Bernd Heidergott                                               

 

Affiliation                   Vrije Universiteit
Faculty of Economics and Business Administration
Department of Department of Econometrics and Operations Research

Research fellow          Tinbergen Institute and EURANDOM.
___

E-mail: bheidergott [at] feweb [dot] vu [dot] nl
Phone: 0031 +20 +5986016
Fax: 0031 +20 +5986020




    Bernd Heidergott has received the Best Lecturer Award of the faculty of Economics and Business Administration of the VU for the academic year 2008/2009.




Index: Research Activities, Publications, Career, Grants and Fellowships and Teaching


Research Activities:

My main current research directions are as follows:

Gradient Estimation is concerned with providing unbiased gradient estimators for stochastic networks. We have worked on the main approaches in this area such as infinitesimal perturbation analysis (IPA), smoothed perturbation analysis (SPA), finite perturbation analysis (FPA), rare perturbation analysis (RPA), score function (SF) and phantom estimators (originating from measure-valued differentiation). In case studies comparing the performance of the estimators we could show that phantom estimators have the potential to outperform single-run gradient estimators like IPA, SPA of SF (which goes against the common belief in the simulation community). Recently we obtained first analytical results supporting this finding. This is joint work with F. Vázquez-Abad.

Differentiation theory for Markov chains seeks sufficient conditions for differentiability of performance characteristics of Markov chains and tries to obtain closed-form analytical expressions for the derivatives. The research in this area is the driving force in our attempt to unify gradient estimation and to develop a deeper understanding of stochastic optimization. We have intensively worked on this topic and could show that MVD in deed allows for a unified approach to gradient estimation. A recent breakthrough result is that we could show existence of derivatives of a particular class of Markov chains (phase-homogeneous random walks) under substantially weaker conditions than known in the literature. Furthermore, we have established an important link between differentiation theory and Banach space theory. This is joint work with A. Hordijk. An overview on the state-of-the-art in MVD can be found here (pdf file) .

 

Taylor series expansions are a powerful tool in performance analysis. By evaluating a finite number of higher-order derivatives, Taylor series allow to obtain the performance characteristic as a function of the parameter of interest. Key question are that of (1) convergence of the (finite) Taylor series to the true performance function and that of (2) establishing bound on the remainder term (i.e., the error made by approximating the true performance function via a finite Taylor polynomial). We have established results on (1) for max-plus linear systems (see below) and Markov chains. We achieved a breakthrough result on (2) by showing that for finite state-space Markov chains the series can be efficiently computed and a sharp bound on the error term can be obtained. This is joint work with A. Hordijk.

Max-plus algebra is an algebraic approach to discrete event dynamic systems, like queuing networks, that are prone to synchronization. Max-plus algebra has been the main theme of the European research project ALAPEDES. Based on the weak differentiation approach, we developed a calculus of weak differentiation for max-plus-linear stochastic systems. Based on this framework, we established Taylor series expansions for max-plus linear systems. Together with G. J. Olsder and J. van der Woude we have written a book on deterministic max-plus algebra. Our work on stochastic max-plus algebra and Taylor series expansions has been published in a monograph. Recent work is concerned with ergodic theory of stochastic max-plus linear stochastic systems. For more on max-plus algebra please visit the max-plus web portal. Slides of a tutorial held on max-plus algebra at the IFORS 2002 conference in Edinburgh, UK, can be found here (pdf file).

Index


Publications:

Books:

Max-Plus linear Stochastic Systems and Perturbation Analysis.
The International Series of Discrete Event Systems 15, Springer,
ISBN: 978-0-387-35206-0 (2007). (contact publisher)

Max Plus at Work - Modeling and Analysis of Synchronized Systems (with G. J. Olsder and J. van der Woude)
Princeton Series in Applied Mathematics, Princeton University Press,
ISBN: 978-0-691-11763-8 (2006). (contact publisher)
(updates/corrections) .

