merging-orbits

N-body gravitational simulator where object collisions are handled by a simplified body merger heuristic.

Basic setup

• Install Anaconda3 on your system
• Create a new conda virtual environment with Python version 3.10:

(or you can create a new project with PyCharm and select Conda as your interpreter, it will automatically create a new conda env)

(If using windows powershell first run conda init powershell, restart the shell and try to activate the env)

! The current latest compatible version of Python is v3.10 !

conda create -n merging_orbits_env python=3.10
conda activate merging_orbits_env

• Install all dependencies with PIP:

pip install vtk
pip install numpy
pip install Pandas
pip install moviepy
pip install mayavi

Optional setup for the GPU accelerated version only

For the GPU accelerated version you must have an NVIDIA GPU with CUDA API version 11 support. NVIDIA CUDA Toolkit DOWNLOAD

• Download version 11 for your architecture
• Reboot your system
• Activate the conda env again and install the pycuda python library

pip install pycuda

( if you don't have Visual Studio installed you need to install it for the NVCC compiler and cl tool - install with the C/C++ toolkit - and remember to add the cl.exe directory C:\Program Files\Microsoft Visual Studio\2022<yourversion>\VC\Tools\MSVC<yourversion>\bin\Hostx64\x64 to PATH )

• For the accelerated version run the simulator_accel.py script

How to run a simulation

• Set the initial conditions and rendering options in config.json
• Run simulator.py (simulator_accel.py for the GPU accelerated version)
• Run render.py

To do list

• Implement Barnes-Hut algorithm with octree
• Implement a 3D spatial partitioning strategy to speed up the collision handling/detection
• GPU Acceleration with CUDA

Gallery