We present a fast, robust, regularized nested Benders' decomposition code which has been used to solve stochastic programming problems which arise from practical financial planning and logistics applications. The code features a stochastic program presolver, as well as aggregation procedures. For certain classes of problem, we have found that an appropriate regularization term in the objective function can speed convergence, and alleviate the numerical instability associated with standard decomposition methods. We also demonstrate how Benders' decomposition can be used to solve problems with a convex nonlinear objective function and how it can be specialized to handle problems with a non-Markovian constraint structure. Solution times are given which compare favourably to those of currently available deterministic and stochastic solvers.
Current 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996