2D Rigid Body Physics Engine
I developed a custom 2D rigid body physics engine inspired by Box2D, focusing on clarity of physical modeling and numerical stability. The engine implements rigid body dynamics with linear and angular motion, impulse-based collision resolution, and iterative constraint solving. Core features include broad-phase and narrow-phase collision detection, contact manifolds, restitution and friction modeling, and a Sequential Impulse (Gauss–Seidel) solver derived from Lagrange multiplier formulations. You can run the Impulse 2D Demo directly in your browser by clicking this link. The notes for the Impulse 2D physics engine are available here.
Surface Model for Water Drop Animation
As a postdoctoral researcher, we proposed a deformable surface model to simulate the flow behavior of water droplets on different solid surfaces. In the Lagrangian coordinate system, at each time step, the implicit curvature flow operator is used to simulate the surface tension of water, the contact angle operator is used to change the shape of water droplets on the solid surface, and the topological structure of the mesh is optimized to improve the robustness of the mesh. Through a series of numerical experiments, it is proved that real-time rendering can be achieved. It was published in IEEE.
First-principles simulation of spin-transfer torques
Recent experiments on magnetic tunneling systems and domain-wall dynamics require a full quantum-mechanical description. We study spin-transfer torques (STTs) in layered magnetic systems with noncollinear magnetization using a first-principles quantum transport approach.
The scattering problem of the open system is solved by matching the wave function in the scattering region with Bloch states in the leads. Based on the resulting scattering wave function, spin-transfer torques are evaluated from the spin-dependent electronic states.
This work provides a physically grounded and efficient framework for simulating current-induced torques in nanoscale magnetic systems. [pdf]
Volume Rendering Engine
Lead the team to develop the core algorithm( Ray-Casting) based on OpenGL/HLSL and optimize the algorithm to meet the requirements of real-time 3D rendering of medical image data and various complex interactive operations , such as splitting, hiding, group,etc.
VR/AR/3D printing in Medical Image Application
Lead the team to develop many interesting product prototypes, which is based on VR/AR and 3D printing techniques. They were exhibited at the CMEF exhibition and were reported and noticed by the many medias. It can render real medical image data and provide effective supporting for surgical planning.
Medical Post-Processing Algorithm
Develop colon segmentation algorithms, colon centerline extraction, and colon polyp CAD diagnosis algorithms. Visualize the neuro fibers and MR brain Images based on OpenGL(HLSL). Intel IPP technology is used for acceleration. Algorithms have been integrated into the products. It has more than 10 related patents.