Ranjit Kumar Nanda Department of Physics Area of interest: Electronic Structure of Solids

B. R. K. Nanda (Ranjit) is currently an Associate Professor in the department of Physics at IIT Madras. He was an assistant professor in the same institute from July, 2011 to July, 2016. Prior to this he worked as a postdoctoral fellow at University of Missouri, Columbia for five years. He obtained his M. Sc. (Physics) degree from Ravenshaw College, Cuttack and holds a PhD degree from IIT Bombay. His research interest is centered around examining the electronic structure of solids.

How does your group keep HPCE cluster busy?

We solve many-electron Hamiltonian in a mean-field framework. Computationally this leads to diagonalization of large-scale matrices (of the order of 10⁴) at each point of a grid built in the momentum space. The convergence of the net results is carried out through a self-consistent loop. Sorting, integrations, and solving second-order differential equations in space and time domains are the other computational tasks that we execute on daily basis.

How do you see HPCE landscape in the domain of your research area change over the years?

In the last decades of 20th century, the area of computational condensed matter physics was treated as an interface between theoretical and experimental condensed matter physics and materials science. Now, it has evolved as an alternate and independent research platform which is capable of both developing new theoretical concepts and replacing the experimental labs with virtual ones. The success lies behind the availability of multi-node-multi-core systems with huge memory (to feed the SCF loop and allied subroutines). Therefore, HPCE plays a significant role in this research area.

What would you suggest to new faculty members and new students in your research area?

As the need will be to carry out large number of computational tasks in less time, it is crucial to learn ever evolving optimization algorithms and to regularly update the compilers. Furthermore, the machine learning tools and high-throughput studies are becoming important in our area as predictive tools. This will not only reduce the computational time, but will also help in mimicking the experiments.