Activity at the Department of Operations Research and Decision Systems
WINGDSS: Group Decision Support System
TRANSOPT: database manager and transport optimizer

Decision Support of Electricity Planning
Teaching activity
Decisions in Economics

Activity at the Laboratory of Operations Research and Decision Systems

During the recent years, the members of the Laboratory did research in several fields of operations research and decisions systems. Although these fields are substantially interrelated, they can be subdivided conforming to the diversity of the mathematical methods used in the analysis and the mathematical objects to be investigated. Some of these fields are listed below: 

• analysis of stochastic and deterministic systems
• decision systems
• economic time series
• global optimization
• linear and quadratic programming
• mathematical methods in decision support of industrial applications
• nonsmooth optimization
• numerical methods in operations research and mathematical physics
• parametric uncertainty in adaptive control
• tensor optimization.

Besides proceeding with research in the earlier successful fields, further research has been or is planned to be started in some new themes. 

Basic research
 

• equilibrium models, structural issues of variational and quasi-variational inequalities; applicability of Finsler geometry instead of the Riemannian geometry to describe the structure of the problems and methods in optimization models
• recognition and analysis of the special structure of global optimization problems
• combinatorial and global optimization tools for the analysis of process network synthesis problems, complexity issues
• linear differential-algebraic equations, solutions of their adjoints and transformations of the latter, asymptotics
• solvability of Riccati difference equations obtained by transformations of linear difference equations and their adjoints, analysis of limit behaviour, properties and stability of solutions of asymptotically time invariant Riccati equations
• structural properties of models of time dependent stochastic systems - both stationary and those having time dependent structure - , analysis of the structure of poles and zeros of stationary systems, model reduction, analysis of the associated Riccati and Hamilton equations, detection of transition time in systems of variable structure
• analysis of risk-sensitive control, i.e., linear exponential quadratic control in multiplicative systems
• axiomatic foundation of the models based on decision tables and Bridgman principle for treating multicriteria decision problems.

Methodological research
 

• analysis of interior point methods for linear and nonlinear optimization problems; analysis of the numerical behaviour of logarithmic barrier methods; efficient exploitation of the structure of optimization problems in interior point algorithms
• development and integration of combinatorial and global optimization methods based on structural analysis of process network synthesis problems and exploitation of the benefits from the special structure
• development of linear programming model and methodology for middle-range production planning of electric power systems enabling to consider multiple group-limitations on the generation of units, e.g. environmental protection limitations, fuel limitations, etc.
• development of a numerically more accurate version, based on new methodological ideas, of the segmentation method of production costing simulation
• control problems of continuous time stochastic systems, numerical solutions of SDE's, uncertainty and performance, interaction of identification and control, stochastic complexity
• identification problems of stochastic systems, parametrization of the models, dynamic factor-analysis, estimation of the model-parameters, analysis of related optimization problems (global optimization on manifolds, semidefinite programming, convex programming)
• analysis of the relation of differential-algebraic equations to the problem formulation and solution algorithms of the linear, parametrical constrained optimization and those of constrained quadratic optimization
• elaboration and solution of multi-criteria decision models with our decision supporting software system.

• developing and solving decision models for treating decision problems at higher hierarchical levels
• solving problems in decision making arising from the applications of the stipulations of the Public Procurement Law
• financial stochastics: analysis of stochastic interest rate models, theory of option pricing, application of models with stochastic differential equations in the theory of exchange rates, application of stable processes and fractional Brown processes
• option pricing: effects of uncertain volatility and transaction costs in the original Black-Scholes model, impact of new condition on the expectation value; problems of implementation: numerical solution of stochastic differential equations, simulation methods
• portfolio-analysis: development of risk sensitive strategies under new conditions, efficient solution of systems of matrix-Riccati equations, applications of technics of linear matrix inequalities, development of optimal investment models
• economic time series: nonstationary time series, effect of cointegration, nonlinear and treshold models, long time dependence, prices of stocks and exchange rates, prediction of interest rates
• financial optimization: analysis of the properties of convex quadratic optimization problems, investigation of efficient solvers on optimization problems from financial applications
• development of algorithms for computation of some models of synthetic spectra (energy levels and transition probabilities in strong magnetic fields) by solving the corresponding Schrödinger equation (application in astronomy, continuation of a Hungarian cooperation)
• development and application of algorithms for computation of models of semiconductor devices by solution of Poisson-Schrödinger system (international cooperation)
• on-line interpretation of ECG signals, detection of changes, application of SPSA optimization method.

