Recursive Model Index Search
Use a hierarchy of learned models to predict a key position, then search inside a bounded error range.
3 notes
Use a hierarchy of learned models to predict a key position, then search inside a bounded error range.
Cache-efficient and hardware-tuned search: branchless binary search, Eytzinger and vEB layouts, learned indexes, SIMD search, and galloping intersection.
Use a predictive model to estimate where a key should appear, then search within a bounded local range.