Work address: Office G3.06, Middelheimlaan 1, B-2020 Antwerp, Belgium
Work phone: +32 3 265 38 95
Email: siegfried.cools@uantwerp.be
Date and place of birth: (available on demand)
Nationality: Belgian
Marital status: (available on demand)
Driving license: Category B
Main areas of expertise
Preconditioning and iterative solution techniques for forward and inverse wave scattering problems
(Helmholtz equation, Schroedinger equation, Seismic inversion, Computed Tomography)
Communication-reducing Krylov subspace methods for massive scale linear systems on HPC distributed memory hardware
(Parallel performance, Latency hiding, MPI-based implementation)
Numerical analysis (stability, robustness, convergence delay, etc.) of the impact of local rounding error propagation on Krylov subspace methods and their HPC communication-reducing variants
Other general interests: numerical optimization, large scale eigenvalue problems, matrix factorization algorithms, graph partitioning algorithms, ...
Short academic resume
2016-present:
Postdoctoral Fellow FWO Flanders, Applied Mathematics Research Group, Dept. of Mathematics and Computer Science, University of Antwerp
Project 12H4617N - Scalable and error resilient iterative solvers for large scale linear algebra problems.
2015-2016:
Postdoctoral researcher, Applied Mathematics Research Group, Dept. Mathematics and Computer Science, University of Antwerp.
EXA2CT European project on EXascale Algorithms and Advanced Computational Techniques. www.exa2ct.eu.
2011-2015:
PhD in Mathematics, Applied Mathematics Research Group, Dept. Mathematics and Computer Science, University of Antwerp.
2006-2011:
Graduated (with highest honor) Master in Applied and Financial Mathematics, University of Antwerp.
Publications and preprints
[A1][preprint][manuscript] S. Cools, J. Cornelis, W. Vanroose: Numerically Stable Recurrence Relations for the Communication Hiding Pipelined Conjugate Gradient Method.
Accepted for publication in IEEE Transactions on Parallel and Distributed Systems, May 2019.
[P3] [preprint] S. Cools, J. Cornelis, P. Ghysels, W. Vanroose: Improving strong scaling of the Conjugate Gradient method for solving large linear systems using global reduction pipelining.
Submitted to EuroMPI’19: 25th European MPI User’s Group Meeting, 10 - 13 September 2019, ETH Zürich, Switzerland. ACM, New York, NY, USA.
[A1][preprint] E. Agullo, S. Cools, E.F. Yetkin, L. Giraud, W. Vanroose: On soft errors in the Conjugate Gradient method: sensitivity and robust numerical detection.
Submitted to SIAM Journal on Scientific Computing (SISC), November 2018.
[A1] [preprint][manuscript]S. Cools: Analyzing and improving maximal attainable accuracy in the communication hiding pipelined BiCGStab method.
Parallel Computing, PMAA'18 Special Issue, 86, pp. 16-35, Elsevier, August 2019.
[A1] [preprint]S. Cools: Numerical analysis of the maximal attainable accuracy in communication hiding pipelined Conjugate Gradient methods.
Submitted to SIAM Journal on Scientific Computing (SISC), August 2018.
[A1] [preprint]J. Cornelis, S. Cools, W. Vanroose: The communication-hiding Conjugate Gradient method with deep pipelines.
Submitted to SIAM Journal on Scientific Computing (SISC), January 2018.
[P3] [preprint][manuscript]S. Cools, W. Vanroose: Communication Hiding Pipelined Krylov Methods: On Parallel Performance and Numerical Stability of Pipelined Conjugate Gradients.
Proceedings of the International Conference on Parallel Computing (PARCO 2017), Advances in Parallel Computing, 32, pp. 77-86, IOS Press Amsterdam, Bologna, September 2017.
[A1] [preprint][manuscript]S. Cools, E.F. Yetkin, E. Agullo, L. Giraud, W. Vanroose: Analyzing the effect of local rounding error propagation on the maximal attainable accuracy of the pipelined Conjugate Gradient method.
