We introduce the convex combinatorial optimization problem, a
far-reaching generalization of the standard linear combinatorial
optimization problem. We show that it is strongly polynomial time
solvable over any edge-guaranteed family, and discuss several
applications.