Principal
investigator: Assoc. Prof. Dr. Cristinca FULGA
Research team: C. Fulga, A. Agapie, R.
Ciumara, S. Dedu, C.C. Popescu
Duration: 2009-2011
Acronym: DeMaMod
Presentation of the project
The aim of the project is the decision making
modeling under uncertainty by combining multicriteria optimization
techniques, dynamic and stochastic optimization techniques, as well as the
development of efficient algorithms, adequate to the specific
characteristics of this type of problems. The models and the algorithms
developed can be used to solve problems concerning economic area. In this
project they are applied to portfolio optimization.
Four research directions are taken into
account.
The first direction consists in determining
the characteristics of the multicriteria problem objective functions. The
use of the utility functions with generalized convexity properties and new
risk measures are proposed, in order to characterize the attitude towards
risk of the decision maker.
The second direction consists in modeling and
developing algorithms for multicriteria decision making. The situation in
which the problem can be decomposed into subproblems will be analyzed and
new specific algorithms using evolutionary algorithms will be developed.
This model will be applied to economic problems with multiple decision
factors, which take decisions as a consequence of their own informations
and whose decisions must be coordinated such as, acting separately, they
generate general optimal solutions.
The third direction regards dynamic decision
making modeling. Models based on approximate dynamic programming techniques
combined with evolutionary algorithms are developed. These models provide solutions
to the dynamic resource allocation problems, such as: dividing the budget
for the productivity activities, portfolio management, planning the new
energetic resources.
The fourth direction regards the correlation
of the obtained results and their application in dynamic portfolio
optimization. In the same time, a relevant database for the construction of
realistic scenarios will be developed and will be used to validate the
developed algorithms.
|