About me

sdr

I am a Senior Research Engineer (Förste Forskningsingenjör) at the Department of Computing Science of Umeå University, Sweden. I also work as an Application Expert at High Performance Computing Center North (HPC2N).

My current research is related to eigenvalue problems, numerical linear algebra, high performance computing and task-based parallelism. My PhD dissertation focused on GPU computation, fast direct linear system solvers and image denoising.

Emails: mirkom@cs.umu.se, mirko.myllykoski@umu.se, first_name.last_name@gmail.com

Other: ORCID [0000-0002-3689-0899], Umeå university (DiVA), Github, LinkedIn, ResearchGate, ArXiv, jyx

StarNEig library

StarNEig library aims to provide a full suite of task-based algorithms for solving nonsymmetric dense (generalized) eigenvalue problems. The library is built on top of the StarPU runtime system and targets both shared memory and distributed memory machines. Some components of the library support GPUs.

The library is open source and published under BSD 3-Clause licence. The library and the accompanying documentation are available at: https://nlafet.github.io/StarNEig.

Research papers

In press and submitted

Peer reviewed publications

  • Mirko Myllykoski, Carl Christian Kjelgaard Mikkelsen: Task-based, GPU-accelerated and Robust Library for Solving Dense Nonsymmetric Eigenvalue Problems, Concurrency and Computation: Practice and Experience, 2020, doi: 10.1002/cpe.5915 (full paper, postprint, DiVA, arXiv:2002.05024, source code)
  • Mirko Myllykoski, Carl Christian Kjelgaard Mikkelsen: Introduction to StarNEig — A Task-based Library for Solving Nonsymmetric Eigenvalue Problems, In Parallel Processing and Applied Mathematics, 13th International Conference, PPAM 2019, Bialystok, Poland, September 8–11, 2019, Revised Selected Papers, Part I, Lecture Notes in Computer Science, Vol. 12043, Wyrzykowski R., Deelman E., Dongarra J., Karczewski K. (eds), Springer International Publishing, pp. 70-81, 2020, doi: 10.1007/978-3-030-43229-4_7 (postprint, DiVA, arXiv:1905.04975, source code)
  • Carl Christian Kjelgaard Mikkelsen, Mirko Myllykoski: Parallel Robust Computation of Generalized Eigenvectors of Matrix Pencils, In Parallel Processing and Applied Mathematics, 13th International Conference, PPAM 2019, Bialystok, Poland, September 8–11, 2019, Revised Selected Papers, Part I, Lecture Notes in Computer Science, Vol. 12043, Wyrzykowski R., Deelman E., Dongarra J., Karczewski K. (eds), Springer International Publishing, pp. 58-69, 2020, doi: 10.1007/978-3-030-43229-4_6 (postprint, DiVA, arXiv:2003.04776, source code)
  • Mirko Myllykoski: A Task-Based Algorithm for Reordering the Eigenvalues of a Matrix in Real Schur Form, In Parallel Processing and Applied Mathematics, 12th International Conference, PPAM 2017, Lublin, Poland, September 10-13, 2017, Revised Selected Papers, Part I, Lecture Notes in Computer Science, Vol. 10777, Wyrzykowski R., Dongarra J., Deelman E., Karczewski K. (eds), Springer International Publishing, pp. 207-216, 2018, doi: 10.1007/978-3-319-78024-5_19 (postprint, DiVA, source code)
  • Mirko Myllykoski, Tuomo Rossi, Jari Toivanen: On solving separable block tridiagonal linear systems using a GPU implementation of radix-4 PSCR, Journal of Parallel and Distributed Computing, 115, pp. 56-66, 2018, doi: 10.1016/j.jpdc.2018.01.004 (postprint, jyx, DiVA, source code)
  • Mirko Myllykoski, Roland Glowinski, Tommi Kärkkäinen, Tuomo Rossi: A GPU-Accelerated Augmented Lagrangian Based L¹-mean Curvature Image Denoising Algorithm Implementation, In WSCG 2015 : 23rd International Conference in Central Europe on Computer Graphics, Visualization and Computer Vision’2015 : Full Papers Proceedings, Gavrilova M., Skala V. (eds.), Union Agency, pp. 119-128, ISBN 978-80-86943-65-7, 2015 (link, full paper, jyx, source code)
  • Mirko Myllykoski, Roland Glowinski, Tommi Kärkkäinen, Tuomo Rossi: A New Augmented Lagrangian Approach for L¹-mean Curvature Image Denoising, SIAM Journal on Imaging Sciences, 8 (1), pp. 95-125, 2015, doi: 10.1137/140962164 (full paper, jyx, source code)
  • Mirko Myllykoski, Tuomo Rossi: A parallel radix-4 block cyclic reduction algorithm, Numerical Linear Algebra with Applications, 21 (4), pp. 540-556, 2014, doi: 10.1002/nla.1909 (postprint, jyx)
  • Mirko Myllykoski, Tuomo Rossi, Jari Toivanen: Fast Poisson Solvers for Graphics Processing Units, In Applied Parallel and Scientific Computing, 11th International Conference, PARA 2012, Helsinki, Finland, Lecture Notes in Computer Science, Vol. 7782, Manninen P., Öster P. (eds), Springer Berlin Heidelberg: Berlin, Germany, pp. 265-279, 2013, doi: 10.1007/978-3-642-36803-5_19 (postprint, jyx, source code)

