Your problem is ill-posed, becaus the state of a beam is not described by a state vector $\psi$, but by the corresponding density matrix $\psi\psi^*$, which determines $\psi$ only up to a phase. Thus it is meaningless to talk about ''the'' superposition of two beams.
If the two beams are known to have a fixed, but unknown, relative phase, the question becomes well-defined. The only way to gain this knowledge is to check that the beams have been coherently generated from a common source. In this case, a half-silvered mirror will produce the required superposition, and one can change the relative phase by a rotator (as described in the book by Mandel and Wolf).
For preparing arbitrary states of a collection of beams from a single coherent beam see my paper http://de.arxiv.org/abs/quant-ph/0306123. See also
M.Reck, A. Zeilinger, H.J. Bernstein and P. Bertani,
Experimental realization of any discrete unitary operator,
Phys. Rev. Lett. 73 (1994), 58–61.
These papers specify how to create appropriate transformations; to prepare a particular state, find a transformation that creates it from a state you know how to prepare, and then implement this transformation.