Generate a dataset of hundreds of millions of chaotic double pendulum simulations at various angles, lengths, and simulation times. Then use Deep Learning with pytorch train a Deep Multi-Layer Perceptron Neural Network to predict where the pendulums will be.

I highly suggest reading through Simulations.c to make sure variables like STEPS, SIMS, and buffer_size are tuned to your computer's hardware capabilities

Open Double-Pendulums-AI in a single Terminal

• 1 export OMP_NUM_THREADS=<threads>replace <threads> with your desired thread count.
• 2 gcc -O3 -march=native -ffast-math -fno-unsafe-math-optimizations -funroll-loops -fno-trapping-math -fno-signaling-nans -fopenmp Simulations.c -lm -o simulations
• 3 ./simulations

I have not officially tested this so make an issue if it does not work. I will remove this once i have tested it

Open Double-Pendulums-AI in a single Powershell / Windows Terminal DO NOT USE COMMAND PROMPT.

• Install gcc (MinGW-w64 is recommended) if it is not already installed: https://www.google.com/search?q=how+to+install+gcc+on+windows you can check if it is installed with gcc --version
• Confirm your version of gcc supports OMP gcc -fopenmp -v update if not.
• 1 Set threads to use with $env:OMP_NUM_THREADS=<threads> replace <threads> with your desired thread count.
• 2 Compile with gcc -O3 -march=native -ffast-math -fno-unsafe-math-optimizations -funroll-loops -fno-trapping-math -fno-signaling-nans -fopenmp Simulations.c -lm -o simulations
• 3 Run with ./simulations.exe

Open a single terminal

• If you already have a pip venv active you can skip to step 3.
• 1 python3 -m venv ~/p1p
• 2 source ~/p1p/bin/activate
• 3 pip install --upgrade pip
• 4 pip install numpy wandb torch heavyball
• If you already have tkinter installed you can skip to step 6, you can check for tkinter with python3 -m tkinter
• 5 sudo apt update && sudo apt install python3-tk
• 6(only for IDE users) In your IDE of choice set the python env variable to ~/p1p/bin/python or your own pip venv if you already have one.
• 7 python3 CSVtoBinary.py or run with the IDE of your choice.
• 8 python3 Train.py or run with the IDE of your choice.
• After your model has finished training open Evaluate.py, choose the model of your choice from Models/ and run with python3 Evaluate.py

I have not officially tested this so make an issue if it does not work. I will remove this once i have tested it

Open a single powershell terminal

If you already have a pip venv active you can skip to step 3

• Install python if you do not already have it installed: https://www.python.org/downloads/windows/
• Modern installations of python should include tkinter, check it exists with python -m tkinter
• 1 python -m venv C:\Users\<YourUsername>\p1p replace <YourUsername> with your windows username.
• 2 C:\Users\<YourUsername>\p1p\Scripts\Activate.ps1 replace <YourUsername> with your windows username.
• 3 python -m pip install --upgrade pip
• 4 python -m pip install numpy wandb torch heavyball
• 5(only for IDE users) In your IDE of choice set the python env variable to C:\Users\<YourUsername>\p1p\Scripts\python.exe replace <YourUsername> with your windows username.
• 6 python CSVtoBinary.py or run with the IDE of your choice.
• 7 python Train.py or run with the IDE of your choice.
• After your model has finished training open Evaluate.py, choose the model of your choice from Models/ and run with python Evaluate.py