For some problems, using the included MIPS solver for DC OPFs is not the most robust option. If you have the Optimization Toolbox installed, you can try using it's LINPROG/QUADPROG solvers by setting the OPF_ALG_DC option to 300.
But, for academics running DC OPFs, I now recommend using CPLEX, Gurobi or MOSEK. All of them offer options for free academic licensing and they include high-performance LP/QP solvers that are quite robust. -- Ray Zimmerman Senior Research Associate 419A Warren Hall, Cornell University, Ithaca, NY 14853 phone: (607) 255-9645 On Mar 18, 2013, at 9:48 AM, [email protected] wrote: > Dear Professor Zimmerman, > I am trying to run several DCOPF for a stochastic analysis. When I > executed the DCOPF in Windows XP and Matlab 2008 (32bits), with MEX > Interface for the BPMPD Interior Point Solver, the DCOPF is optimal. But > the same program is executed in Windows 7 and Matlab 2012 (64bits), the > results are very different and often the DCOPF does not converge. In these > ocassions the error is: > > Error using \ > Error during solve with symmetric indefinite factorization. > > Error in mips (line 422) > dxdlam = [M dg; dg' sparse(neq, neq)] \ [-N; -g]; > > Error in qps_mips (line 185) > [x, f, eflag, output, lambda] = mips(p); > > Error in qps_matpower (line 230) > [x, f, eflag, output, lambda] = ... > > Error in dcopf_solver (line 216) > [x, f, info, output, lambda] = qps_matpower(HH, CC, A, l, u, xmin, xmax, > x0, opt); > > Error in opf_execute (line 71) > [results, success, raw] = dcopf_solver(om, mpopt); > > Error in opf (line 225) > [results, success, raw] = opf_execute(om, mpopt); > > Error in runopf (line 96) > [r, success] = opf(casedata, mpopt); > > Error in rundcopf (line 94) > [varargout{1:nargout}] = runopf(casedata, mpopt, fname, solvedcase); > > Error in OPF_EjemploAB (line 156) > R = rundcopf(mpc,opt); > > > This problem is for the optimizer?. > Can you recommend an optimizer for Windows 7 and matlab 2012 (64bits) > because the error does not allow me to run distributed computing. > >
