Second six-monthly periodic report

Program

10:00-10:10	Megnyitó
I. szekció: Sun Grid Engine
II. szekció: Az NIIF Sun E10000 szuperszámítógép első éves tapasztalatai
III. szekció: Grid rendszerek
16:50-17:00	Zárszó Conclusions

Location: Budapest, XIII. ker., Victor Hugo utca 18-22., PC-Cluster room

Parallel, Distributed and Network-based Processing
PDP-2002

Gran Canaria Island
January 9th-11th, 2002

No. of participants: 60

No. of Hungarian participants: 3

The best 5 papers are published by the Euromicro Journal of System Architectures (JSA)
This event was organized and scientifically managed , chaired and organized by MTA SZTAKI.
Workpackage progress report for Workpackage No.: 5

Formal language, agent based computing
Erzsebet Csuhaj-Varju, Ph. D. senior researcher, head,

Research Group on Modelling Multi-Agent Systems

An active international co-operation is undertaken by this group. Several key experts ‘s visit to our Institute is being prepared.

The ERCIM Consortium has helped us to recruit young scientists from European partner research institutes, to come and work in MTA SZTAKI. Maurice ter Beek will start his work in March.

Prof. Markus Holzer will also join this lab for March and April.
Prof. Victor Mitrana has postponed his visit to the second half of 2002.


E-mail:[e-mail],

[e-mail]
http://www.sztaki.hu/~csuhaj/

http://www.sztaki.hu/mms/

Workpackage progress report for workpackage No.WP 6

High-performance scientific computing
We have carried out (the planned) data mining experiments to test various algorithmic ideas related to the APRIORI algorithms. We worked with datasets obtained from a major Hungarian Internet service provider. We have written a research paper and a tutorial related to the subject.
F. Bodon, L. Ronyai: Trie: an alternative data structure for data mining; manuscript, submitted to Computers and Math. with Applications;

F. Bodon: Data mining algorithms (in Hungarian) 1--90. A draft collection of lecture notes. It is still under development. Part of the material has been used at several university courses in Hungary (ELTE, BMGE). The latest version available at:

http://www.mit.bme.hu/~bodon

REGARDING WORKPACKAGE B:

The members of group played active role in university education related to our field. We have related courses at Budapest University of Technology, Central European University, Eotvos Lorand University. We have given seminar lectures at the Technical University Budapest, Szeged University on related topics.

REGARDING WP C

Ulrich Meyer, Max Planck Institute, Saarbrucken visited us in Jan-Feb 2002.
We worked on problems of joint interest on external memory algorithms. He gave a lecture on his speciality at SZTAKI and gave a highly successful tutorial at the Algorithms-class of L. Ronyai at BMGE. These - in addition to students of the class - were attended by some specialists from industry.

To respond to the great interest here on outer memory computations, Dr. Meyer established a webpage of iportant links on the subject, which is designed to help beginners to acquire rudimental information in the field.

Dr. Tamas Lukovszki, Univ. of Paderborn, March-April 2001.
We worked on algorithmic issues related to routing problems in ad hoc mobile communication networks. He gave an insightful overview of his work, and joint research with Andras Benczur.

REGARDING WP D:

We are active in the oragnization of the largest Hungarian Conference related to computing: Informatics in Higher Education IF'2002 (please see their webpage at www.date.hu/if2002).

Janos Demetrovics is the president of the conference, Lajos Ronyai is a memebr of the PC. At the latest PC-meeting we finalized the scientific program.

REGARDING WORKPACKAGE E:

Dr. Gabor Ivanyos visited Univ. Paris Sud, Orsay in November-December 2001, and Jan-Feb 2002.

Dr. Katalin Friedl Univ. Paris Sud, Orsay February-June 2002.
The objective of these visits is to set up a long term cooperation in the field of quantum computations.

We have a joint paper: G. Ivanyos, M. Santha: Efficient quantum test for abelian groups. (manuscript, 2002)

PLANS to WP6:

We intend to develop solutions to subproblems related to data mining applications. We intend to use these to demonstrate (to students, industry partners) the applicability of scientific computing results and paradigms. We plan to maintain our strong presence in university education of these subjects.
To WP E:

June, 2002 Dr. Ronyai visits the ETH of Zurich The purpose of the visit is discussion of issues in symbolic computation with the Swiss partner.

To WP D:

Successful organization of IF'2002 in August 2002.

