ENGLISH-magnetic-field-simulator
🧲 Magnetic Field Simulator (FEM Based)
Click here for Korean!
Click Here to download!
🛠️ Language and Libraries used in this project
In order to implement High-performance GPU computations and interactive 3D visualization, I used libraries below.
| Language | Library | Purpose | Official Document url |
|---|---|---|---|
| Python | CuPy | GPU based Parallel computing | CuPy |
| Â | NumPy | Post-processing of GPU computation results | Numpy |
| Â | Flask | Constructing Local web server for Magnetic field data interactions | Flask |
| JavaScript | Three.js | Real-time Magnetic field visualization via web browser | Three.js |
“Reveal the invisible world of Magnetic field!”
This project calculates and visualizes magnitude and direction of magnetic field in each point generated by user-given electric current, using Finite Element Method.
For research and education purpose, you can use this whatever you like. (Code modification is allowed, secondary distribution is not allowed.)
COMSOL is too expensive! I made this for moneyless students
You can run this in Google Colab WITHOUT any graphics card!
Introduction & Google Colab
Youtube video instructions are ready! Click image to watch. (English subtitle is ready)
| 🎥 Overall Instruction | 💻 How to use this on Google Colab |
|---|---|
 |
 |
| Introducing overall features | Run this on web browser without GPU |
Key Features
- 3D Vector Visualization: Real-time visualization of vector direction and magnitude at each point. (This code visualizes any
.csvfile containing 6 columns!) - User-given Electric Currents: Free to arrange shape(circle or line), intensity(Ampere), and position(3D vector) of the Electric current.
- Mesh dense and size adjustment: Adjust simulation space and resolution via Mesh Dense and size variables.
- Download Magnetic field data: Download computation results as
.csvformat in order to see it on Microsoft Excel or else.
🛠To run this in your computer, You need NVIDIA graphics card! If you don’t have one, run this in Google Colab
This project implements High-performance computations via NVIDIA GPU(CUDA)
- GPU: Can run ONLY on NVIDIA graphics card.
- Auto Setup: Execute
run.batfile by double click. It will install necessary libraries automatically in virtual environment. - My setup: CUDA 13.1, CuPy 13.6.0, NumPy 2.4.0 (If something goes wrong with the libraries, try similar versions.)
🚀 Brief instructions for who has GPU(Congrats! you are rich!)
0. Download this porject
- click here and download ZIP file.
- Unzip and open folder.
1. Execute run.bat file
- You will see “Unknown publisher” warning at first. That’s due to the absence of a digital signature certificate. There are no evil programs here. If you execute
run.batfile, it will automatically generate virtual environment and download necessary libraries. If download is completed,run.batwill automatically initiate local server. - If interface window doesn’t pop up, enter 127.0.0.1:5000 in your web browser.
2. Simulation settings
- Step 0: Define size and mesh dense of your simulation space.
- Step 1: Add your electric current information into “Request list” by clicking “Request” button after you finish entering electric current informations(shape, intensity, coordinates).
- Step 2: Click “Start simulation” after your “Request list”. It will compute via CuPy Library.
3. See results and terminate
- After the computation is complete, Three.js will automatically visualizes 3D magnetic field.
- Press Download as .csv button to save results as
.csvformat. Use this for further accurate analysis. - Press Simulation Turn Off button to terminate the program. This button will completely terminate any background processes.
🤝 Any Help is Welcome!
This project is a one-person project developed to provide students an academic tool to study magnetic fields for free.
Any feedback and assistance on these issues below is welcome!
- Magnetic field line visualization: Need to visualize smooth lines, instead of dirty/unintuitive arrows.
.exeformat packaging: PyInstaller crashes if I try to export this project into single.exepackage.- Better visualization: Reduce number of overlapping vectors. Or adjust individual opacity to see only important vectors.
- Make this into a server: construct GPU server that can be accessed via low-performance devices. (Partially resolved via Google Colab)
- Improve UI/UX: Children-friendly UI/UX. Or fix bug that
window.close()does not work. (I think this is because of Chrome’s security policy)
đź“ť License
This project can be used freely for Education and Research purpose. You can use this without any permission for your class or personal studies!🎓
Code modification is always allowed. Secondary distribution is not allowed. If you want to share, send url of this page.
If you have any feedbacks or questions, ask me via github or email(lcp020620@kaist.ac.kr).