PINN-s-for-Heat-Transfer-Problem

Name: Anubhav Joshi

Scientific Machine Learning

In recent years, the use of physics-informed neural networks (PINNs) has gained popularity across several engineering disciplines due to their effectiveness in solving linear and non-linear partial differential equations (PDEs) and real-world problems despite noisy data. The basic approach used to solve the PINNs is to construct a neural network and define a loss function as a function of the PDE and boundary/initial conditions (B.C./I.C). To get the dependent variable from the PDE, our aim is to minimise the loss function formed from fundamental equations by employing effective optimisation techniques.