Lookup Table¶
A prepared, addressable set of associations that returns a stored value or output for a supplied key instead of deriving or selecting it afresh.
Core Idea¶
A lookup table stores addressable input-to-output associations so a query can retrieve a prepared value instead of deriving or selecting it afresh. It has a supported key domain, stored entries, a rule for finding the relevant entry, and stated behavior for keys without one. Numerical samples, configured Boolean truth outputs and keyed data translations all fit; no grid, interpolation rule, expensive original calculation or universal no-miss guarantee is required.[ref-1298e871dc54][ref-fa7cbb259a07][^ibm]
The frozen Wikipedia sources are Trigonometric table and Generating trigonometric tables. They describe narrower table identities and remain candidates for separate review; this broader draft does not absorb them as exact aliases.
Scope of Application¶
Arm's CMSIS-DSP sine routine reads from 512 stored values and linearly interpolates. AMD's UltraScale FPGA LUT selects a configured Boolean output from six input bits without numerical interpolation. IBM documents keyed event tables whose lookup returns an associated value or an empty string if no key matches. These unlike implementations share prepared keyed retrieval but differ in precision, completeness, media and miss behavior.[ref-1298e871dc54][ref-fa7cbb259a07][^ibm]
Clarity¶
Ask what the key is, what output information is already stored, how the key selects it, and what happens outside the stored key set. A mathematical function need not be tabulated; a disk of arbitrary records is not a lookup table unless a key selects a result. A cache may be prefetched, so the distinction is not “cache fills late, table fills early”: caching exploits a fast local copy and reuse, while a lookup table exposes an intended keyed association. A branch table is a narrower use that selects a code target and transfers control.
Manages Complexity¶
The table can move repeated calculation or choice off the use-time path, but it incurs storage, configuration and sometimes update work. In a numerical table, fewer samples plus interpolation trades memory for runtime computation and approximation error. In a hardware LUT, finite input width and configured truth values govern resource use. In a sparse keyed table, omitted keys save entries but require explicit defaults. The name alone does not establish constant-time access, perfect accuracy or complete domain coverage.[ref-1298e871dc54][ref-fa7cbb259a07][^ibm]
Abstract Reasoning¶
For supported key \(k\), an addressing rule selects stored association \(T[k]\); optional postprocessing can combine selected entries. Remove the retained associations and the answer must be computed or chosen another way. Remove the addressing rule and the records cannot be queried as a table. An absent key may return a default or error rather than making the entire structure cease to be a lookup table. This model separates the shared retrieval core from optional interpolation, hashing or control transfer.[ref-1298e871dc54][ibm]
Knowledge Transfer¶
The portable diagnostic—prepared answer, key, selection rule and boundary behavior—travels from numerical software to digital hardware and event enrichment. Their extra rules do not: sine interpolation is not an FPGA requirement, and AMD's fixed logic delay is not a guarantee for software lookups. Live Function Mapping, Caching and Precomputation are related but have fuller distinct signatures; no strict DAG parent is asserted here. The narrower trigonometric-table source identities remain unresolved rather than silently promoted or rejected by this broader entry.
[^ref-1298e871dc54]: Arm, CMSIS-DSP “Sine”, description of 512-value table lookup and linear interpolation. [^ref-fa7cbb259a07]: AMD, UltraScale Architecture Configurable Logic Block User Guide UG574, “Look-Up Table”, six-input Boolean LUT capability. [^ibm]: IBM, “Lookup table operations,” Netcool/OMNIbus 8.1, keys, values and documented missing-key result.
Neighborhood in Abstraction Space¶
Lookup Table sits in a sparse region of the domain-specific corpus (68th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.
Family — Storage & Lookup Data Structures (21 abstractions)
Nearest neighbors
- Key–Value Database — 0.86
- Retrieval Data Structure — 0.85
- Existential Quantification — 0.84
- Multimap — 0.84
- Database application — 0.83
Computed from structural-signature embeddings · 2026-10-08