Home Resume Research Courses Publications Links
Parikshit Dutta
Postdoctoral Fellow
INRIA Rhône Alpes & Laboratoire Jean Kuntzmann

Contact:
STEEP Team
655 avenue de l'Europe
38330 Montbonnot, France
Phone:   +33 476 61 54 14 (Office)
Mobile:   +33 760 59 49 21
Email: firstname dot lastname at inria dot fr


Welcome to my home page! I am a postdoctoral fellow at STEEP Team, which is a part of INRIA Rhône Alpes and Laboratoire Jean Kuntzmann (LJK). I am currently working in the field of Uncertainty Quantification, with focus on model calibration and validation under uncertainty. I am predominantly interested in using Bayesian methods to model uncertainty. Currently I am working on an integrated land use and transportation model for Grenoble, France.

Prior to joining INRIA, I did my Ph.D. from Texas A&M University in Aerospace Engineering . My advisor was Dr. Raktim Bhattacharya.

Research interests

  • Model validation and calibration under uncertainty using Bayesian methods. (Current research)
  • Filtering and state estimation of high dimensional dynamical systems. (Ph.D. research)
  • Randomized algorithms and MCMC. (Ph.D. and current research)
  • Stochastic dynamical systems. (Ph.D. and current research)
    •

Other interests and attractions

  • Stochastic optimization - current interest
    • Specially interested in optimization with large number of random parameters. Currently working on an integrated land use and transportation model, involving more than 100 parameters.
  • Fokker Planck and Perron Frobenius operator and their applications.
    • How to approximate the evolving densities in a SDE.
  • Polynomial chaos and Karhunen Loève expansion
    • Use of stochastic FEM methods to approximate the noise as well as the states in a SDE.
  • Planetary entry descent landing (EDL).
    • EDL of a hypersonic vehicle into a planet's atmosphere. Estimating states and parameters of the vehicle in presence of uncertainties.
  • Stochastic optimal control.
    • Worked with problems involving optimizing functions of expectation. Have some basic knowledge about stochastic HJB.
  • Computational sustainability.
    • This relates to use of computer science and applied mathematics in real estate usage and transportation. Find out more here.

Some programming stuff

  • Python - preferred progamming language.
    • Is number 6 in popularity. Check this out . By popular acclaim it is most popular in academia due to ease of object orientation, simplicity and wide range of classes available.
  • MCMC using python. I primarily use pyMC but often write my own code.
  • High performance computing.
    • Right now dealing with symmetric multiprocessing, MPI, OAR, ssh, grid computing etc. Check out our own grid; as in grid 5000