Interactive dynamic programming. A framework for solving multicriteria decomposable problems

Interactive programming is one of the approaches proposed to deal practically with multicriteria decision-making problems. We show that straight forward application of interactive methods to solve dynamic programming models fails to exploit the main feature of the models, which is the possibility to decompose the problem into easily solved subproblems. Consequently, sequential optimization cannot be fully applied, and standard interactive methods become impractical. We suggest combining interactive optimization with 'surrogate' value functions as an alternative to the difficult task of assessing a 'true' value function, to derive the full power from decomposition.