Home
Hi, I am Tapas Tripura. I am a PMRF fellow at the Department of Applied Mechanics, IIT Delhi. I obtained my master's and bachelor's in Civil Engineering from IIT Guwahati and Nerist, respectively. In my master's, I extensively worked on stochastic dynamics, stochastic calculus, and machine learning to develop stochastic integration schemes and real-time system identification tools. Currently, at IIT Delhi, I am working on model-based stochastic control. Instead of using the approximated model, I have developed an algorithm for identifying the governing physics of stochastic dynamical systems. The discovery architecture uses a Bayesian machine learning algorithm to discover the governing physics of stochastic dynamical systems. To enable faster predictions during control, I have also developed data-driven and physics-informed deep learning solvers. The deep learning solvers integrate the robustness of existing neural operators with wavelet decomposition for learning parametric differential equations. In a parallel path, I have developed a robust model agnostic predictive control framework for stochastic control of dynamical systems. I am currently exploring deep learning alternatives to avoid costly optimization steps in model predictive control.
Hi, I am Tapas Tripura. I am a PMRF fellow at the Department of Applied Mechanics, IIT Delhi. I obtained my master's and bachelor's in Civil Engineering from IIT Guwahati and Nerist, respectively. In my master's, I extensively worked on stochastic dynamics, stochastic calculus, and machine learning to develop stochastic integration schemes and real-time system identification tools. Currently, at IIT Delhi, I am working on model-based stochastic control. Instead of using the approximated model, I have developed an algorithm for identifying the governing physics of stochastic dynamical systems. The discovery architecture uses a Bayesian machine learning algorithm to discover the governing physics of stochastic dynamical systems. To enable faster predictions during control, I have also developed data-driven and physics-informed deep learning solvers. The deep learning solvers integrate the robustness of existing neural operators with wavelet decomposition for learning parametric differential equations. In a parallel path, I have developed a robust model agnostic predictive control framework for stochastic control of dynamical systems. I am currently exploring deep learning alternatives to avoid costly optimization steps in model predictive control.
Glimpses on my work at IIT Delhi as PMRF
Glimpses on my work at IIT Delhi as PMRF
🔗 Data-driven Physics Discovery of Dynamical systems
🔗 Data-driven Physics Discovery of Dynamical systems
Discovering interpretable Lagrangian of dynamical systems from data
Discovering interpretable Lagrangian of dynamical systems from data
MAntRA: A framework for model agnostic reliability analysis
MAntRA: A framework for model agnostic reliability analysis
🔗 Model Agnostic Stochastic Model Predictive Control (MASMPC)
🔗 Model Agnostic Stochastic Model Predictive Control (MASMPC)
Robust model agnostic predictive control algorithm for randomly excited dynamical systems
Robust model agnostic predictive control algorithm for randomly excited dynamical systems
🔗 Predictive Digital Twin
🔗 Predictive Digital Twin
🔗 Deep Neural Operator for Computational Mechanics
🔗 Deep Neural Operator for Computational Mechanics
Wavelet Neural Operator based elastography for tumors
Wavelet Neural Operator based elastography for tumors
Fault detection and isolation using probabilistic wavelet neural operator
Fault detection and isolation using probabilistic wavelet neural operator
Neural Combinatorial Wavelet Neural Operator for continual learning
Neural Combinatorial Wavelet Neural Operator for continual learning
Preview: https://arxiv.org/abs/2310.18885