The Parallel Computing Airlift Allocation Model (PaCAAM)

Air-crafts and air-crews are allocated by the military at its different home bases in anticipation of the demands for several missions that are to be conducted in the period of one month. Each aircraft flies from a base location and serves a source destination pair. It is important that the allocations of air-crafts to the different missions at a base are made carefully because re-allocations are costly. This project aims at modeling and solving this complex MIP (mixed-integer programming problem) with stochastic demands using parallel computing technology.

To account for the stochastic nature of demands, we model the problem as a two stage problem with recourse. There are two main sources of parallelism. The problems at each stage can be decomposed further into several sub-problems that can potentially be mapped onto different processors to leverage the benefit of parallel computing. Furthermore, multiple instances of the second stage problems need to be solved to calculate the recourse function.

Investigators: Udatta Palekar, Laxmikant V. Kale, MITRE