https://kanand-cfd.github.io/Karan Anand, PhDCFD engineer specialized in numerical simulation, multiphase flow modeling and HPC. 2025-07-30T16:12:34+02:00 Karan Anand https://kanand-cfd.github.io/ Jekyll © 2025 Karan Anand /assets/img/favicons/favicon.ico /assets/img/favicons/favicon-96x96.png PINNs: Hard and Soft Constraints for 2D Heat Equations2025-07-26T08:00:00+02:00 2025-07-30T16:12:17+02:00 https://kanand-cfd.github.io/posts/pinn-2d-heat-equation/ Karan Anand Solving the 2D Heat Equation using Physics-Informed Neural Networks (PINNs) In this notebook, we demonstrate how to solve the 2D time-dependent heat equation using a Physics-Informed Neural Network (PINN). Unlike traditional machine learning models that rely heavily on large datasets, PINNs embed physical laws (PDEs) directly into the loss function. We solve the following PDE: \[\frac{\parti... Human Pose Estimation from Images2025-07-19T08:00:00+02:00 2025-07-19T23:15:32+02:00 https://kanand-cfd.github.io/posts/pose-estimation-human.md/ Karan Anand Human Pose Estimation from Images By Karan Anand, PhD Motivation Human pose estimation is a fundamental task in computer vision, aiming to predict key joint positions (like head, shoulders, wrists, etc.) from an image. This project explores a bottom-up keypoint detection approach using heatmaps and builds the training pipeline from scratch using COCO keypoints data. Project Objectives ... Balancing Chaos: A Physicist’s Dive into Reinforcement Learning with CartPole2025-06-21T08:00:00+02:00 2025-06-24T01:31:46+02:00 https://kanand-cfd.github.io/posts/cartpole/ Karan Anand Balancing Chaos: A Physicist’s Dive into Reinforcement Learning with CartPole What do rocket nozzles, humanoid robots, and an inverted pendulum have in common? They all live on the edge of instability. And now, so does our code. In this project, I built a physics-based CartPole environment from scratch, visualized its motion, and trained a reinforcement learning agent (PPO) to master the ... Bubble rise through water in 2D using Basilisk2025-06-11T08:00:00+02:00 2025-06-11T16:18:12+02:00 https://kanand-cfd.github.io/posts/bubble-riser-2D/ Karan Anand Predicting Bubble Rise Velocity in Liquids using ML and Basilisk The rise of a gas bubble in a liquid is a classic multiphase flow problem, rich in fluid mechanics and practical relevance — from boiling and cavitation to bioreactors and nuclear cooling. In this project, we simulate 2D bubbles rising in a quiescent liquid using the Volume of Fluid (VOF) method in Basilisk, and then train a mac... Surrogate Modeling for Pressure Drop using PartiNet2025-06-08T08:00:00+02:00 2025-06-09T13:26:14+02:00 https://kanand-cfd.github.io/posts/fluidized-bed-surrogate/ Karan Anand This notebook uses synthetic fluidized bed theory and trains machine learning models to predict pressure drop. 🧪 Objective To build fast surrogate models that predict pressure drop in a gas-solid fluidized bed using Random Forest, Linear Regression, and MLP — trained on synthetic data derived from fluid mechanics theory. ⚙️ Libraries and Data Generation import numpy as np import pandas ...