Train Movement Analysis at Railway Stations: Procedures and Evaluation of Wakob's Approach (with A. de Kort, R. van Egmond and G. Hooghiemstra)

Studies in Transport Series No. S99/1, TRAIL Research School,

ISBN: 90-407-1855-5 (1999). (contact publisher)

Sample-Path Analysis for Stochastic Networks (in German)

Forschungsergebnisse zur Informatik 10, Verlag Dr. Kovac,

ISBN: 978-3-86064-086-9 (1996). (contact publisher)

 

Contributions to books

 

A paradigm for derivatives of positive systems.
In Lecture Notes in Control and Information Sciences, vol. 294, pages 312-328, Springer,

ISBN: 3-540-4034206 (2003).

Weak Differentiation and Gradient Estimation for Discrete Event Processes (with  F. Vázquez-Abad).
In Discrete Event Systems: Analysis and Control, pages 433-440, Kluwer,

ISBN: 0-7923-7897-0 (2000)

GSMP with discrete lifetime distributions.
In Lecture Notes in Control and Information Sciences vol. 199, pages 456-462. Springer, ISBN: 0-387-19896-2 (1994).


Journal Papers:

Weak differentiability of product measures (with H. Leahu) Mathematics of Operations Research (in press)

Gradient estimation for a multi-component maintenance system using measure-valued differentiation (with T. Farenhorst-Yuan) Operations Research (in press)

Series expansions for continuous-time Markov chains (with A. Hordijk and N. Leder) Operations Research (in press)

An Approximation Approach for the Deviation Matrix of Continuous-Time Markov Processes with Application to Markov Decision Theory (with N. Leder and A. Hordijk). Operations Research (in press)

A perturbation analysis approach to phantom estimators for waiting times in the G/G/1 queue (with F. Vázquez-Abad and T. Farenhorst-Yuan). Discrete Event Dynamic Systems (accepted January 2009)

Gradient estimation for discrete-event systems by measure-valued differentiation (with F. Vázquez-Abad
, G. Pflug and T. Farenhorst-Yuan). Transactions on Modeling and Computer Simulation (accepted January 2009).

Strong bounds on perturbations (with H. Leahu).
Mathematical Methods of Operations Research vol. 70, pages 99-127, 2009.

Gradient estimation for a class of systems with bulk arrivals: a problem in public transportation (with F. Vázquez-Abad).
Transactions on Modeling and Computer Simulation
vol. 19, issue 3, article 13 (27 pages), 2009.

Derivatives of Markov kernels and their Jordan decomposition (with A. Hordijk and H. Weiβhaupt)
Journal of Applied Analysis
vol. 14, pages 13-26, 2008.

Measure valued differentiation for Markov chains (with F. Vázquez-Abad).
Journal of Optimization and Applications vol. 136, pages 187-209, 2008. (Web Supplement)

Sensitivity Estimation for Gaussian Systems (with F. Vázquez-Abad and W. Volk-Makarewicz).
European Journal of Operational Research
vol. 187, pages 193-207, 2008.

A Note on Gradient Estimation for Maintenance Systems (with T. Yuan).
IEEE Transactions on Automatic Control vol. 52, pages 2131-2145, 2007.

Complexity reduction in MPC for stochastic max-plus-linear discrete event systems by variability expansion (with T.J.J. van den Boom and B. de Schutter).
Automatica
vol. 43, pages 1058-1063, 2007.

Series expansions for finite-state Markov chains (with A. Hordijk and M. van Uitert).

Probability in Engineering and Informational Sciences vol. 21, pages 381-400, 2007.

Measure-valued differentiation for random horizon problems (with F. Vázquez-Abad).

Markov Processes and Related Fields, vol. 12, pages 509-536, 2006.

Measure-valued differentiation for stationary Markov chains (with A. Hordijk and H. Weiβhaupt)
Mathematics of Operations Research, vol. 31, pages 154-172, 2006.

Single-run gradient estimation via measure-valued differentiation (with A. Hordijk).
IEEE Transactions on Automatic Control , vol. 49, pages 1843-1846, 2004

Taylor series expansions for stationary Markov chains (with A. Hordijk).
Advances in Applied Probability, vol. 23, no. 4, pages 1046-1070, 2003.

See also Correction: Taylor series expansions for stationary Markov chains;
Advances in Applied Probability, vol. 36, no. 4, page 1300, 2004.

A probabilistic approach for determining railway infrastructure capacity (with A. de Kort, and H. Ayhan).

European Journal of Operational Research
, vol. 148, pages 644-661, 2003.

