Research
Few selected publications are given below. You can find the latest list of publications in my Google Scholar page.
2026
Learning Physical Operators using Neural Operators
OpsSplit introduces a physics-informed machine learning framework that decomposes partial differential equations into fixed linear approximations and learned non-linear neural operators within a Neural ODE, delivering superior generalization to unseen physics, parameter efficiency, and interpretability.
2025
Calibrated Physics-Informed Uncertainty Quantification
Calibrated uncertainty quantification of neural PDE solvers using physics residual errors as non-conformity scores for data-free conformal prediction.
2024
Uncertainty Quantification of Surrogate Models using Conformal Prediction
Guaranteed and valid error bars across spatio-temporal domains using conformal prediction.
Valid Error Bars for Neural Weather Models using Conformal Prediction
Marginal conformal prediction as a method of guaranteed error bars across neural weather models.
Plasma Surrogate Modelling using Fourier Neural Operators
Multi-variable FNO designed to model the plasma evolution within a Tokamak across both simulations and experiment on the MAST Tokamak.
2023
Fourier-RNNs for Modelling Noisy Physics Data
Recurrent Fourier neural operators with hidden state representations for non-Markovian physical modelling.
Loss Landscape Engineering via Data Regulation on PINNs
Impact Data-Regulation has on smoothening the loss landscape of physics-informed neural networks for better convergence.
2022
Image Mapping the Temporal Evolution of Edge Characteristics in Tokamaks using Neural Networks
Branched fully convolutional network designed to emulate the plasma at the scrape-off layer with coupled plasma and neutral behaviour.
