We are using the mosek solver inside of matlab, the primal simplex version. We get an optimal solution and a basis. But the basis is length < m, where the problem is in standard form with Ax=b and A is m by n, x>=0 (all upper bouinds are set to infinity).

[rPhase1a, resPhase1a] = mosekopt(‘minimize echo(0)’,probPhase1a,param);

cellSol = cellstr(resPhase1a.sol.bas.skx);

indsBas = find(strcmp(cellSol,‘BS’)); % Find inds of cells that are BS

xbas = resPhase1a.sol.bas.xx(indsBas);

i.e. the length of indsBas (xbas) is < m.