PINNs Package: Dynamics, Heat Transfer & CFD

Physics-Informed Neural Networks are transforming computational engineering, but almost every resource out there is either a dense academic paper or a surface-level blog post with no working code.

This bundle changes that.

The Complete Series takes you from first principles all the way to solving coupled Navier-Stokes with an energy equation, entirely in PyTorch, through hands-on Jupyter notebooks you can run in your browser for free. No prior machine learning experience required.

What's included:

Mastering Dynamics PINNs (9 notebooks): The foundation. Covers the PINN framework from scratch — neural networks, automatic differentiation, and residual-based training — applied to real mechanical systems. Starts with a damped nonlinear pendulum, builds through parameterised PINNs, hybrid data and physics training, and finishes with an inverse PINN that identifies unknown equation terms from noisy measurements. Three animation tutorials turn your trained results into shareable mp4 videos. By the end you will have built every core PINN component from the ground up.

Mastering Heat Transfer PINNs (8 notebooks): Extends the framework to PDEs with spatial dimensions. Covers 1D steady conduction with Dirichlet and Neumann boundary conditions, 2D steady heat on a square domain, 1D transient heat with full space-time training, inverse identification of unknown thermal conductivity from noisy data, and spatially varying conductivity in functionally graded materials. Three animation tutorials visualise the learning process — including watching k-hat converge toward its true value in real time. Each worksheet is validated against an analytical solution.

Mastering CFD PINNs (10 notebooks): The flagship series. Starts with 1D and 2D convection-diffusion, introduces pressure-velocity coupling and incompressibility through 2D Stokes flow, solves the full nonlinear Navier-Stokes equations at Re=100 with benchmark comparison against Ghia et al. (1982), validates a Kovasznay flow PINN against Raissi et al. (2019) with full L2 error reporting, and finishes with coupled Navier-Stokes and energy equations for buoyancy-driven natural convection. Three animation tutorials animate the 2D flow and temperature fields as they are learned.

The complete value stack:

27 interactive Jupyter notebooks, each in fill-in-the-blank worksheet and complete answers versions. 3 technical handbook PDFs. 1 Google Colab quick-start guide. 9 animation tutorials. Validated against analytical solutions, Ghia et al. (1982), Raissi et al. (2019), and Mao et al. (2020).

How it works:

Every worksheet is a Jupyter notebook with the complex code pre-written. You fill in clearly signposted blanks — always a single value or short formula — answer conceptual questions, and run the cells in order. Upload directly to Google Colab and run it in your browser on a free GPU with zero setup required. Answer keys are included for every worksheet.

Who this is for:

Engineering students, researchers, and practising engineers who already know their physics and want to learn how to solve it with PINNs. Basic Python assumed. No prior machine learning or PyTorch experience needed.

👉 Buy the complete series