2006
BealeOrchardHays Prize Citation
Nick Sahinidis and Mohit Tawarmalani,
"A polyhedral branchandcut approach to global optimization",
Mathematical Programming, Series B 103 (2005), pp. 225249.
The approaches described in this paper are implemented in the BARON
system. It represents a powerful approach for the global optimization
of nonlinear optimization problems, including problems with integer
variables.
The paper develops techniques that enhance previous versions of
BARON. In particular, it uses factorable decompositions of nonlinear
functions into subexpressions to construct polyhedral outer
approximations that exploit convexity more thoroughly and yield
tighter underestimators, in comparison to techniques that do not use
decompositions. BARON also incorporates techniques from automatic
differentiation, interval arithmetic, and other areas to yield an
automatic, modular, and relatively efficient solver for the very
difficult area of global optimization.
