Pawan Kumar

Employment

Assistant Professor, 2017-now, IIIT, Hyderabad, India
Postdoc, 2014-2016, FU Berlin, Germany
Postdoc, 2013- 2014, Fraunhofer ITWM, Kaiserslautern, Germany Postdoc, Sept. 2011-July 2013, KU Leuven and exascience lab, Leuven, Belgium

Education

PhD (Highest honours (``très honorable'' in French)), 2010, INRIA, France
MS, 2007, Indian Institute of Technology, Guawahati, India

Languages

English (Fluent), French (Intermediate), German (Beginner)

Computer
Skills

Familiarity with: HPC programming: C, C++, Fortran 77/90, MPI, GPI (Global Address Space Programming, PGAS model) OpenMP, Cilk plus, C++-11 threads, Parallel analysis tools: Scalsca, TAU analysis tool, Vtune, Likwid, HDF5, Paraview, Web tools: XML, HTML, PHP, CSS, Version Control: SVN, GIT, Other Protyping tools: MATLAB, Octave, Maple, mexfile (Matlab, C, Fortran interface), Documentation: Latex

Numerical libraries used: FreeFEM++, UMFPACK, SuperLU, BLAS, Sparse BLAS, Goto BLAS, MKL (sparse and dense), PLASMA, METIS, ITSOL, pARMS, Peano (Grid traversal software) PETSc (basic), HYPRE, etc

Softwares created:

1. COMPRE: Fortran 77/90 codes for preconditioning structured PDE problems by combinative preconditioning (see the paper below)
2. ADDS: A class of parallel non-overlapping domain decomposition methods written in cilk plus suitable for shared memory architecture. A distributed memory variant is under construction

Operating Systems: Windows, Unix (preferred!)

Publications
Reports

Solvers for Computer Vision Problems:

• S. Das, S. Katyan, P. Kumar, A Deflation Based Fast and Robust Preconditioner for Bundle Adjustment, accepted in WACV 2021.
• S. Katyan, S. Das, P. Kumar, Multigrid Preconditioned Solver for Bundle Adjustment, accepted in WACV 2020.
• S. Das, S. Katyan, P. Kumar, Domain Decomposition Based Preconditioned Solver for Bundle Adjustment, accepted in NCVPRIPG 2019.

On Preconditioners/Linear Solvers:

• P. Kumar, Fast Preconditioned Solver for Truncated Saddle Point Problem in Nonsmooth Cahn–Hilliard Model, Book chapter, Recent Advances in Computational Optimization, 2016
• P. Kumar, L. Grigori, F. Nataf, and Q. Niu, Combination preconditioning based on relaxed nested factorization and tangential filtering decomposition, International Journal of Computer Mathematics, 2015, doi:10.1080/00207160.2014.998208

• P. Kumar, Aggregation based on graph matching and inexact coarse grid solve for algebraic multigrid, accepted, Int. J. Comp. Math., 2013,
  http://dx.doi.org/10.1080/00207160.2013.821115

On Fourier Analysis

• Q. Niu, L. Grigori, P. Kumar, and F. Nataf, Modified tangential frequency filtering decomposition and its Fourier analysis, Numerische Mathematik, Volume 116, issue 1, p 123-148, 2010, doi: 10.1007/s00211-010-0298-3

On High Performance Computing/Scientific Computing

• S. Rampalli, N. Sehgal, I. Bindlish, T. Tyagi, Efficient FPGA Implementation of Conjugate Gradient Methods for Laplacian System using HLS, short paper, FPGA 2019
• P. Kumar, Multilevel Communication Optimal Least Squares Solver, IEEE proceedings, International Conference on Computational Sciences, ICCS, vol. 51, p. 1838-1847, 2015, doi: 10.1016/j.procs.2015.05.410
• P. Kumar, Communication Optimal Least Squares Solver, accepted, IEEE proceedings, 16th international conference on high performance computing and communications, HPCC, 20-22 August 2014, Paris, France
• P. Kumar, Multi-threaded direction preserving preconditioners, IEEE proceedings, 13th international symposium on parallel and distributed computing, ISPDC, 23-27 June 2014, p. 148-153, Marseille (Porquerolles island), France
• P. Kumar, S. Markidis, G. Lapenta, K. Meerbergen, D. Roose, High Performance Solvers for Implicit Particle in Cell Simulation, ICCS, vol. 18, Procedia Computer Science, pp 2396-2405, 2013, http://dx.doi.org/10.1016/j.procs.2013.05.396
• P. Kumar, K. Meerbergen, and D. Roose, Multi-threaded nested filtering factorization preconditioner, vol. 7782, LNCS, pp. 220-234, 2013

