The flag which specifies whether or not to return singular/eigenvectors alongside the singular/eigenvalues is extremely inconsistently named.

• scipy.linalg.eig: left=True/False, right=True/False
• scipy.linalg.eigh: eigvals_only=False/True
• scipy.linalg.svd: compute_uv=True/False
• scipy.sparse.linalg.eigs / scipy.sparse.linalg.eigsh: return_eigenvectors=True/False
• scipy.sparse.linalg.svds: return_singular_vectors=True/False/'u'/'vh'

Also, linalg admits functions scipy.linalg.eigvals, scipy.linalg.svdvals and scipy.linalg.eigvalsh equivalent to running the above functions with the appropriate parameter selection, but sparse.linalg offers no such functions (and in my opinion neither should, according to the Zen of Python: "There should be one-- and preferably only one --obvious way to do it.")