Computational Complexity

This section is the developer-facing reference for PROTEA’s computational complexity. It covers the Big-O profile of every pipeline stage, measured hot paths, and practical advice for profiling.

The corresponding academic narrative (derivations, empirical scaling plots, and the ablation study that justifies the numpy + FAISS design) lives in thesis Chapter 5 (“Computational Analysis”), which is outside the Sphinx tree. The two documents are complementary: this reference tells you where the cost lives and how to measure it; the thesis explains why the current design was chosen over the alternatives.

Division of labour

This Sphinx section	Thesis Ch. 5
Function signatures + source anchors	Formal Big-O derivations
Measured wall-clock numbers (EXP.13)	Empirical scaling plots
Profiling recipes (scalene, etc.)	Architecture motivation
Forward-ref to PERF.1 flamegraphs	Comparison to pgvector / GPU KNN

• Pipeline Overview: Big-O per Stage
• PLM Attention Complexity
• KNN Search Complexity
• LightGBM Complexity
• Anc2Vec Complexity
• Export Pipeline Complexity
• Measuring Performance