Mohammad Samara/Physics-Nemo [Modulus ] : Advanced Topics
Buy now

Physics-Nemo [Modulus ] : Advanced Topics

  • Course
  • 73 Lessons

Advanced Simulations with AI

  • 10 hours on-demand video

  • Certificate of completion

  • Full access

55$ Special Discount price : 10$

Buy Now

What you'll learn

  • I-PINNs for 2D heat sink flow problem .

  • DeepONet for  Integration problem.

  • Fourier Neural Operator FNO for  Darcy problem.

  • PINNs for  3D Linear Elasticity Problem.

  • PINNs for  3D Fluid/ Solid Multi Domain Calculation.

  • PINNs for 3D Geometric Optimization for Heat Exchanger Flow Problem.

Contents

Section 1: Introduction

lecture 1: Introduction
Preview
lecture 2: Course Structure
Preview
ref.txt

Section 2: Inverse PINNs

Section-2 Files
lecture 3: Inverse PINNs Theory
lecture 4: Define the Problem
lecture 5: Define the Data
Preview
lecture 6: Define the Config File
lecture 7: Import Needed Libraries
lecture 8: Define the Governing Equation
lecture 9: Define the Deep Neural Network
lecture 10: Add the Data
lecture 11: Add Inverse Value Monitor
lecture 12: Solve
lecture 13: Results Post Processing

Section 3: Deep Neural Operator (DeepONet)

Section-3 Files
Lecture 14: Deep Neural Operators Theory
lecture 15: Define the Problem
Preview
lecture 16: Define the Data
lecture 17: Define the Config File
lecture 18: Import Needed Libraries
lecture 19: Define the Deep Neural Network
lecture 20: Load the Data
lecture 21: Add the Data Constraint
lecture 22: Add Results Validator
lecture 23: Solve
lecture 24: Results Post Processing

Section 4: Deep Neural Operator (FNO - Fourier Neural Operator)

Section-4 Files
lecture 25: Fourier Neural Operator (FNO) Theory
Preview
lecture 26: Darcy Flow Problem
lecture 27: Define the Config File
lecture 28: Import Needed Libraries
lecture 29: Define the Data
lecture 30: Define the Deep Neural Network
lecture 31: Add the Data Constraint
lecture 32: Add Validator
lecture 33: Solve
lecture 34: Results View

Section 5: 3D Bracket Stress Analysis

Section-5 Files
lecture 35: 3D Stress Analysis Problem
lecture 36: Define the Config File
lecture 37: Import Needed Libraries
lecture 38: Define the Gov. Eq.
lecture 39: Define the DNN
lecture 40: Define the geometry - part a
lecture 41: Define the geometry - part b
lecture 42: Define the B.C, Interior Constraints
lecture 43: Solve
lecture 44: Results Post Processing
Preview

Section 6: PINNs Multi Domain Calculation

Section-6 Files
lecture 45: 3D Flow Solid Multi Domain Problem
Preview
lecture 46: Flow configuration file
lecture 47: Define the geometry - part A
lecture 48: Define the geometry - part B
lecture 49: Define the geometry - part C
lecture 50: Flow: Import Needed Libraries
lecture 51: Flow: Define the Gov. Eq. / DNN
lecture 52: Flow: Define the B.C, Interior Constraints
lecture 53: Flow: Add Monitor
lecture 54: Thermal : configuration file
lecture 55: Thermal : Import Needed Libraries
lecture 56: Thermal : Define the Gov. Eq. / DNN
lecture 57: Thermal : Define the B.C, Interior Constraints
lecture 58: Run Flow Code
lecture 59: Run Thermal Code
lecture 60: Results Review

Section 7: Geometric Optimization using PINNs

Section-7 Files
lecture 61: Parameterized 3D Geometry
lecture 62: Geometry Update
lecture 63: Flow Network Update
lecture 64: Thermal Networks Updated
Preview
lecture 65: Run (Retrain All)
lecture 66: Results view

Requirements

  • High School Math

  • Basic Python knowledge

Description

Description

This course is related with Advanced topics related with PINNs using NVIDIA Modulus. We will cover the topics of Inverse PINNs, Deep Neural Operator Network with DeepONet, Deep Neural Operator Network using Fourier Neural Operator (FNO), PINN for 3D Linear Elasticity Problem, PINNs for Multi Domain Calculation, and Geometric Optimization using PINNs.

What skills will you Learn:

In this course, you will learn the following skills:

  • Understand the Math behind solving partial differential equations (PDEs) with PINNs, I-PINNs,  Deep Neural Operator Network for DeepONet, along with FNO, Multi Domain Calculation and finally Geometric Optimization using PINNs.

  • Write and build Machine Learning Algorithms to solve PINNs using Nvidia Modulus.

  • Postprocess the results.

  • Pre-process the data and upload it to Nvidia Modulus.

  • Use opensource libraries.

We will cover:

  • Inverse Physics-Informed Neural Networks (I-PINNs) Solution for 2D heat sink flow problem .

  • Deep Neural Operator Network (DeepONet) Solution for  Integration problem.

  • Deep Neural Operator Network Fourier Neural Operator (FNO) Solution for  Darcy problem.

  • Physics-Informed Neural Networks (PINNs) Solution for   3D Linear Elasticity Problem.

  • Physics-Informed Neural Networks (PINNs) Solution for   3D Fluid/ Solid Multi Domain Calculation.

  • Physics-Informed Neural Networks (PINNs) Solution for 3D Geometric Optimization for Heat Exchanger Flow Problem.

If you do not have prior experience in Machine Learning or Computational Engineering, that's no problem. However it is recommended to have knowledge in the basics of the use and code running using Nvidia Modulus.

Let's enjoy Learning Nvidia Modulus together.

Hear from our happy students

"Would be great if we could have more of the FE related examples as well. For Example, how can we leverage PINNS when we have files from solvers, for instance if we have a .bdf , .dat file and .HDF5 file how can we use PINNS to solve related problems"

Chirag Shetty

"Thank you for your explanation, I enjoyed this course."

June Alba

"Interesting Topics."

Sara Harmon

"Simple Explanation with explanation of the basic math behind the topic of interest."

Koritos anagostabilo