Joyjit Kundu Email & Phone Number

Leuven, Flanders, Be

Leuven, Flemish Region, Belgium

- Performance modelling and optimization of Machine learning (e.g., LLM inference, training) + HPC workloads- AI accelerated physics simulations- Molecular dynamics + ML for Drug Discovery

Hyderabad, Telangana, India

Durham, North Carolina

Supported by Simons Collaboration on Cracking the Glass Problem.Advisor: Dr. Patrick Charbonneau, Dr. Ludovic BerthierAlthough amorphous materials such as grains, bubbles, foams, glasses are ubiquitous in nature, their complete theoretical description remains one of the unsolved fundamental problems in theoretical condensed-matter physics. On the molecular scale, the structure of a glass resembles that of a liquid, however, they do not flow, thus, behaves like a solid. Developing a theoretical description of such systems is notoriously difficult as the conventional techniques fail. The problem is solvable in the infinite dimensional limit, but the phenomenology is quite different in physical dimensions. Many systems in fields as diverse as computer science, neuroscience, biology and information theory exhibit a phenomenology akin to glassiness and will benefit from our advances. • using numerical simulations to bridge the gap between the infinite dimensional limit and the physical dimensions. • developed statistical mechanical models of glass forming liquid.• developed Monte Carlo codes to simulate systems of polydisperse hard sphere fluids in different spatial dimensions.• implemented the swap algorithm that has recently been very successful to simulate glassy systems. • optimized various parameters (e.g. polydispersity, swap probability) to obtain the best performance in various spatial dimensions. • developed scripts for analyzing the trajectories and measuring different observables to understand the dynamic and statistical properties (e.g pair correlation, overlap function, fluctuations). • research articles in preparation.

Berkeley, Ca

Supported by UC Berkeley’s Center for Gas Separations Relevant to Clean Energy Technologies, an Energy Frontier Research Center and The Soft Matter Electron Microscopy ProgramAdvisor: Dr. Stephen Whitelam, Dr. David PrendergastDeveloping technologies for gas separation and storage are crucial for clean, renewable energy and various industrial applications. Metal organic frameworks (MOFs), a class of porous crystalline materials with largely tunable properties and huge surface area, are promising candidates in this regard. • developed statistical mechanical models of gas uptake within porous media.• collaborated with experimentalists to build coarse-grained simulation models. • developed kinetic Monte Carlo codes to simulate the model systems.• performed multi-scale simulations to parametrize the model and measured important thermodynamic and dynamic observables related to gas uptake.• used tools of statistical physics to understand the physical properties.• proposed strategies for selective gas capture under nonequilibrium conditions. • explained the physics of cooperative adsorption in MOFs.• came up with exact solution of a model of cooperative adsorption. • published research articles (see Publications)The assembly of sequence-defined polymers into ordered nanostructures opens up a class of well- defined protein-mimetic materials with complex functions such as catalysis and molecular recognition. Specific arrangement of different functional groups in a peptoide leads them to self-assemble in different structures like nano sheets, nanotubes. • constructed atomistic models of self-assembled peptoid nanostructures (e.g. nanosheet).• performed all-atom molecular dynamics simulations to understand self-assembly mechanism.• computed free enrergy of the cis, and the trans conformations using 2-Phase Thermodynamics Method.• discovered the molecular structure and arrangement within the nanostructures.• published research articles (see Publications)

Chennai Area, India

Advisor: Dr. R. RajeshEntropy-driven phase transitions are interesting and counter-intuitive as they lead to increase in visible order with gain in entropy. Systems exhibiting entropy-driven transitions include colloids of polymethyl methacrylate (PMMA), tobacco mosaic virus, fd virus, silica particles, boehmite particles, and Brownian platelets. Hard-core lattice gas models or systems with only excluded volume interactions on lattices are minimal models to study entropy-driven ordering transitions. • studied the phase transitions in systems of hard rectangles/rods on lattices.• developed a new, efficient and highly parallelizable Monte Carlo algorithm with nonlocal moves.• developed Monte Carlo codes to perform simulations for the model of hard rectangles of arbitrary size on lattices.• developed GPU version of the code with CUDA C.• adapted MPI and openMP versions of the code. • discovered different phases, phase diagrams, detailed nature of the phase transitions.• performed free energy calculations using thermodynamic integration and entropy estimations to obtain phase diagrams.• developed codes and scripts to analyze the data.• studied different limits of the problem using various analytical techniques such as mean filed theory, viral expansion, high density perturbation expansion.• solved the model of rods exactly on a Bethe- like lattice.• published research articles (see Publications).

Chennai Area, India

Advisor: Prof. Prasanta Kumar Tripathy• worked on developments in Cosmology.• studied Hořava–Lifshitz gravity.

Indore Area, India

Advisor: Dr. Manoranjan P. SinghCoherent Population Trapping (CPT) is a phenomenon occurring due to the coherent effect of atom-field interaction. If the atoms are prepared in coherent superposition states, it is possible to cancel the absorption or emission under certain conditions. These atoms are then effectively transparent to the incident field even in the presence of resonant transition. The cause of the modified optical response of an optical medium is the laser induced coherence of atomic states, which leads to quantum interference between the excitation pathways that controls the optical response. In this way the absorption and refraction of the medium can be modified. Electromagnetically Induced Transparency (EIT) is a method eliminating the effects of a medium on a propagating beam of electromagnetic radiation. It gives rise to greatly enhanced nonlinear susceptibility in the spectral region of induced transparency of the medium and is associated with steep dispersion. • studied light-atom interaction for various atomic systems.• examined Coherent Population Trapping (CPT) and Electromagnetically Induced Transparency (EIT) for multi-level systems.• used semi classical approach where the atoms are treated quantum mechanically and the field is treated classically. • derived various aspects of CPT and EIT to understand enhancement of refractive index with low absorption, pulse storage and information processing, propagation of pulse without distortion, and lasing without inversion.

Education record

Education record

Doctor Of Philosophy - Phd, Theoretical And Mathematical Physics

Master'S Degree, Physical Sciences, Ranked 4Th In The Class.

Bachelor'S Degree, Physical Sciences

Education record