In an -dimensional real vector space, each member of a conic hull has a representation using at most generators. To prove it, take a finite positive-coefficient representation with more than terms. Its generators are linearly dependent, say , with some after reversing the relation if necessary. Subtract from each coefficient, taking . All coefficients stay nonnegative and at least one vanishes. Iterate. The bound differs from the bound for a convex hull because the coefficient sum is unrestricted.
For a convex cone in an inner product space, use the nonnegative-pairing convention
Here the pairing is the Frobenius inner product on real symmetric matrices. Let be the conic hull of the nonnegative rank-one matrices , and let . For , . This extends to conic combinations, and by continuity to their limits. Hence .
For the converse, if , separation from a closed convex cone supplies a symmetric with
In particular for every , so . The negative pairing then excludes from . Therefore
The same generator test gives . This is the duality of copositive and completely positive cones; the next argument removes the closure.
The real vector space of symmetric matrices has dimension . By the conic Carathéodory theorem, every member of is a conic combination of at most generators. Absorb each nonnegative coefficient into its vector through .
If converges to , write, padding with zero vectors if necessary,
The matrix trace satisfies
The left side is bounded because converges. Thus the finite tuple is bounded. The Bolzano-Weierstrass theorem gives a subsequence on which every vector converges, say . Continuity of the outer product now gives . Hence
This proves closedness of the completely positive cone. The uniform bound on the number of factors and the matrix trace bound are both essential: an arbitrary conic hull of a closed generating set need not be closed.