Hi, I'm Guru.

I'm a fourth-year Physics PhD student at UC San Diego, advised by Nigel Goldenfeld. I work on the statistical mechanics of non-equilibrium systems, in problems spanning turbulence, ecology, and machine learning.

I'm mostly interested in how non-equilibrium systems transition into rich emergent states. A few questions drive my work:

1. How do non-equilibrium systems move between states, and what governs the transition? I study this in transitional flows: the onset of turbulence in pipe flow, Taylor–Couette flow, and related systems. I'm interested in how new states emerge as a reorganization of length, energy, and time scales, and how the interplay between these scales controls the spatiotemporal organization and dynamics as the system passes from one state to another. In pipes with body forces, for example, I showed that the transition is governed by tricritical directed percolation (Physical Review Letters, 2025).
2. Can such core principles be generalized and utilized to seek simpler modeling priors in complex yet realistic settings? The goal is to find simple descriptions of phenomena endowed with real-world complexity. The problem I work on here is stratified turbulence — fluid flow under buoyancy, as in the ocean and atmosphere, where buoyancy and turbulence together mix and reorganize the fluid, shaping both the route to turbulence and the structure of fully developed turbulence. I use techniques from statistical mechanics to understand aspects such as the onset of turbulence in these flows, isolating the key physical mechanisms in a more tractable form and using them to inform qualitative features and make quantitative predictions.
3. What lies beyond linear instability? Often invoked to analyze transitions to non-equilibrium states, linear instability tells us when a state is unstable, but crucially fails to answer what the long-time steady state is. Moreover, it fails to address the complex, multi-scale structure that develops (mostly non-analytically) on the way to new non-equilibrium states, and how information is processed en route to such states. I investigate this in machine learning. May's famous complexity–stability theorem in ecology predicts that large, densely interacting systems become linearly unstable as they grow, as shown using random-matrix theory. However, a neural network post training is exactly such a system and functions stably. I ask how training lets it escape this instability.

To address these problems, I combine methods from non-equilibrium statistical field theory (Martin–Siggia–Rose and Onsager–Machlup functionals), the theory of phase transitions, finite-size scaling, asymptotic analysis, stochastic modeling, and random-matrix theory.

Before UCSD, I did a Dual Degree in Engineering Physics at IIT Bombay, working on quantum condensed matter problems: with Hridis Kumar Pal, I studied topological insulator–superconductor junctions. I also spent time at TIFR Mumbai, understanding electron–phonon equilibration via non-equilibrium field theory methods, and at Aalto University, studying fluctuations in non-centrosymmetric superconductors.

Affiliations

Current

2022 – present

UC San Diego Physics

Ph.D. candidate in theoretical physics. Thesis work: transition to turbulence under body forces, stratified flows, and the statistical mechanics of machine learning.

Education

2017 – 2022

IIT Bombay

Dual Degree (B.Tech + M.Tech) in Engineering Physics, specialization in Nanoscience. Master's thesis on topological insulator–superconductor junctions.