Diatom Bitmap: Density-Derived Container Boundaries and Cooperative Thermoregulation for Compressed Set Membership

The Diatom Bitmap is a bitmap primitive that replaces fixed arithmetic partitioning with density-derived container boundaries and replaces static container managementwith cooperative thermoregulation. These two mechanisms target a recurring constraint classin analytics infrastructure: related bitmap families whose members are built, queried, andmaintained together, where shared morphology and coordinated maintenance can eliminatestructural waste that fixed-boundary designs impose. The current implementation providesthree pillars: truthful read semantics across thermal states, an explicit fixed-floor boundarypolicy for standalone honesty, and a shared-family mode in which related bitmaps preservealigned morphology under coordinated maintenance. Against Roaring Chambi et al. [2016],the strongest results appear in family-level operations: aligned AND is 7.4x faster and alignedAND-cardinality is 19.3x faster at 100K. Standalone point lookups and bulk construction remainslower than Roaring, and fixed-floor behavior currently outperforms adaptive morphology onstandalone paths. The paper therefore positions Diatom as a research primitive for coordinatedbitmap families, not as a blanket replacement for existing bitmap infrastructure. TemporalStrata, Summary, and Symbiosis are optional experimental adaptations layered on that core.