Introduction to GraphBLAS: A linear algebraic approach for concise, portable, and high-performance graph algorithms

This tutorial describes the theoretical background of GraphBLAS. First, we discuss the need for a standard for graph algorithms. Then, we define the key concepts in GraphBLAS such as sparse matrix multiplication, semirings, and masked matrix operations. We illustrate their usage through textbook graph algorithms including BFS, single-source shortest paths, triangle count, PageRank as well as more advanced graph algorithms such as community detection, local clustering coefficient, and bidirectional BFS. Finally, we provide a collection of GraphBLAS tools and resources for learning more about GraphBLAS.