L2hforadaptivity Ef F1 F3 F5 [repack] Jun 2026

Let Fi_score = w1 F1_norm + w3 F3_norm + w5*F5_norm (w1+w3+w5=1). Policy mapping:

h(x) = g(W * (x + ϵ))

For Poisson’s equation with a sharp interior layer (e.g., f5‑dominant region), pure L² adaptivity refines too late, while H¹‑only refines the entire layer uniformly. The combination: l2hforadaptivity ef f1 f3 f5