sci.op-research

In article <1cdd67b4-9168-415c-b75b-e65709365f0b@h25g2000hsf.googlegroups.com>,
GG writes:
>Hello,
>
>I am looking for a program or library for convex analysis.
>
>For an m x n matrix M, I looking for v which satisfies Mv = 0. There
>exists a convex set of vectors all non-negative linear combinations of
>which satisfy the equation above, so-called extreme rays. I am looking
>for a program to get extreme rays out of a given matrix M.
>
>I look forward to hearing from you. Any comment will be welcomed.
>Thank you very much.

something must be wrong here:
the solution set of Mv=0 is a subspace, which best is computed using
the svd of M. Maybe you meant Mv >=0, v>=0 ?
Computing all extremal directions of an unbounded polyhedral set is as complex as
finding all vertices from which an LP step detects unboundedness
(the current correction direction computed in this step is just such an
extremal direction) I don't know whether this is feasible

hth
peter