A note on the relation between realization probabilities and weak derivatives (with X.R. Cao ).
IEEE Transactions on Automatic Control, vol. 47, pages 1112-1115, 2002.

Towards a control theory for transportation networks (with R. de Vries).
Discrete Event Dynamic Systems, vol. 11, pages 371-398, 2001.

A differential calculus for random matrices with applications to (max,+)-linear stochastic systems.
Mathematics of Operations Research
, vol. 26, pages 679-699, 2001.

Option pricing via Monte Carlo Simulation: A weak derivative approach.
Probability in Engineering and Informational Sciences, vol. 15, pages 335-349, 2001.

A weak derivative approach to optimization of threshold parameters in a multi-component maintenance system.
Journal of Applied Probability, vol. 38, pages 386-406, 2001.

Analyzing sojourn times in queuing networks: a structural approach.
Mathematical Methods of Operations Research, vol. 52, pages 115-132, 2000.

A characterization for (max,+)-linear queuing systems.
Queuing Systems: Theory and Applications, vol. 35, pages 237-262, 2000.

Customer-oriented finite perturbation analysis for queuing networks.
Discrete Event Dynamic Systems, pages 201-232, 2000.

Optimization of a single-component maintenance system: a smoothed perturbation analysis approach.
European Journal of Operations Research
, pages 181-190, 1999.

Infinitesimal perturbation analysis for queuing networks with general service time distributions
Queuing Systems: Theory and Applications, pages 43-58, 1999.

The zero utility principle for scale families of risk distributions ( with D. Pfeifer).
Deutsche Gesellschaft fürVersicherungsmathematik, pages 711-722, 1996.

Sensitivity analysis of a manufacturing workstation using perturbation analysis techniques.
International Journal of Production Research, vol. 33, pages 611-622, 1995.

Conferences with Refereed Proceedings:

Railway timetable stability analysis using stochastic max-plus linear systems (with R. Goverde, and G. Merlet). Proceedings 3rd International Seminar on Railway Operations Modelling and Analysis, Zürich, Switzerland, pages 1-18, 2009.

Perturbation analysis of Markov chains. International Workshop on DES (WODES' 08), Göteborg, Sweden, pages 99-104, 2008.

Stochastic optimization for control for maintenance systems (with T. Farenhorst-Yuan). International Workshop on DES (WODES' 08), Göteborg, Sweden, pages 93-98, 2008.

A fast approximation algorithm for the Lyapunov exponent of stochastic max-plus systems (with R. Goverde and G. Merlet). International Workshop on DES (WODES' 08), Göteborg, Sweden, pages 49-54, 2008.

Bounds on Perturbations for DES (with A. Hordijk and H. Leahu)
International Workshop on DES (WODES 06), Ann Arbor, USA, pages 378-383, 2006.

Asymptotic growth rate of stochastic max-plus systems that with a positive probability have a sunflower-like support (with J. van der Woude).
International Workshop on DES(WODES 06), Ann Arbor, USA, pages 451-456, 2006.

Series Expansions of Generalized Matrix Products (with H. Leahu).
Proceedings of the CDC, Seville, Spain, pages 7793-7798, 2005.

On choosing the right gradient estimator.
5th St. Petersburg Workshop on Simulation, St. Petersburg, Russia, June 26 - July 2, pages 313-318, 2005.

An ergodic theorem for stochastic max-plus linear systems (with J. van der Woude and G. J. Olsder).
International Workshop on DES (WODES'04), Reims, France, pages 97-102, 2004.

Gradient estimation for a problem in public transportation: A comparison of SPA, SF and MVD (with F. Vázquez-Abad).
International Workshop on DES (WODES'04), Reims, France, pages 241-246, 2004.

A paradigm for derivatives of positive systems.
Lecture Notes in Control and Information Sciences, vol. 294, Springer, pages 312-328, 2003.

Complexity reduction in MPC for stochastic max-plus-linear systems by variability expansion (with T.J.J. van den Boom and B. de Schutter).
Proceedings of the CDC, Las Vegas, Nevada, USA, pages 3567-3572, 2002.

Variability expansion for performance characteristics of (max,plus)-linear systems.
International Workshop on DES (WODES'02), Zaragoza, Spain, pages 245-250, 2002.

