c336346c777707e09cab2a3c79174d90-AuthorFeedback.pdf

It is unclear whether these subsets will be algorithmically well behaved, i.e., the optimal7 Lagrange multiplier forsuch constraints will besmall. Indeed, asshowninFig1(d)-(e), asprox center gets close to8 the point which violates MFCQ, the convex polyhedron flattens and bound on Lagrange multiplier blows up. These are quite standard solvers for convex and nonconvex sparse models. Ifsame constant isadded on both side of (5) then algorithm does not45 change at all. The constraintg(x) is nonconvex and nonsmooth according to Assumption 2.1.52