Activity at the Department of Operations Research and Decision Systems

As mentioned above, the basic activity at the Department is research and development in the frame of projects financed from contracts and grants. The outcomes of these projects are usually computer programs and studies. We strive to undertake projects whose topics can be associated with some themes studied at the Laboratory as well. Besides the financial benefits, additional results of such projects are often publications in high-quality international journals. For demonstration, three themes of the research and development activity of the recent years are presented below. 

WINGDSS: Group Decision Support System

Decision problem of WINGDSS

• one or more decision makers
• ranking a number of alternatives or selecting the best
• qualification by decision makers of both subjective and factual criteria
• ranking the alternatives by quantified criteria

•  preliminary phase
• registration of decision makers
• assignment of voting powers and authorization
• building up the decision tree
• definition of the qualification procedures of the criteria
• setting the alternatives

• criterion tree editor
• data form editor
• dynamic link to external database

• individual decision making
• weighting the decision criteria: individual importance of criteria (optionally, ranges)
• qualification of the alternatives: utility functions
• individual ranking

• utility function editor
• modifiable database for the qualification of the alternatives
• aggregation preserving the preference of the criteria

• group decision making
• aggregation of individual ranking weighted by voting power values

• sensitivity analysis
• investigation of the effect of the weights, voting powers, individual qualification values and their uncertainties on the group weights, the group qualifications and the ranking of the alternatives

Software environment

• WINDOWS, EXCEL

1. Csáki, P., Rapcsák, T., Turchányi, P, Vermes, M.: Research and development for group decision aid in Hungary by WINGDSS, a Microsoft Windows based group decision support system, Decision Support Systems 14 (1995) 205-21.
2. Mészáros, Cs., Rapcsák, T.: On sensitivity analysis for a class of decision systems, Decision Support Systems 16 (1996) 231-240.
3. Rapcsák, T., Sági, Z., Tóth, T., Kétszeri, L.: Evaluation of tenders in information technology, Decision Support Systems 30 (2000) 1-10.

• Consultancy in Tender Evaluation for Hungarian Power System Operator Company
• Comparative analysis of the alternatives of the subway line 4. in Budapest 
• Environmental impact assessment for the revitalization of Lake Sasfészek at the village of Páty
• Ranking of Projects related to Raiffeisen Bank
• Office of Prime Minister: Evaluation of 17 Parallel Informatics Tenders
• Ministry of the Interior: Evaluation of Informatics Tenders and Evaluation of Software Tenders for Referendum
• Hungarian Post Office Ltd.: Installation of WINGDSS
• Tender Bureau of the Hungarian Telecommunication Company: Installation of WINGDSS
• Ministry of Welfare: Evaluation of Social Tenders
• Ministry of Environment and Regional Planning: Environment Impact Assessment,
• Evaluation of the State of the Environment Evaluation Strategies

For more information please contact
Tamás Rapcsák, Room L512. 
Department of Operations Research and Decision Systems 
e-mail: rapcsak@sztaki.hu
phone: (36-1) 279-6113, fax: (36-1) 209-5267 

TRANSOPT: database manager and transport optimizer

Aims of TRANSPORT

• handling the data of gas cylinder delivery
• assigning the customers to filling stations
• scheduling the daily delivery

Database of TRANSPORT

The database of 

• settlements in Hungary
• the Hungarian road network
• filling stations
• bulk consumers
• sales points
• vehicles

Consumers and filling stations

• The consumers are arranged by filtering.
• For each consumer, the shortest route to the filling station is computed.
• The consumers are assigned to filling stations.
• Sales points are assigned to filling station regarding the lower and upper capacity limits of the latter, as well.

• Demands should be satisfied possibly fully. In some cases, ranking of consumers must be taken into account.
• Tracks may have two routes daily, the working time, however, may be restricted.

• The maximization with respect to the fulfilment of the daily demands and
• the minimization of overall costs of vehicles' routes should be achieved simultaneously.

• Consumers' chain served by each route of vehicles
• Detailed timetable of vehicles

• New and/or modified data may be archived.
• Data concerning the demand may undergo to a statistical analysis.
• Based on the data, forecast is possible.

