Use the Charnes-Cooper transformation on the positive-denominator region:
It gives , and .
We must check that allowing introduces no false feasible solution. If , then . The recession cone of the nonempty linear polyhedron is . Indeed for , for every . Boundedness of therefore forces , contradicting . Thus every transformed feasible point has .
Conversely set . The constraints give and , with the same objective value. The assumed original optimizer with positive denominator supplies a transformed feasible point. Every transformed point corresponds to an original point and has objective at most that optimizer's value. Hence the transformed linear program attains the original optimum, and its optimizer recovers . Positivity on every point of is not needed here; the given positive-denominator optimum suffices.
Apply the Charnes-Cooper transformation to any feasible point of :
Then
so is feasible for , and
In particular, the given solution of produces a feasible point of with the same objective value, so
Moreover, because every , the equation implies
The linear objective is therefore bounded on the feasible set. The feasible lie in the compact simplex . Their subset arising in is closed: if , any nonzero component of determines continuously from , while if the condition is simply . Hence the feasible form a compact set, on which the continuous objective attains a finite maximum. Thus has a finite maximum at least as large as that of .
Recession cone 2026-10-06
The recession cone of a nonempty convex set consists of directions for which for every and . For a linear polyhedron it is exactly . A nonzero recession direction gives an unbounded ray, so a bounded nonempty polyhedron has recession cone . This excludes spurious zero-scale feasible points in the Charnes-Cooper transformation.