Pratik Nayak
Computer Scientist, Mathematician, Engineer.
Currently
Doctoral Candidate in Applied Mathematics and Computer Science.
Research interests
High Performance Computing, Numerical Linear Algebra, GPU Computing, Asynchronous iterative methods.
Education
Oct 2017 -- present
Karlsruhe Institute of Technology, Karlsruhe, Deutschland
- Doctoral Candidate
August 2015 -- June 2017
Technical University of Delft, Delft, Netherlands
- Masters (with Honors) in Mechanical Engineering, Specializing in Solid and Fluid Mechanics.
Awards
September 2015
Wind Energy Scholarship, Delft Wind Energy Institute, TU Delft. - Full scholarship for Masters studies
Teaching and Organizational Experience
November 2022
Lead student volunteer, SC22, Dallas Texas, USA.
November 2021
SCALE student (Lead volunteer): part of the Junior Technical Papers committee, SC21, Virtual and St. Louis, Missouri, USA.
April 2021 -- July 2021
Numerical Linear Algebra for High Performance Computing,
Instructor and Assistant, Department of Mathematics, Karlsruhe Institute of Technology
October 2020 -- February 2021
Numerical Linear Algebra for High Performance Computing,
Teaching Assistant, Department of Mathematics, Karlsruhe Institute of Technology
November 2020
Student volunteer, SC20, Virtual, World
October 2018 -- February 2019
Numerical Linear Algebra for High Performance Computing,
Teaching Assistant, Department of Mathematics, Karlsruhe Institute of Technology
July 2018
Student volunteer, PASC18, Basel, Switzerland
January 2018
HPC for Tomorrow- Scientific Computing short course,
Teaching Assistant, National Taiwan University, Taiwan
October 2017 -- February 2018
Numerical Linear Algebra for High Performance Computing,
Teaching Assistant, Department of Mathematics, Karlsruhe Institute of Technology
Publications
2022
- Batched sparse iterative solvers on GPU for the collision operator for fusion plasma simulations, International Parallel and Distributed Processing Symposium (IPDPS), Lyon, May-June 2022 (accepted)
2021
-
Ginkgo: A Modern Linear Operator Algebra Framework for High Per- formance Computing, December 2021, ACM Transactions on Mathematical Software (ACM-TOMS) doi:10.1145/3480935
-
Advances in Mixed Precision Algorithms: 2021 Edition, August, 2021, Sandia National laboratory technical report. SAND2021-10227R
-
Batched Sparse Iterative Solvers for Computational Chemistry Simulations on GPUs, 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems (ScalA), November 2021, 10.1109/ScalA54577.2021.00010
-
A collaborative peer review process for grading coding assignments in coursework, July 2021, ICCS 2021, [paper]
2020
-
Two stage asynchronous iterative solvers for multi-GPU clusters , Nov 2020, ScalA Workshop, Supercomputing (SC20), [paper]
-
Ginkgo: A high performance numerical linear algebra library, August 2020, The Journal of Open Source Software, doi: 10.21105/joss.02260
-
Evaluating asynchronous Schwarz solvers for Exascale, Aug 2020, The International Journal of High Performance Computing [paper link]. Alternative: arXiv:2003.05361, cs.DC,
-
A Survey of Numerical Methods Utilizing Mixed Precision Arithmetic, July 2020, arXiv:2007.06674 cs.MS
-
Ginkgo: A Modern Linear Operator Algebra Framework for High Performance Computing, July 2020, arXiv:2006.16852, cs.MS
-
Load-balancing Sparse Matrix Vector Product Kernels on GPUs, ACM-TOPS, March 2020, doi: 10.1145/3387354
2019
- Towards Continuous Benchmarking: An Automated Performance Evaluation Framework for High Performance Software, Platform for Advanced Scientific Computing (PASC) Conference, June 12-14, 2019, Zurich , doi: 10.1145/3324989.3325719
Talks
-
A collaborative peer review process for grading coding assignments in coursework, July 2021, ICCS 2021, [slides]
-
Two stage asynchronous iterative solvers for multi-GPU clusters , Nov 2020, ScalA Workshop, Supercomputing (SC20) , [slides]
-
Evaluating asynchronous Schwarz solvers for Exascale, PACO19 conference, Max Planck Institute-Magdeburg, 5-6 November 2019. [slides]
-
Using iterative methods for local solves in asynchronous Schwarz methods, University of Tennessee, Knoxville - Innovative Computing Laboratory, Lunch talks [slides].