Is it possible to get the complete thin SVD?

I've noticed that the `nev` and `nsv` arguments are capped at `<n` rather than `<=n` if `n` is the smaller of the dimensions `n ` & `m` of the matrix. 

I am working with some maximum entropy models on sparse matrices, and it would nice to work with the complete thin SVD, rather than the `n-1` rank approximation. Is this possible with arpack?

