CUDA optimized c++ implementation of the North Star Algorithm for gravitational wave localization.

Localizing gravitational waves requires algorithms capable of processing large scales of data efficiently. The NorthStar pipeline originally written in python provides a substructure for this task however it faces some challenges in speed and performance. This project attempts to rewrite the NorthStar pipeline using C++ as well as optimizing it using the CUDA libraries to essentially accelerate the GPU. Initially google colab was used to optimize the code using the CUDA libraries, but hardware compatibility issues ended up restricting the performance. The project also faced challenges in regard to algorithmic outputs that did not fully align with the original python code. To overcome these challenges a laboratory computer was used instead and AI tools were implemented to debug code and improve Kernel execution. The resulting C++ rewrite successfully compiled and executed however it was only partially optimized as output inconsistencies remained. Nevertheless, this project provided a valuable understanding on the complexity of CUDA based optimization as well as the difficulties that arise when trying to ensure consistency across different programming languages to lay out the groundwork of gravitational wave localization.