Sensitivity analysis of joint characteristics of (max,+)-linear queuing networks.
IFAC satellite workshop on (max,+) algebra, Prague, Czech Republic, pages 221-226, 2001.

Bounding the Coupling Time of (Max,+)-Linear Systems (with Soto y Koelemeijer).
Transport, Infrastructure and Logistics, 6th TRAIL annual congress, The Hague, the Netherlands, 2000.

Weak differentiation and gradient estimation for discrete event processes (with  F. Vázquez-Abad).
Discrete Event Systems: Analysis and Control, Kluwer Academic Publisher, proceedings of the WODES (WODES'00), Gent, Belgium, pages 433-440, Kluwer, Boston, 2000.

Optimization of synchronization constraints via weak derivatives.
International Workshop on DES (WODES'98), Cagliari, Italy, pages 261-266. IEE, London, 1998.

Modeling and Analysis of queuing processes at railway stations: a case study (with R.-J. van Egmond, A. de Kort, and G. Hooghiemstra).
Transport, Infrastructure and Logistics, 4th TRAIL annual congress, The Hague, the Netherlands, 1998.

An overview of waiting time approximations for single server queues (with R.-J. van Egmond, A. de Kort, and  G. Hooghiemstra).
Transport, Infrastructure and Logistics, 4th TRAIL annual congress, The Hague, the Netherlands, 1998.

A stochastic minimal headway model for trains (with R.-J. van Egmond, A. de Kort, and G. Hooghiemstra).

Transport, Infrastructure and Logistics, 4th TRAIL annual congress, The Hague, the Netherlands, 1998.

GSMP with discrete lifetime distributions.
Lecture Notes in Control and Information Sciences, vol. 199, Springer, London, pages 456-462, 1994.

Sensitivitätsanalyse eines Fertigungssystems mit Hilfe der infinitesimalen Perturbationsanalyse (Sensitivity analysis for a manufacturing system using IPA).
ASIM91, Hagen, Germany, 1991.

Infinitesimal perturbation analysis: an overview.
Methods of OR, vol. 64, Vienna, Austria, pages 163-172, 1990.

Perturbation analysis, concept of an implementation.
SCS European Simulation Multiconference, Erlangen-Nürnberg, Germany, pages 192-197, 1990.

Index


Career:

from 9/2002 until today Department of Econometrics and Operations Research, Faculty of Economics and Business Administration, Vrije Universiteit Amsterdam

from 3/2001 to 8/2002 Operations Research and Statistics, Department of Mathematics and Computer Science, Division of Mathematics, TU Eindhoven

from 8/1999 to 2/2001 EURANDOM

from 4/1997 to 7/1999 DIAM, Department of Information Technology and Systems

from 5/1996 to 3/1997 Econometric Institute, Faculty of Economics, Erasmus University Rotterdam

from 3/1992 to 4/1992 Center of Mathematical Statistics and Stochastic Processes, Department of Mathematics, University of Hamburg (Ph.D. student)

Index


Grants and Fellowships:

STW grant for research project Modeling and Analysis of Operations in Railway Networks: the Influence of Stochasticity. This is a joint project together with F.M. Dekking, I. Hansen and L. Meester, which will run from October 2003 to September 2007.

Deutsche Forschungsgemeinschaft (DFG) Fellowship, from August 1, 1999, to July 31, 2001.

Verein zur Förderung der Versicherungswissenschaft in Hamburg , Fellowship from October 1, 1995, to May 31, 1996.

Wilhelm-Blaschke-Stiftung, travel grants, June 1994 and July 1995.

Index


Teaching:

At the Vrije Universiteit Amsterdam I am mainly involved in teaching mathematics and statistics for economists. In addition, I am teaching a course on Convex Analysis and Optimization for econometricians.

Lecture "Convex Analysis and Optimization"

The complete syllabus for the lecture is here (handout).

The handouts for week 1 of the lecture here (handout).

The handouts for week 2 of the lecture here (handout).

The handouts for week 3 of the lecture here (handout).

The handouts for week 4 of the lecture here (handout).

The handouts for week 5 of the lecture here (handout).

The handouts for week 6 of the lecture here (handout).

The Matlab file containing some useful plotting functions can be found here (m-file). For use change the extension from "txt" to "m".

The exam 2006 with solutions can be found here (exam-2006).