Comparison of a deterministic and a stochastic formulation for the optimal control of a Lanchester-type attrition process

Abstract

The structure of the optimal fire distribution policy obtained using a deterministic combat attrition model is compared with that for a stochastic one. The same optimal control problem for a homogeneous force in Lanchester combat against heterogeneous forces is studied using two different models for the combat dynamics (the usual deterministic Lanchester-type differential euqation formulation and a continuous parameter Markov chain with stationary transition probabilities). Both versions are solved using modern optimal control theory (the maximum principle (including the theory of state variable inequality constraints) for the deterministic control problem and the formalism of dynamic programming for the stochastic control problem). Numerical results have been generated using a digital computer and are compared. (Author)