Workpackage progress report for Workpackage No.: 7

SZTAKI in PLANET: An Introduction

L. Monostori and J. Váncza

The Computer and Automation Research Institute of the Hungarian Academy of Sciences – SZTAKI in short – performs basic and application-oriented research in the fields of computer science, intelligent systems, process control, wide-area networking and multimedia. The Institute is active in graduate and post-graduate education, cooperating with most of the technical universities in Hungary. The European Research Consortium of Informatics and Mathematics (ERCIM) granted full membership to the Institute in 1994. SZTAKI is a PLANET node since 2000.
In SZTAKI, members of the Laboratory of Engineering and Management Intelligence have been working for more than a decade on planning, scheduling and production management problems of manufacturing. The Laboratory has a long-standing interest in modular design, manufacturing process planning, distributed scheduling, holonic manufacturing systems, agent-based control and the simulation of production systems, as well as in the management of production networks. The Laboratory has regularly contributed to the Holonic Manufacturing Systems (HMSs) project of the worldwide Intelligent Manufacturing Systems (IMS) initiative.
Our recent planning and scheduling related research topics are as follows:
Emergent production planning and scheduling We have an interest in the emergent control of manufacturing systems that operate under uncertain and complex conditions [6]. We have developed a multi-agent model for solving integrated order processing and dynamic production scheduling problems. The model is based on the economic rationality of individual agents, though, by the application of a special incentive mechanism, reconciles autonomous and cooperative behavior. Hence, the model paves the way toward so-called holonic manufacturing systems [1,7]. Further on, we have developed an adaptable order selection and resource allocation model where production balancing is made by reinforcement learning.
Constraint-based computer-aided manufacturing process planning Recently, we have modeled computer-aided manufacturing process planning (CAPP) as a large-scale constrained optimization problem.The model captures several kinds of resource, operation ordering and grouping (so-called setup) constraints. Two planner systems have been built on the top of general-purpose constraint programming systems [2,8]. The planners can cope with conditional and antagonistic (soft) constraints as well. The model is generic: it was validated in the domains of machining prismatic parts and of bending sheet metals. Recently, customized search methods have been developed that combine constraint satisfaction, branch-and-bound search and multi-dimensional Pareto optimization. In the bending domain, planning was integrated with geometric reasoning.

Modelling industrial manufacturing systems: A method was developed to model complex production process and process chains. The method that combines simulation and artificial neural network techniques was applied in the optimization of process parameters [5]. We participate also in an international research consortium whose goal is to establish a virtual institute focused on knowledge sharing and integration of competencies in modelling and designing industrial manufacturing systems. This project applies on-line distance learning (ODL) techniques to develop a portal applicable both for research and educational purposes.
The follow-up of these works is a currently launched mid-term national R&D project on digital factories. The project, which is running with substantial industrial participation, focuses on resource-constrained project management, production planning, tele-presence and interactive multimedia, as well as on the monitoring of complex production processes.
SZTAKI helped to organize international conferences on the application of AI in manufacturing and engineering - such as the 2nd World Congress on Intelligent Manufacturing Processes and Systems [3] and IEA/AIE-2001, the 14th International Conference on Industrial & Engineering Applications of Artificial Intelligence & Expert Systems [4].

References
[1] Bongaerts, L.; Monostori, L.; McFarlane, D.; Kádár, B.: Hierarchy in distributed shop floor control. Computers in Industry, 43(2),123-137, (2000).

[2] Márkus, A., Váncza, J.: Process planning with conflicting and conditional advice. Annals of the CIRP, 50(1), 327–330, (2001).

[3] Monostori, L. (ed.), Proc. of the 2nd World Congress on Intelligent Manufacturing Processes and Systems, Budapest, June 1997, Springer.

[4] Monostori, L., Váncza, J., Ali, M. (eds.): Proc. 14th Int. Conf. on Industrial & Engineering Applications of Artificial Intelligence & Expert Systems, Budapest, June 2001, Springer LNAI 2070.

[5] Monostori, L.; Viharos, Zs.J.: Hybrid, AI- and simulation-supported optimisation of process chains and production plants. Annals of the CIRP, 50(1), 353-356, (2001).

[6] Ueda, K.; Márkus, A.; Monostori, L.; Kals, H.J.J.; Arai, T.: Emergent synthesis methodologies for manufacturing. Annals of the CIRP, 50(2), 535-551, (2001).

[7] Váncza, J., Márkus, A.: An agent model for incentive-based production scheduling. Computers in Industry, 43(2), 173–187, (2000).

[8] Váncza, J., Márkus, A.: A constraint engine for manufacturing process planning. In: Walsh, T. (ed.): Principles and Practice of Constraint Programming - CP 2001, Springer LNCS 2239, 745–759, (2001).

Links

www.sztaki.hu

www.sztaki.hu/ake/ai/

www.sztaki.hu/conferences/ieaaie2001/

http://www-lag.ensieg.inpg.fr/~vimims/main.htm

Laboratory of Engineering and Management Intelligence