MATLAB How to make `eigs` work correctly or alternative to `eigs` for sparse matrix -

eigs(a,k,'sm') returns k smallest magnitude eigenvalues. however, gives warning not eigenvalues converge , cannot give k of smallest eigenvalues.

on other hand, using eig(full(a)), performing sorting myself , extracting smallest k of eigenvalues give me same answer.

by searching on internet long time, have heard lot of negative comments eigs. however, since matrix a sparse , large. eigs seems best way go in speed, there way can make eigs(a,k,'sm') right , output correct eigenvalues for. or there more reliable , faster numerical scheme finding eigenvalues of sparse matrix?