Reports under submission

• P. Kumar, L. Grigori, Q. Niu, F. Nataf, Fourier analysis of Modified Nested Factorization Preconditioner for Three-Dimensional Isotropic Problems, HAL, INRIA report, 2019.
• L. Grigori, P. Kumar, F. Nataf, and K. Wang, A class of multilevel parallel preconditioners, submitted as INRIA tech. report no. 7410, available online at: http://hal.archives-ouvertes.fr/docs/00/52/41/10/PDF/Paper.pdf

Grants

• Ripple Center of Excellence: Distributed Optimization and Blockchain
• INAE Travel grant
• MATRICS grant: Solvers for saddle point problems
• ERCIM: Marie Curie Actions Fellowship

Awards

• Marie-Curie ERCIM fellowship for independent postdoctoral research, 2013-2014
• European CORDIS scholarship (applied by Supervisor) for PhD studies at INRIA, Saclay, 2007-2010
• Secured a percentile of 96.6 (rank 88/2500) in all India applied Mathematics category in GATE (Graduate Aptitude Test in Engineering, India), 2007
• Junior summer research scholarship from JNCASR bangalore, India to conduct undergraduate research for two months, 2005
• Certificate of excellence for 3rd rank in Mathematics department in north eastern hills university in 2005

Teaching

• Spring 2021
  • Adavnced Optimization
• Monsoon 2020
  • Discrete Structures
  • Probability and Statistics
• Spring 2020
  • Introduction to Parallel Scientific Computing
• Monsoon 2019
  • Topics in Applied Optimization
  • Discrete Structures
• Spring 2019
  • Introduction to Parallel Scientific Computing
  • Topics in Optimization on Manifolds
• Monsoon 2018
  • Discrete Structures
  • Topics in Applied Optimization
• Spring 2018
  • Introduction to Parallel Scientific Computing
  • Linear Algebra 2018
• Monsoon 2017
  • Discrete Mathematics and Algorithms
  • Algorithms
• Spring 2017
  • Introduction to Parallel Scientific Computing

Proposal
accepted

Scalable robust Schur complement preconditioners using PGAS (performed under ERCIM fellowship), accepted and completed under MArie-Curie Fellowship at Fraunhofer ITWM.

Talks

• Multilevel Communication Optimal Least Squares Solver, HPCC, 1-3 June 2015, Reykjavic, Iceland
• High Performance Solvers for Implicit Particle in Cell Simulations, ICCS, 5-7 June 2013, Barcelona, Spain
• Multi-threading and auto-vectorization for direction preserving preconditioners, SIAM conference on Computational Science and Engineering, 25 February - 1 March 2013, Boston, USA
• Parallel aggregation based algebraic multigrid, International Conference on Domain Decomposition, 25-29 June 2012, Rennes, France
• Purely algebraic domain decomposition methods for the incompressible Navier-Stokes equation, SIAM LA, 18-22 June 2012, Valencia, Spain
• Multithreaded row and column sum based preconditioners, PARA 10-13 June 2012, Helsinki, Finland
• Purely algebraic domain decomposition methods for the incompressible Navier-Stokes equation, Workshop on Recent Developments in the Solution of Indefinite Systems, April 17, 2012, Eindhoven, The Netherland
• Combination preconditioning based on relaxed nested factorization and tangential filtering preconditioner, IHP, 2008, Paris, France
• Gershgorin circles and Poincare separates, ESS sem. series, KU Leuven, Belgium

Visits
Schools
Workshops

• Research visit (Prof. M. Gander) University of Geneva, Geneva, 2-6 June 2014
• Research visit (Prof. A. Napov) Universite Libre de Bruxelles, 7-11 April 2014
• Winter school on Hierarchical matrices, Leipzig, Germany, 2014
• Visit Lawrence Berkeley National Lab.:, 4th March 2013, Berkeley, California
• Ninth VI-HPS Tuning Workshop, 23-27 April 2012, St-Quentin-en-Yvelines, France
• INRIA school on solution of large sparse linear systems, Sophia-antipolis, France, 2008
• Winter school on Hierarchical matrices, Leipzig, Germany, 2008
• Workshop on FreeFEM++, 14-15 September, 2009, IHP, Paris, France

Hobbies

• Playing chess, biking, travelling and cooking

About this document ...

This document was generated using the LaTeX2HTML translator Version 2019.2 (Released June 5, 2019)

The command line arguments were:
latex2html -no_math -html_version 3.2,math -split 0 cv.tex

The translation was initiated on 2021-02-23