Curriculum Vitae for Per Kreuger

 

Professional experience and involvements

 

During my early years at SICS I worked mainly with computational logics and logic programming languages. I was involved in the specification of the GCLA programming language and modeling tool and in developing the core algorithms used in its implementation. That was also the subject of my Fil. Lic. and Ph.D. (abstract) thesi. Since then I worked primarily with scheduling, resource allocation and matching problems, using mainly constraint programming and other combinatory reasoning and optimisation techniques.

 

  • For a period around 1994-1995 I was involved in the ALBAM project with Pharmacia Biotech applying Markov modeling techniques in DNA analysis. We trained Markov models to interpret the signals from a electrophoresis instrument as sequences of DNA bases (DNA sequencing). Within the relatively short duration of the project we were able to reproduce and in some cases improve the results of a the specialized algorithm used in the commercial product sold by Pharmacia Biotech at the time.
  • In 1994 I was involved in the modeling and specification phase of the MITOS project which originally aimed to integrate a production planning system for the steel plant of OVAKO Steel in Hofors with mechanisms to determine suitable production parameters, such as batch sizes, storage volumes and cycle times, for given prognosi of demand and other economic parameters. This project was jointly financed by OVAKO Steel and NUTEK.
  • The MITOS project later turned its focus to the economic model and choose to ignore the scheduling issues but the production planning activities continued in the SICS financed continuation project MISC (during 1995-97). In this project we focused on the combined problem of finding a production schedule and an allocation of the produced material to (customer or internal) orders. In this period my interest moved more an more from an AI approach (Agent modeling) to a more low level and technical focus on basic mechanisms for scheduling and resource allocation such as those used in constraint programming.
  • Toward the end of 1996 we initiated actives (within the TUFF project) around using constraint programming techniques for transport planning in railway networks This was done in cooperation with the Swedish State Railways (SJ). Since then this has been one of my main concerns as this turned out to be a very rich domain full of interesting sub problems. In the planner implemented within the project we use a variant of the job shop scheduling model with two alternate durations depending on the order chosen between the tasks. This model and its implementation was presented at two workshops: the "International Logic Programming Symposium 1997 (ILPS'97)" in Long Island, USA 1997 and the "International Conference the Principles and Practice of Constraint Programming (CP 97)" in Linz, Austria 1997.
  • The production planning activities meanwhile continued within the Esprit project TACIT again with OVAKO Steel as partner. This time the focus is on actually implementing a full scale scheduler for the ingot rolling facilities in Hofors using state of the art constraint programming techniques. The aim of the project is to verify that constraint programming is a mature technology that can in fact be used to solve real life problems.
  • The TUFF activities continued throughout 1998 and 1999 focusing on coordination of different planning processes. Constraint Programming (CP) remained our main tool but techniques from traditional Operations Research (OR) was introduced for some subproblems. Research on integrating techniques from these two frameworks was conducted in cooperation with Mats Carlsson in the Combinatorial Problem Solving and Prolog Technology group. During 2002-2003 a selection of methods developed within the TUFF project has been further developed and introduced into the production environment of Green Cargo within the LOOP project.
  • In the IPLDOPT project, conducted in cooperation with and funded by Telia Research, members of the DSPS group and the Computer and Network Architectures (CNA) Laboratory applied flow optimisation techniques to packet routing in IP networks.
  • The HEMCO project funds two doctoral students (Markus Bohlin and Waldemar Kocjan) at SICS Västerås office which I supervise. I have also developed and taught a course in Constraint programming at Mälardalens Högskola in Västerås. Course materials in are available here.
  • The GENFUNC project studied an optimal matching problem in gene expression analysis in cooperation with the startup biotechnology company Global Genomics. I contributed to the domain modeling and in particular a constraint model and matching algorithm based on constraint programming. This model and solver were never fully explored since an Integer Programming (IP) model developed in parallel was found to work sufficiently well.
  • The KABAN project, initiated within DSPS group, is funded the R&D Program of Swedish National Rail Administration (Banverket) with the aim to develop standardized methods of capacity assessment and analysis for signaling design and capacity planning. My contributions has so far been mainly minimal headway and traversal time computation and approximation methods.
  • The SPOK project, initiated within DSPS group, is funded the R&D Program of Swedish National Rail Administration (Banverket) with the aim to develop application formats and conflict detection and solution methods for the timetabling process in the presence of several operators on the shared rail track resource.