GSoC 2018: Parallel Graph Algorithms in Julia LightGraphs
This technical report summarizes the outcomes of a Google Summer of Code 2018 project focused on enhancing the LightGraphs.jl library within the Julia programming ecosystem. The developer implemented parallel versions of critical graph analysis algorithms and optimized sequential implementations to improve performance. Key achievements include parallel implementations for Breadth-First Search, PageRank, and various centrality measures, alongside heuristics for NP-Hard problems like Minimum Vertex Cover. Benchmarks conducted on a 64-bit Linux machine with four cores demonstrated significant speed-ups; for instance, parallel Breadth-First Search achieved up to 2.59x acceleration on Astro-Physics collaboration data. Sequential optimizations also yielded substantial gains, with Prim's MST algorithm showing up to 7.65x improvement. While many features were merged into the main codebase, some parallel implementations were excluded due to slower performance compared to sequential counterparts. The project highlights advancements in high-performance graph computing using Julia, contributing valuable tools for scientific and social network analysis.
Wire timeline
GSoC 2018: Parallel Graph Algorithms in Julia LightGraphs
This technical report summarizes the outcomes of a Google Summer of Code 2018 project focused on enhancing the LightGraphs.jl library within the Julia programming ecosystem. The developer implemented parallel versions of critical graph analysis algorithms and optimized sequential implementations to improve performance. Key achievements include parallel implementations for Breadth-First Search, PageRank, and various centrality measures, alongside heuristics for NP-Hard problems like Minimum Vertex Cover. Benchmarks conducted on a 64-bit Linux machine with four cores demonstrated significant speed-ups; for instance, parallel Breadth-First Search achieved up to 2.59x acceleration on Astro-Physics collaboration data. Sequential optimizations also yielded substantial gains, with Prim's MST algorithm showing up to 7.65x improvement. While many features were merged into the main codebase, some parallel implementations were excluded due to slower performance compared to sequential counterparts. The project highlights advancements in high-performance graph computing using Julia, contributing valuable tools for scientific and social network analysis.
JuliaLang - The Julia programming language