About the one multiagent optimization problem in positional games

We consider a variant of the classical problem of operations research arising in the analysis of positional games. In contrast to the classic game of “sea battle” and other games in which the actions of subjects are analyzed with the help of cellular automata, there is a new fundamental moment in this task. The attacking side has the “intellectual ability” to overestimate the situation during the attack and take into account which of the figures are already defeated.
We solve the resulting optimization problem by means of simulated annealing. The solution turned out to be paradoxical. As the resource of the attacking side increases, it became necessary to change the attack model. The results obtained are illustrated by the example of the sea battle, but they can be useful in solving a number of problems of optimal control of limited resources.