SIAM Journal on Matrix Analysis and Applications (SIMAX), 39 (1), pp. 426-450, March 2018.
[P3] [preprint][manuscript] E. Agullo, S. Cools, L. Giraud, A. Moreau, P. Salas, W. Vanroose, E.F. Yetkin, M. Zounon: Hard Faults and Soft-Errors: Possible Numerical Remedies in Linear Algebra Solvers.
Conference Proceedings of the 12th International Conference on High Performance Computing in Computational Science (VECPAR 2016), High Performance Computing for Computational Science, 10150, Springer, Cham, pp.11-18, July 2017.
[A1] [preprint][manuscript] S. Cools, W. Vanroose: The communication-hiding pipelined BiCGStab method for the parallel solution of large unsymmetric linear systems.
Parallel Computing, 65, pp. 1-20, Elsevier, July 2017.
[P3] [manuscript]S. Cools, W. Vanroose, E.F. Yetkin, E. Agullo, and L. Giraud: On rounding error resilience, maximal attainable accuracy and parallel performance of the pipelined Conjugate Gradients method for large-scale linear systems in PETSc.
Proceedings of the Exascale Applications and Software Conference, ACM, pp. 20–29, Stockholm, Sweden, April 2016.
[C4] [preprint][manuscript]S. Cools, E.F. Yetkin, E. Agullo, L. Giraud, and W. Vanroose: Analysis of rounding error accumulation in CG to improve the maximal attainable accuracy of pipelined CG.
Technical report. University of Antwerp Dept. Mathematics and Computer Science and INRIA Bordeaux, January 2016.
[B2] [preprint][manuscript] S. Cools, W. Vanroose: On the optimality of shifted Laplacian in the class of expansion preconditioners for the Helmholtz equation.
Book chapter in "Modern solvers for Helmholtz problems", Lahaye, D., Tang, J., Vuik, K. (Eds.), Springer, February 2017.
[A1] [preprint][manuscript] S. Cools, W. Vanroose: A fast and robust computational method for the ionization cross sections of the driven Schrödinger equation using an O(N) multigrid-based scheme.
Journal of Computational Physics, 308, pp. 20-39, December 2015.
[A1] [preprint][manuscript]S. Cools, P. Ghysels, W. van Aarle, J. Sijbers, W. Vanroose: A multi-level preconditioned Krylov method for the efficient solution of algebraic tomographic reconstruction problems.
Journal of Computational and Applied Mathematics, 283, pp.1-16, August 2015.
[P3] [proceedings]S. Cools, B. Reps, W. Vanroose: An efficient multigrid calculation of the Far field map for Helmholtz and Schrödinger equations.
Proceedings of the 11th International Conference on the Mathematical and Numerical Aspects of Waves, pp. 275-278, Gammarth, Tunisia, June 2013.
[A1] [preprint][manuscript]S. Cools, B. Reps, W. Vanroose: An efficient multigrid calculation of the Far field map for Helmholtz and Schrodinger equations.
SIAM Journal on Scientific Computing, 36 (3), pp. B367-B395, July 2014.
[A1] [preprint][manuscript]S. Cools, B. Reps, W. Vanroose: A new level-dependent coarsegrid correction scheme for indefinite Helmholtz problems.
Numerical Linear Algebra with Applications, 21 (4): pp. 513-533, August 2014.
[A1] [preprint][manuscript]S. Cools, W. Vanroose: Local Fourier Analysis of the Complex Shifted Laplacian preconditioner for Helmholtz problems.
Numerical Linear Algebra with Applications, 20 (4): pp. 575-597, August 2013.
[C3] [contact] (Manuscript available on demand) S. Cools: Multigrid methods for wave scattering problems governed by the Helmholtz and Schrodinger equation.
PhD dissertation, University of Antwerp, May 22, 2015.
[C3] [contact] (Manuscript available on demand) S. Cools: Complex Shifted Multigrid Methods on the indefinite Helmholtz equation.
Master dissertation, University of Antwerp, in Dutch, June 2011.
