Search for $\varepsilon$-optimal strategies for observation of a markov process

The moment of observation of a Markov process are determined so that the losses sustained in the absorbind state and consts of observation over a finite time interval be different from the optimal value by no more than a specified $\varepsilon$. Search algorithms are proposed for all possible programmed observation programs. An example is given of solving an applied problem of determining test points in an industrial process.