Reproduce “FenixWalk: Adaptive Sampling for Large-Scale Dynamic Graphs on Heterogeneous Architectures” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Pinhuan Wang, Jianian Fu, Jie Zhang, Shixuan Sun, Qing Wang, Xiaoliang Wang, Chen Tian, Chengying Huan
FenixWalk keeps the full graph and sampler state in host memory, while placing frequently reused vertex samplers on the GPU and transferring compact walker state during execution. Its Fenwick-based representation supports rejection sampling and prefix sampling with updates. On DeepWalk, it reports a 3.00× geometric-mean speedup over KnightKing across completed datasets, under the paper’s synthetic-update and synthetic-weight protocol.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.