of the distributions you are adding.  
Add DISTRIBUTION XX (e.g., 38) after the current distributions, 
in the SUBROUTINE DISTNM_INIT().

2. Add the new_dist_name.f90 files in the main LHS directory. You may want 
to pattern them after an existing distribution. 

3.  Change Wrtpar.f90 to write the new distribution type (by using a 
conditional clause such as IF(ID .EQ. 38) THEN ...). 

4.  Change the Chkdat.f90 file to check for the correct parameter inputs 
for the new distribution.  You need to add a distribution keyword near the top 
of this file, then upon finding that distribution, you add the proper checks 
for the associated parameters and also write statements. 

5.  Add the distributions to lhs.f90, so that when lhs is called with 
the new parameter, it will execute that distribution with a CALL routine. 

6.  Add the same logic to Lhs_run.f90, the callable subroutine version of LHS. 

7.  Add the new distribution.f90 file names to Makefile.am

8.  Build and test, then commit the changes and the new files to the LHS 
directory. 

9.  To ensure compatibility with DAKOTA, you need to also add the new 
distributions to the dakota input specification, and make the 
corresponding additions to the DakotaVariables, DataVariables,
DakotaNonD, DakotaModel, DataFitSurrModel, and NonDSampling classes. 
The actual call to LHS is in NonDSampling.C.  The new distribution must 
be entered in the same order as it is found in the input specification. 
 
