We present a new structure-conveying algebraic modelling language for mathematical programming. It is based on AMPL, with object-oriented extensions that allow models to be constructed from sub-models. The implementation is a combination of pre- and post-processing for AMPL. Unlike traditional modelling languages, our approach does not scramble the block structure of the problem, thus enabling this structure to be passed on to the solver. Interior point solvers that exploit block linear algebra (e.g. OOPS) and decomposition solvers can therefore directly exploit the structure of the problem.
One major advantage of such structure-exploiting solvers is their ease of parallelisation. To keep this advantage, our modelling language is designed to allow sub-model processing to be run in parallel as well. Through this design our modelling language is scalable to model truly large problems.
Current 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996