We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Jim Kitchen & Erik Welch - GraphBLAS for Sparse Data and Graphs | SciPy 2023
Join Jim Kitchen and Erik Welch at SciPy 2023 as they introduce GraphBLAS, a Python library for sparse data and graphs, offering fast, stable, and scalable solutions for graph processing and algorithms.
- Python GraphBLAS for single-source shortest path: uses adjacency matrix, level, and parents; iterates until converged
- GraphBLAS is fast, stable, and scalable; uses sparse accelerations and hardware being worked on
- Key features: semirings, sparse linear algebra API, simple and expressive
- Math specification: defines a syntax and rules for graph algorithms; allows for expressing complex algorithms in a simple way
- Example algorithms: page rank, single-source shortest path, breadth-first search
- Target audience: graph theorists, computer scientists, researchers
- Python GraphBLAS is designed for longevity and maintainability
- Need for a graph library: current solutions are not efficient or scalable
- Example use cases: social networks, biology, genomics, power grids
- Open-source and community-driven project; collaboration with NetworkX
- Vision: to accelerate libraries that use NetworkX; create a community around graph processing
- QGraph is another solution for graph processing, which is designed for scalability and efficiency.