简介:Inthispaperwediscussthediscrete,timenon--homogeneousdiscountedMarkoviandecisionprogramming,wherethestatespaceandallactionsetsarecountable.Supposethattheoptimumvaluefunctionisfinite.Wegivethenecessaryandsufficientconditionsfortheexistenceofanoptimalpolicy.Supposethattheabsolutemeanofrewardsisrelativelybounded.Wealsogivethenecessaryandsufficientconditionsfortheexistenceofanoptimalpolicy.