Reports and deliverables

  • Mirko Myllykoski, Carl Christian Kjelgaard Mikkelsen, Angelika Schwarz, Bo Kågström: D2.7 Eigenvalue solvers for nonsymmetric problems, public NLAFET deliverable, 2019 (download, link, DiVA)
  • Lars Karlsson, Mahmoud Eljammaly, Mirko Myllykoski: D6.5 Evaluation of auto-tuning techniques, public NLAFET deliverable, 2019 (download, link, DiVA)
  • Bo Kågström et al.: D7.8 Release of the NLAFET library, public NLAFET deliverable, 2019 (download, link, DiVA)
  • Mirko Myllykoski, Lars Karlsson, Bo Kågström, Mahmoud Eljammaly, Srikara Pranesh, Mawussi Zounon: D2.6 Prototype Software for Eigenvalue Problem Solvers, public NLAFET deliverable, 2018 (download, link, DiVA)
  • Mirko Myllykoski, Carl Christian Kjelgaard Mikkelsen, Lars Karlsson, Bo Kågström: Task-Based Parallel Algorithms for Reordering of Matrices in Real Schur Forms, NLAFET Working Note WN-11, 2017. Also as Report UMINF 17.11, Department of Computing Science, Umeå University, SE-901 87 Umeå, Sweden (download, UMINF, DiVA)
  • Carl Christian Kjelgaard Mikkelsen, Mirko Myllykoski, Björn Adlerborn, Lars Karlsson, Bo Kågström: D2.5 Eigenvalue Problem Solvers, public NLAFET deliverable, 2017 (download, link, DiVA)

