Graph Diffusion Residuals for Control-Function Instrumental Variables
The paper proposes Adaptive Anisotropic Instrumental Heat Flow (A-IHF), a deterministic graph-diffusion residual extractor for control-function instrumental variable estimators. A-IHF treats treatment as a signal on a feature graph, uses pilot diffusion to detect treatment jumps, attenuates conductance across those jumps, and computes control via a sparse graph resolvent. The observational selection rule combines graph generalized cross-validation, roughness, residualized-treatment relevance, and graph-admissibility filtering without outcome data. The analysis decomposes error into structural leakage, residual attenuation, and residualized treatment variation, yielding finite‑sample bounds and graph‑admissibility rates under latent piecewise‑smooth geometry. Across 54 synthetic benchmark cells, guarded observational A‑IHF achieves the lowest average structural‑response MSE and the A‑IHF family beats the best non‑A‑IHF baseline in 32 cells, excelling when the graph captures piecewise‑smooth first‑stage structure.