• User-friendly frontends for both the data input and displaying the data are provided.
• Data management functions are easily available.
• The geographical information may be visualized during data input, modification and merge. Graphic displays have search, editor and zoom functions.
• The results, assignments, routes are demonstrated on a map.

1. Fölsz, F., Mészáros, Cs., Rapcsák, T.: Distribution of gas cylinders, European Journal of Operational Research 87 (1995) 613-623.
2. Fölsz, F., Mészáros, Cs., Rapcsák, T.: Transport optimization of gas cylinders, Zeitschrift für Angewandte Mathematik und Mechanik 46 (1996) 425-426.

Decision support of electricity planning

Production costing simulation of electricity with multiple group-limitations

A probabilistic model is elaborated for production costing simulation of electricity with multiple group-limitations. It serves as a tool in 

• medium- and long-term planning,
• planning the expansion of production and capacity,
• fuel resource scheduling

Scheduling the pumped storage power plants

A model and the corresponding methodology for simulating the operations of a pumped storage plant have been elaborated for medium- and long-term production costing simulation. 

Construction of load duration curve

• Peak load forecast uncertainty is handled in the probabilistic model used in system-level reserve calculations and in production costing simulation.
• The preservation of the main characteristics of a load duration curve, such as monotonicity, energy, peak and minimal load, is ensured by a new polynomial approximation based on semiinfinite programming.
• The construction, based on dynamic programming, of piecewise constant approximations of load duration curves used in capacity expansion planning.

In electricity planning programs, for the approxima-tion of load duration curves, the load data are subject to a treatment aiming at 

• Automatic computation of the missing but computable data.
• Correction of the contradictory data elimination or inconsistency. The latter is done by an automatic concistency analyser.

The database and the programs implemented in SAS Information System aim at 

• integrating the data coming from different sources into a unified database
• analysing the consistency
• supporting the monthly technological analysis.

• Hungarian Power Companies
• International Atomic Energy Agency
• DECADES research project
• WASP-IV program package
• ENEP electricity planning program package
• Argonne National Laboratory, USA

1. Fülöp, J., Hoffer, J., Dörfner, P.: LDC Module: A load data based tool in electricity planning, International Journal of Electrical Power and Energy Systems 14 (1992) 5, 307-312.
2. Hoffer, J., Prill, M.: On the models of peak load uncertainty in probabilistic production costing algorithms, International Journal of Electrical Power and Energy Systems 18 (1996) 6, 153-160

Teaching activity

Our teaching activity is closely related to research.

• The main objective of the Department of Decisions in Economics, started in 1995, and operating as a Department of the CORVINUS University of Budapest, located at the Computer and Automation Institute is the model and case-oriented, high-level education, based on the different branches of operations research and decision theory. The courses are held in the well-equipped Laboratory of Education at the Institute.

• Education at the Technical University of Budapest and at L. Eötvös University of Sciences as well as at their Doctoral Schools aims at making the students be acquainted with the recent research results and the relevant subjects, preparing them for research activity and involving postgraduate students into the ongoing projects.

Decisions in Economics

Department of the
CORVINUS University of Budapest
at the Computer and Automation Institute, Hungarian Academy of Sciences

Attention in education is focused on models, modelling and case-studies, which provides assistance in solving real-life problems. Besides the continuous, discrete, 
stochastic and dynamic problems, furthermore, the related case-studies, great emphasis is laid on developing individual models, their concrete realization by the help 
of widely used software systems. Ground for all this is ensured through the even internationally significant application activity pursued for several decades at the 
Computer and Automation Institute, Hungarian Academy of Sciences. 

Practice is organized in the lecture-room of the Institute where each student works on Intel Celeron 2 GHz, 512 MB RAM - Windows 2000, with the application softwares most appropriate for the curriculum (optimization models: GAMS; forecast models: MINITAB; multi-criteria decision models: EXPERT CHOICE, DECISION LAB 2000, WINGDSS; dynamic models: MATLAB). 

The branch of studies is equal to training in "operations research and system analysis" and provides with favourable opportunities for finding jobs after the students 
have graduated, as managers system and/or production engineers, mostly in industrial spheres, public administration, insurance companies, banks and financial 
institutions. 

For more information please contact
Prof. Tamás Rapcsák 
Head of Department 
Tel: (36-1) 279-6113, Fax: (36-1) 209-5267 
e-mail: rapcsak@sztaki.hu
CAI, HAS, H-1111 Budapest, Lágymányosi u.11.