Conference talks and seminars
CSE 2019: [slides] SIAM Conference on Computational Science and Engineering (CSE’19), "Analyzing and Improving Attainable Accuracy for the Communication Hiding Pipelined Conjugate Gradient Method",
Spokane Convention Center, WA, US, February 25 - March 1, 2019.
PMAA 2018: [slides] 10th International Workshop on Parallel Matrix Algorithms and Applications, "Parallel Performance and Numerical Stability of Communication Hiding Pipelined Krylov Subspace Methods",
Zurich, Switzerland, June 27-29, 2018.
SIAM ALA 2018: [slides] SIAM Conference on Applied Linear Algebra, "Parallel Performance and Numerical Stability of Communication Hiding Pipelined Krylov Subspace Methods",
Hong Kong, China, May 7-10, 2018.
PARCO 2017: [slides] International Conference on Parallel Computing, "Hiding Global Reduction Latency in Pipelined Krylov Methods: On Parallel Performance and Numerical Stability
of the Pipelined Conjugate Gradient and BiCGStab Algorithms for Solving Large Scale Linear Systems",
Bologna, Italy, September 12-15, 2017.
GAMM 2017: [slides]17th GAMM Workshop, "Hiding Global Reduction Latency in Pipelined Krylov methods: Parallel Performance and Numerical Accuracy
of Communication-Hiding Krylov Subspace Methods for Solving Large Scale Linear Systems",
Cologne, Germany, September 7-8, 2017.
LBNL 2017: [slides]Lawrence Berkeley National LabCS Seminar, "Latency hiding of global reductions in pipelined Krylov methods: On Parallel Performance and Numerical Accuracy
of Communication Hiding Pipelined Krylov Subspace Methods for Solving Large Scale Linear Systems",
Lawrence Berkeley National Laboratory, Berkeley, CA, US, August 7, 2017.
EXA2CT 2016:[slides]EXA2CT European FP7 Project Meeting, "WP1: Overview of recent solver developments: A general framework for deriving pipelined Krylov methods: application to BiCGStab for large and sparse unsymmetric linear systems"
Bordeaux, France, September 22-23, 2016.
EASC 2016: [slides]Exascale Applications and Software Conference "Latency hiding of global reductions in pipelined Krylov methods"
Stockholm, Sweden, April 25-29, 2016.
SIAM PP 2016: [slides]SIAM Conference on Parallel Processing for Scientific Computing, "Latency hiding of global reductions in pipelined Krylov methods"
Paris, France, April 12-15, 2016.
WAVES 2015:[slides] 12th International Conference on the Mathematical and Numerical Aspects of Wave Propagation "A scalable and robust multigrid-based solver for the far field map of Helmholtz and Schrödinger equations"
Karlsruhe, Germany, July 20-24, 2015.
PREC 2015:[slides]International Conference On Preconditioning Techniques For Scientific And Industrial Applications "A robust computational method for the Schrödinger equation cross sections using an MG-Krylov scheme"
Eindhoven, The Netherlands, June 17-19, 2015.
EMG 2014:[slides]European Multigrid Conference 2014 "A new Level-dependent coarse grid correction scheme for indefinite Helmholtz problems"
Leuven, Belgium, September 9-12, 2014.
ZSS 2014: [slides]SAM Zürich Summer School on Advanced Numerical Methods for Non-local Operators
Participant's presentations: "A Short Research Overview - by Siegfried Cools"
Zürich, Switzerland, August 18-22, 2014.
WAVES 2013:[slides] 11th International Conference on the Mathematical and Numerical Aspects of Wave Propagation "An Efficient Multigrid Calculation of the Far field map for Helmholtz problems"
Gammarth, Tunisia, June 3-7, 2013.
CMCMM 2013:[slides] 16th Copper Mountain Conference on Multigrid Methods "An Efficient Multigrid Calculation of the Far field map for Helmholtz problems"
Copper Mountain, USA, March 17 - 22, 2013.
