An implementable decomposition method based on branch-and-bound techni
ques is proposed for finding a global optimal solution of certain conv
ex programs with an additional convex-concave constraint h (x, y) less
-than-or-equal-to 0. A nonadaptive simplicial and an adaptive bisectio
n are used for the branching operation, which is performed in y-space
only. The bounding operation is based on a relaxation of the convex-co
ncave constraint h(x, y) less-than-or-equal-to 0. The algorithm can be
applied efficiently for linear programs with an additional affine mul
tiplicative constraint.