THIS DIRECTORY IS UNDER CONSTRUCTION; PROBLEMS ARE
GRADUALLY BEING ADDED.  (SEJ)

The files in /pub/nonconvex contain test problems for
linear reverse convex programs (LRCP's), and concave
minimization problems. The filenames that begin with
"sr" contain problems that were generated by the 
Sung/Rosen method, and those file names of the form
"*c.*" are concave minimization problems. Filenames that
begin with "jm" contain linear reverse convex programs that
have been generated by a method of Jacobsen/Moshirvaziri.
Other files contain problems that have appeared elsewhere.

All of these problems have been solved with a basic
edge search procedure (all globally optimal solutions are on
edges of the polytope when all optimal solutions of the
linear program are infeasible for the "reverse convex" nonlinear
constraint). For the concave minimization problems, we use the
phrase "best found x" (not "optimal x") to indicate that our 
implementation of edge search for LRCP's may not find the 
globally optimal solution for concave minimization problems; 
however, we believe that it is has done so for all such problems 
in this directory.

The computational results, using a straight forward edge
search algorithm, for these problems are contained in
the file results.ps. The two tables that are contained
in results.ps are fully explained in the paper
"Computational Experience Using an Edge Search Algorithm
for Linear Reverse Convex Programs", which is to appear
in J. Global Optimization.

All of the problems have non-global local minima.


  Steve Jacobsen & Khosrow Moshirvaziri