Efficient compact linear programs for network revenue management

Abstract We are concerned with computing bid prices in network revenue management using approximate linear programming. It is well-known that affine value function approximations yield which not sensitive to remaining capacity. The analytic reduction compact programs allows the efficient computation of such prices. On other hand, capacity-dependent can be obtained separable piecewise approximations. Even though have been derived for this case also, they still computationally much more expensive compared functions. propose requiring substantially smaller times while, simultaneously, significantly improving performance capacity-independent This simplification achieved by taking into account capacity only if it becomes scarce. Although our proposed relaxations unreduced programs, we conjecture equivalence and provide according numerical support. measure quality an approximation difference between expected induced policy corresponding theoretical upper bound. Using paradigm experiments, demonstrate competitiveness programs.