Residual adaptation with a uniform-refinement safeguard (source code)

= Residual adaptation with a uniform-refinement safeguard

A reliable residual estimator certifies an error tolerance only when its proved upper bound is below that tolerance. Reliability alone does not prove that a particular refinement strategy will eventually pass the stopping test. Occasional uniform bisection provides a simple safeguard in a setting where the estimator vanishes as the maximum cell length tends to zero.

For example, consider one-dimensional $-(pu')'+qu=f$, with $p\in W^{1,\infty}$, $p\geq p_*>0$, bounded $q\geq q_*>0$ and $f\in L^2$, using continuous piecewise-linear <finite elements> and homogeneous compatible boundary data. Discrete <energy> stability bounds $u_h$ in $H^1$, so the cell residual $R_h=f+p'u_h'-qu_h$ is bounded in <L2 norm> uniformly over refinements. The <explicit one-dimensional dual residual bound in L2> gives
$$
\|u-u_h\|_2\leq C\eta_h,\qquad
\eta_h=\left(\sum_i h_i^4\|R_h\|_{2,I_i}^2\right)^{1/2}
\leq h_{\max}^2\|R_h\|_2.
$$
Here $C$ depends on coefficient bounds, not on the mesh. If the refinement schedule forces $h_{\max}\to0$, then $C\eta_h\to0$ and every positive tolerance is eventually certified. This is a termination argument for the stated residual/regularity setting; it does not claim optimal complexity for an arbitrary adaptive marking rule.