Gopinath Mishra recently joined IMSc as a faculty member in the Theoretical Computer Science group. He completed his PhD and Masters degrees from the Indian Statistical Institute (ISI), Kolkata, and has a Bachelor’s degree from Veer Surendra Sai Institute of Technology, Odisha. He pursued postdoctoral research at the University of Warwick, UK, and the National University of Singapore. His broad research interests are in theoretical computer science with a focus on model-centric computation.
How did your interest in research develop?
I really liked solving puzzles in my childhood, and had a particular liking for mathematics. When I was in college during my Bachelor’s, I liked theoretical computer science, particularly algorithms. When I joined ISI for my Masters, I became more curious about the area of theoretical computer science and algorithms, which eventually led me to research.
What do you work on?
I am interested in algorithms when the data is very large and we have very limited access to it. The data can be so huge that we may not be able to read the dataset in its entirety even once. Or even if we are able to read the data once, we aren’t able to store it and access it at our disposal. It may be the case that the data is distributed across many systems instead of being available with a central computer.
My research is primarily theoretical. We study fundamental problems in the simplest possible theoretical models. When considering practical applications, many additional specifics must be taken into account, and suitably modified versions of the algorithms that retain theoretical guarantees may be applied.
In my PhD research, I worked on streaming and property testing with limited access to data. In the streaming model, we can read the entire data once but cannot store it. In property testing, we cannot read the entire data even once. During my postdoc, I worked on massively parallel models and distributed computing, where the data is stored across multiple computers. The massively parallel computation model is a standard framework for studying parallel computing today. It shares several algorithmic techniques with streaming and property testing. This overlap is one reason I became interested in this model during my first postdoctoral position, after completing my PhD in streaming and property testing. Towards the end of my first postdoc, I developed an interest in distributed computing, which is closely related to massively parallel computation models. This interest eventually led to my second postdoc, which focused solely on distributed computing. There is a common theme across my PhD and postdoc research involving randomized algorithms and graph problems under limited access to data, with a gradual shift across different computational models.
For the next few years of research, I want to focus on streaming, property testing, and distributed computing. My current work is focused on graph algorithms, including problems related to graph coloring, clustering, shortest paths, and graph parameter estimation.
What are some of your other interests?
I enjoy reading, particularly non-fiction books related to science, politics, and people’s struggles, which help keep me grounded. I also enjoy traveling, going on long walks, listening to music, and spending time with family and friends.
