Reproduce “FM-index for dummies” 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
Szymon Grabowski, Marcin Raniszewski, Sebastian Deorowicz
Researchers at Lodz University of Technology and Silesian University of Technology engineered new FM-index variants that achieve 2-3 times faster count query performance than existing methods by introducing a cache-friendly rank primitive and specialized index architectures. These variants prioritize search speed, resulting in a moderate increase in memory footprint (1.5-5 times larger).
Key insights
- Prioritizing search speed through cache-aware design and hardware-optimized primitives is critical for translating theoretical compressed data structure advantages into practical, high-performance implementations.
- Multiary wavelet trees, particularly Huffman-shaped 4-ary and 8-ary versions, can offer competitive space-time trade-offs when implemented with practical optimizations, challenging assumptions about their real-world utility.
- Tailoring FM-index architectures and encoding strategies to specific alphabet characteristics (e.g., very small vs. larger general alphabets) significantly enhances search performance and efficiency.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.