Persistent Homology of Architectural Spatial Configurations: Topological Invariants Beyond Graph Metrics
Аннотация
This paper introduces persistent homology as a complementary framework for the topological analysis of architectural floor plans. While Space Syntax relies on graph-based metrics to quantify spatial accessibility, it remains insensitive to higher-order topological features including circulation loops. Applying cubical persistent homology with distance-transform filtration to five synthetic typologies (Palladian, Corbusian, Wrightian, corridor, and courtyard) and validating against Villa Rotonda and Villa Savoye, this study demonstrates that plans with nearly identical Space Syntax integration values exhibit fundamentally different first-homology signatures. The Wasserstein-2 distance between persistence diagrams provides a stable metric for typological classification, capturing loop structures invisible to graph methods. A correlation analysis over an augmented sample confirms that the strength of the dominant loop (H 1 maximum persistence) is statistically independent of integration, while resolution tests, morphological-robustness tests, and comparison with simpler topological baselines establish the stability and specific advantage of the proposed framework. Persistent homology and Space Syntax together provide a more complete characterization of architectural spatial configurations.