Signal A Generative Framework for the Creation of Multi-Attribute Geographically-Explicit Synthetic Populations

Summary

Generating multi-attribute synthetic populations with realistic joint distributions and geographic variation is a foundational requirement for geo-simulation techniques such as micro-simulation and agent-based modeling, which are widely used in urban planning, public health and social policy analysis. Existing methods, however, have struggled to reconstruct region-specific joint distributions from aggregated census or survey data alone. This paper proposes a hierarchical diffusion-based generative framework to produce geographically explicit synthetic populations that better preserve realistic attribute correlations across regions. The hierarchical structure of the diffusion model is intended to capture finer regional detail than prior approaches. It is framed as a technical and methodological contribution to computational social science infrastructure rather than a study of any specific policy or population.