NodeCompressionTable

NodeCompressionTable is a compact data structure used to map node identifiers to a dense, reduced set of codes in graph or tree representations. Its primary purpose is to decrease memory usage and improve cache locality when handling large graphs by replacing potentially large or sparse node IDs with small, contiguous codes.

A NodeCompressionTable typically consists of a forward mapping from original node identifiers to compressed codes, and often an optional inverse mapping to recover the original IDs. Additional metadata may accompany the table, such as the total number of nodes, the bit width of the codes, and integrity checks. Some implementations store only the forward mapping and reconstruct the inverse on demand, trading some lookup time for reduced memory.

Construction and usage commonly involve renumbering nodes into a dense range (for example, 0 to N-1) or employing a minimal perfect hash function to provide a compact, collision-free mapping. Delta encoding or other compressed representations may be used for the table itself. In practice, the NodeCompressionTable is used alongside adjacency or edge lists by substituting original node IDs with their compressed codes, thereby enabling more compact graph representations and faster traversal in memory.

Performance characteristics depend on the original identifier space and the chosen encoding method. Benefits include reduced memory footprint and improved data locality, while costs include the overhead of maintaining the table and potential lookup latency, especially for the inverse mapping. Concurrency and update behavior vary by implementation, with some designs favoring static graphs and others supporting incremental updates.

Applications span large-scale graph databases, network simulations, and graph-serialization formats. Related concepts include node ID remapping, dictionary encoding, minimal perfect hash functions, and various graph compression techniques.