> To use them however the problem has to be transformed into some specific matrix form, and the coefficients of all the matrices has to be passed to the solver using some API. This transformation is a tedious and error prone step that forces you to work with matrix indexes instead of readable variable names.<p>I found Zimpl easy to work with. (<a href="http://freshmeat.net/projects/zimpl" rel="nofollow">http://freshmeat.net/projects/zimpl</a>) A small domain specific language for linear modelling. See their user guide for some examples (<a href="http://cgm.cs.mcgill.ca/~avis/courses/567/cplex/zimpl.pdf" rel="nofollow">http://cgm.cs.mcgill.ca/~avis/courses/567/cplex/zimpl.pdf</a>).
Interesting. Have you done a comparison test against Coopr?<p><a href="https://software.sandia.gov/trac/coopr" rel="nofollow">https://software.sandia.gov/trac/coopr</a><p>Also were you using the glpk solver with CVXOPT on cpython? I find it hard to believe setting up the problem alone added nearly 20 minutes.