Modeling of inner currency in reflexive games with multi-criteria payoff functions
Keywords:reflexive games, multi-criteria payoff functions, inner currency method, linear convolution method, interval estimates
AbstractThe problem of decision-making under conditions of the conflict, multi-objective uncertainty and reflexive interaction of the parties is considered. Modeling of reflexive behavior makes it possible to analyze situations when the decisions taken differ from the non-reflexive rationality, to investigate and reveal the internal causes for such behavior. The solution of this problem, taking into account the multi-valued interests of the parties, is based on a multi-criteria generalization of the proposed V. Lefebvre setting, based on the use of the concept of inner currency. To calculate the initial assessment of the opponent's inner currency on the basis of nominally known criteria, the interval estimates method was used. Its application enables experts to set a range of possible values of weight coefficients, without requiring the definition of their specific values, which simplifies the expert procedure. The further refinement of the weighting factors occurs by solving an auxiliary problem for finding corrections that are introduced into the model for determining the inner currency.
Lefevr V.A. Vnutrennjaja valjuta v refleksivnyh igrah / V.A. Lefevr, P.V. Baranov, V.E. Lepskij // Yzvestyja AN SSSR. Tekhnycheskaja kibernetyka. — 1969. — № 4. — S. 29–33.
Smyrnov S.A. Protsedura obchyslennja vnutrishn'oyi valjuty v refleksyvnykh ihrakh / S.A. Smyrnov, I.M. Tereshchenko // Systemni doslidzhennja ta informatsijni tekhnolohiyi. — 2015. — № 1. — S. 39–44.
Smirnov S.A. Garantirovannyj sintez skaljarnogo kriterija dlja reshenija zadachi mnogokriterial'noj optimizatsii / S.A. Smirnov, I.S. Gontarenko // Systemni doslidzhennja ta informatsijni tekhnolohiyi. — 2006. — № 2. — S. 99–106.