Piston Workshops - Introduction to Physics Informed Neural Networks (PINNs)
The IEEE NITK Piston Special Interest Group inaugurated its new series, Piston Workshops, aimed at providing students with exposure to emerging technologies and software through structured, hands-on learning sessions. The initiative seeks to bridge theoretical understanding with practical implementation, ensuring participants gain both conceptual clarity and technical proficiency.
The first session under this series was conducted on 23rd and 24th October 2025 by Mr. Nishant Patil, Vice Chair of IEEE Piston. The topic of the workshop was Physics Informed Neural Networks (PINNs), a cutting-edge approach that integrates physical laws into neural network training for solving complex engineering problems.
Participants were encouraged to bring their laptops and actively engage in coding exercises using Google Colab in Python.
Date and Time
Location
Hosts
Registration
-
Add Event to Calendar
- National Institute Of Technology Karnataka
- Surathkal
- MANGALURU, Karnataka
- India 575025
- Building: LHC C
- Room Number: CR 5
Speakers
Nishant Vinod Patil
Agenda
Day 1 of the workshop focused on building foundational understanding. Attendees were introduced to the basic framework of Neural Networks (NNs) and Physics Informed Neural Networks (PINNs). The session included demonstrations of key Python libraries such as NumPy, TensorFlow, and Matplotlib, which were used to construct and visualize neural networks. Participants also learned to prepare customized loss functions tailored to differential equations, emphasizing the fusion of machine learning with physics-based modeling.
Day 2 explored practical implementations of PINNs in solving engineering problems. Through guided examples, students appreciated the potential of PINNs in addressing classical problems such as 2D heat conduction and the Poisson equation. The session concluded with discussions on the advantages, limitations, and future scope of applying PINNs in computational engineering.