EMG 2012:[slides]European Multigrid Conference 2012 "Generalised Shifted Multigrid preconditioners for the inde finite Helmholtz equation"
Schwetzingen, Germany, August 13 - 16, 2012.
ICCAM 2012:[slides]International Congress on Computational and Applied Mathematics "Generalised Shifted Multigrid preconditioners for the inde finite Helmholtz equation"
Gent, Belgium, July 9 - 13, 2012.
Software
The following HPC algorithms were contributed to the open-source PETSc (Portable, Extensible Toolkit for Scientific Computation) C library
KSPPIPELCG : [manual page] : pipelined CG with deep pipelines
KSPPIPELCG : [manual page] (KSPPIPELCG v2.0, available in PETSc v3.12 onward) : pipelined CG with deep pipelines and improved robustness
Short research visits (< 1 week)
Attendance in addition to the events participated in as listed under ‘Conference talks and seminars’.
WAVES 2017: 13th International Conference on the Mathematical and Numerical Aspects of Wave Propagation, May 15-19, 2017.
EXA2CT BRU 2016: EXA2CT EU FP7 - Review Meeting, Brussels, Belgium, October 17 - 18, 2016.
EXA2CT BOR 2016: EXA2CT EU FP7 - Project Meeting, Bordeaux, France, September 22 - 23, 2016.
EXA2CT FRAN 2016: EXA2CT EU FP7 - Project Meeting, Frankfurt, Germany, March 14 - 15, 2016
SPRING 2016: Spring Meeting of the Dutch-Flemish Numerical Analysis community, TU Delft, The Netherlands, April 27, 2016.
EXA2CT BRU 2015: EXA2CT EU FP7 - Review Meeting, Brussels, Belgium, October 15 - 16, 2015.
EXA2CT OST 2015: OST 2015: EXA2CT European Project - F2F Meeting, Ostrava, Czech Republic, September 13 - 15, 2015.
WSC 2014: 39th Woudschoten Conference of the Dutch-Flemish Numerical Analysis community, Zeist, The Netherlands, October 8 - 10, 2014.
INV 2012: Summer School on Computational Methods for Inverse Problems in Imaging, Kuopio, Finland, June 11 - 15, 2012.
SPRING 2012: Spring Meeting of the Dutch-Flemish Numerical Analysis community, Antwerp, Belgium, May 11, 2012.
WSC 2012: 37th Woudschoten Conference of the Dutch-Flemish Numerical Analysis community, Zeist, The Netherlands, October 3 - 5, 2012.
OPTPDE 2011: ESF Workshop on Fast Solvers for Simulation, Inversion, and Control of Wave Propagation Problems, Würzburg, Germany, September 26 - 28, 2011.
Extended research visits (1 month)
LBNL 2019: research visit to Lawrence Berkeley National Laboratory, Computational Research Department, Scalable Solvers Group, hosted by Xiaoye "Sherry" Li (group lead) and Pieter Ghysels (research scientist), 1 Cyclotron Rd, Berkeley, CA 94720, US, March 2019.
LBNL 2017: research visit to Lawrence Berkeley National Laboratory, Computational Research Department, Scalable Solvers Group, hosted by Xiaoye "Sherry" Li (group lead) and Pieter Ghysels (research scientist), 1 Cyclotron Rd, Berkeley, CA 94720, US, August 2017.
General audience science communication
MATH 2019:[slides] Research Day Mathematics 2019 Pitch talk: How do supercomputers communicate? (and why mathematicians should know)
Antwerp, Belgium, May 16, 2019.
PHD 2015:[slides]PhD in Science: Mathematics, PhD defense 2015 "Multigrid methods for wave scattering problems governed by the Helmholtz and Schrodinger equation"
Antwerpen, Belgium, May 22, 2015.
PROC 2014:[slides]Plechtige Proclamatie Faculteit Wetenschappen UAntwerpen 2014 "Wiskundige algoritmes in het dagelijks leven: de golfvergelijking en haar toepassingen" (in dutch)
Antwerpen, Belgium, September 18, 2014.
