At MAVENs, we study how local electronic structure evolves into collective material behaviour — and why that transition so often determines whether a material ultimately functions.

A magnetic moment may produce a paramagnet rather than a ferromagnet. A chemically ideal adsorption site may sit in a matrix that suppresses catalytic turnover. A defect with the correct spin state may still fail to support coherent transport if the surrounding defect network does not percolate.

Local electronic features are necessary for functionality, but they are rarely sufficient for it.

What decides whether a local feature survives into a macroscopic response is usually not the feature itself, but the larger physical structure it inhabits: the topology of exchange pathways, the statistics of disorder, the connectivity of defect networks, or the dimensionality of an interacting phase.

This intermediate scale — the physical bridge between local structure and collective response — is the central focus of our research.

Isolating the Intermediate Scale

We choose systems — substitutional alloys, layered surfaces, defect-bearing semiconductors, two-dimensional magnets — where the intermediate scale takes a specific, computationally tractable form, and where one variable at a time can be cleanly isolated. The recurring question across these systems is the same:

When does a local descriptor remain predictive, and when does an intermediate scale become decisive?

Methods Chosen by the Question

Methods enter only when a question requires them.

We use:

• Density functional theory for electronic and magnetic structure, primarily through plane wave and Green’s-function coherent potential approximation based methods.
• First-principles exchange interactions extracted through the Lichtenstein formalism, preserving the topology of competing magnetic pathways when mapping finite-temperature behaviour onto effective spin models.
• Classical Monte Carlo simulations to study ordering temperatures, phase competition, and magnetocaloric response.
• Machine learning constrained by physically interpretable descriptors when compositional spaces become too large for direct first-principles exploration.

Part of what we report is where local descriptors succeed — and where they fail. That boundary is often where the intermediate scale becomes physically important.

Research Software and Computational Infrastructure

We also develop computational tools where existing workflows become restrictive.

cview is an open-source crystallographic interface designed to bridge structure visualisation and ab-initio workflow generation across multiple formats.

A kinetic Monte Carlo engine for defect evolution and ordering dynamics is currently under active development.

Experimental Constraints and Validation

Because collective behaviour is ultimately constrained by synthesis, morphology, and disorder, we work closely with experimental collaborators to ensure that computational predictions remain physically accessible.

Current collaborations include:

• Dr Ashutosh Kumar Singh, CeNS — quantum materials
• Dr Payel Bandyopadhyay, SRMIST — polymers and soft materials
• Dr Pralay K. Santra, CeNS — two-dimensional materials

Research Themes