Conferences, workshops and meetings

  • Cancelled: Task-based Algorithms and Software for Solving Dense Nonsymmetric Eigenvalue Problems, Householder Symposium XXI on Numerical Linear Algebra, Selva di Fasano (Br), Italy, June 14-19, 2020 (abstract)
  • How two-sided matrix transformation algorithms can benefit from task parallelism, Nordic Numerical Linear Algebra meeting, Stockholm, Sweden, October 21-22, 2019 (abstract, slides)
  • Task-based, GPU-accelerated and Robust Algorithms for Solving Dense Nonsymmetric Eigenvalue Problems, Swedish eScience Academy, Lund, Sweden, October 15-16, 2019 (poster)
  • Introduction to StarNEig — A Task-based Library for Solving Nonsymmetric Eigenvalue Problems, 13th International Conference on Parallel Processing and Applied Mathematics (PPAM 2019), Bialystok, Poland, September 10, 2019 (slides)
  • Co-authored: Parallel Robust Computation of Generalized Eigenvectors of Matrix Pencils (presented by Carl Christian Kjelgaard Mikkelsen), 13th International Conference on Parallel Processing and Applied Mathematics (PPAM 2019), Bialystok, Poland, September 10, 2019 (slides)
  • Poster: Task-Based Algorithms for Eigenvalue Problems, Swedish eScience Academy, Uppsala, Sweden, October 16-17, 2018 (poster)
  • Poster: A Task-Based Algorithm for Reordering the Eigenvalues of a Matrix in Real Schur Form, Swedish eScience Academy, Umeå, Sweden, October 11-12, 2017 (poster, presentation)
  • A Task-Based Algorithm for Reordering the Eigenvalues of a Matrix in Real Schur Form, 12th International Conference on Parallel Processing and Applied Mathematics (PPAM 2017), Dublin, Poland, September 13, 2017 (slides)
  • Poster: How Fast Direct Solvers Can Benefit from GPU-acceleration, Householder Symposium XX on Numerical Linear Algebra, Blacksburg, Virginia, USA, June 20, 2017 (poster)
  • Co-authored: Extreme-Scale Eigenvalue Reordering in the Real Schur Form (presented by Carl Christian Kjelgaard Mikkelsen), SIAM Conference on Computational Science and Engineering (CSE17), Atlanta, Georgia, USA, February 28, 2017 (slides)
  • A GPU-Accelerated Augmented Lagrangian Based L¹-mean Curvature Image Denoising Algorithm Implementation, 23rd International Conference in Central Europe on Computer Graphics, Visualization and Computer Vision 2015, Plzen, Czech Republic, June 9, 2015 (slides)
  • Poster: A New Augmented Lagrangian Approach for L¹-mean Curvature Image Denoising, RICAM Special Semester on New Trends in Calculus of Variations, Workshop 2: Variational methods in imaging, Linz, Austria, October 27-31, 2014 (poster)
  • Cyclic Reduction Type Poisson and Helmholtz Solvers on a GPU, SIAM Conference on Parallel Processing for Scientific Computing, Portland, Oregon, USA, February 18, 2014 (slides)
  • Block Cyclic Reduction Type Fast Poisson Solvers for GPU, Workshop on the State-of-the-Art in Scientific and Parallel Computing (PARA2012), Helsinki, Finland, June 13, 2012 (slides)

Theses

  • PhD thesis: On GPU-Accelerated Fast Direct Solvers and Their Applications in Image Denoising, Jyväskylä, Finland: University of Jyväskylä. Jyväskylä studies in computing, 218, ISBN 978-951-39-6277-7, ISSN 1456-5390, 2015 (download, jyx, Google Docs)
  • Master’s thesis (Pro gradu, in Finnish): Separoituvien yhtälöryhmien ratkaiseminen ohjelmoitavalla näytönohjaimella (Solving separable linear systems on GPU), 2011 (download, Google Docs)
  • Bachelor’s thesis (Kandidaatintutkielma, in Finnish): Rinnakkaistuvat nopeat Poisson-ratkaisijat (Parallel Fast Poisson Solvers), 2010 (download, Google Docs)

Miscellaneous

  • A quick introduction to StarPU (Google Slides)
  • Cyclic Reduction Type Poisson and Helmholtz Solvers on a GPU (extended version), Farhat Research Group Seminars, Stanford, California, USA, February 26, 2014 (slides)
  • Alberto Quintero, Alexandra Rohde O’Sullivan Freltoft, Anna Igolkina, Mirko Myllykoski, Nitya Dixit, Fabian Rost (instructor): Morphogenesis and Dynamics of Multi-cellular Systems, ECMI Newsletter 52, 2012 (link)
-----BEGIN PGP PUBLIC KEY BLOCK-----
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=rEO7
-----END PGP PUBLIC KEY BLOCK-----