WSC 2011:[poster]36th Woudschoten Conference of the Dutch-Flemish Numerical Analysis community: Poster presentation "Multigrid rules the Waves: Complex Shifted Multigrid on the indefinite Helmholtz equation"
Zeist, The Netherlands, October 5 - 7, 2011.
Participation in conference organization
ICPP 2019: International Conference on Parallel Processing, Program Comittee Member in the "Algorithms" Program Section
Kyoto, Japan, August 5-8, 2019.
PMAA 2018: 10th International Workshop on Parallel Matrix Algorithms and Applications, Organizer Minisymposium: "Scalable communication-reducing Krylov subspace methods",
Zurich, Switzerland, June 27-29, 2018.
SIAM PP 2018: SIAM Conference on Parallel Processing for Scientific Computing, Organizer Minisymposium MS40 (joint with dr. Erin Carson): "Scalable Communication-Avoiding and -Hiding Krylov Subspace Methods",
Tokyo, Japan, March 4-8, 2018.
International Journal Referee
ACM ICPP 2019: Proceedings of the 48th International Conference on Parallel Processing (ICPP 19)
SIAM Journal on Matrix Analysis and Applications (SIMAX)
Elsevier - Parallel Computing (PARCO)
IEEE Transactions on Medical Imaging (TMI)
Lecture Teaching
University of Antwerp, Master courses in Mathematics - Financial and Applied Mathematics
MSc 2012-2013, Multilevel and Multiscale solution methods, Course leader(s): Prof. W. Vanroose
MSc 2011-2012, Applications of Partial Differential Equations, Course leader(s): Prof. W. Vanroose, Prof. K. In't Hout
Awards and grants
University Research Fund "Small Research Project" (BOF/KP), University of Antwerp
ProjectID 34398: "HPC iterative solution methods for multi-particle physics simulation".
Start date: April 1, 2017. End date: March 31, 2018.
University Research Fund "Doctoral Fund (1 yr)" (BOF/Opvang SB), University of Antwerp
PhD candidate: Jeffrey Cornelis. Supervisor(s): Wim Vanroose, Siegfried Cools
ProjectID 38148: "Asynchronous Krylov methods with deep pipelines".
Start date: January 1, 2018. End date: December 31, 2018.
FWO Postdoctoral Fellowship. Start date: October 1, 2016. End date: September 30, 2019.
Application no. 12H4617N: "Scalable and error resilient iterative solvers for large scale linear algebra problems".
Grant received on June 22, 2016.
Student Paper Award: 16th Copper Mountain Conference on Multigrid Methods (2013), "An Efficient Multigrid Calculation of the Far field map for Helmholtz equations",
see the Copper Mountain Conference home page for a list of past Student Paper winners.
OS: Microsoft Windows OS: expert; Linux OS: beginner
Version controle systems: Git (Github, Bitbucket): advanced; SVN: beginner
Other skills
Board member of Education Committee UAntwerp Dept. Mathematics and Computer Science , 2013-present
Website Content Manager for UAntwerp Applied Mathematics research group, 2011-present
Member of Society for Industrial and Applied Mathematics (SIAM), 2018-present
Member of Dutch-Flemish Scientific Computing Community (WSC), 2011-present
Active member of the UAntwerp Applied Mathematics research group, 2011-present
Co-organizer of annual UAntwerp ’Wiskunde-in-Zicht’ event days, 2011-2016
Student job at Delta Lloyd Bank (Bank Nagelmackers), Shipments and central cash register, July 2006
Higher Degree in Musical Education (guitar), SAMWD Lier, 2006
Animator (A) Degree Youth Monitor, Kena vzw, 2004
Member of the Belgian Scouting Movement, Sint-Jan in d’Olie Vremde, 1994-2002
Cookie consent
The University of Antwerp website uses cookies and similar technologies to ensure the basic functionality of the site and for statistical and optimisation purposes. It also uses cookies to display content such as YouTube videos and for marketing purposes. This last category consists of tracking cookies: these make it possible for your online behaviour to be tracked. You consent to this by clicking on Accept. Also read our Privacy statement
