Consider an edge-weighted graph
G = (V, L), and define a
k-cover C as a subset of the edges
L such that each vertex in
V is incident to at least one edge of
C, and
|C| = k. Given
G and
k, the problem is to find a
k-cover of minimum weight sum. This paper presents characterizations of minimum
k-covers, and shows their weight to be convex with the parameter
k. An efficient algorithm is presented which generates minimum
k-covers continuously as the parameter
k ranges over all feasible values, together with a proof of optimality. The computational order of this algorithm is found to be
|V|
|L|
2.
Presently at the IBM Corporation, Yorktown Heights, New York.