Learned Sorted Table Search and Static Indexes in Small-Space Data Models

Machine-learning techniques, properly combined with data structures, have resulted in Learned Static Indexes, innovative and powerful tools that speed up Binary Searches the use of additional space respect to table being searched into. Such is devoted machine-learning models. Although their infancy, these are methodologically practically important, due pervasiveness Sorted Table Search procedures. In modern applications, model a key factor, major open question concerning this area assess what extent one can enjoy speeding achieved by Indexes while using constant or nearly constant-space paper, we investigate mentioned (a) introducing two new models, i.e., k-ary Model Synoptic Recursive Index; (b) systematically exploring time–space trade-offs hierarchy existing ones reference software platform Searching on Data, together proposed here. We document novel rather complex trade-off picture, which informative for users as well designers Indexing structures. By adhering extending current benchmarking methodology, experimentally show competitive time space. Our second model, bi-criteria Piece-wise Geometric Index, achieve 0.05% more than taken table, thereby, terms proposals. The Index complement each other quite across various levels internal memory hierarchy. Finally, our findings stimulate research since they highlight need further studies regarding relation Indexes.