1: \begin{abstract}

2: For the Poisson problem in two dimensions, posed on a domain partitioned into

3: axis-aligned rectangles with up to one hanging node per edge, we envision an

4: efficient error reduction step in an instance-optimal $hp$-adaptive finite

5: element method. Central to this is the problem: Which increase in local polynomial degree

6: ensures $p$-robust contraction of the error in energy norm? We reduce this problem to a small

7: number of saturation problems on the reference square, and provide strong numerical

8: evidence for their solution.

9: 

10:   ~~

11: 

12:   \smallskip

13:   \noindent\textsc{\subjclassname.}

14:     65N12, %Stability and convergence of numerical methods

15:     65N30, %Finite elements, Rayleigh-Ritz and Galerkin methods, finite methods

16:     65N50. %Mesh generation and refinement

17: 

18:   \smallskip

19:   \noindent\textsc{\keywordsname.}

20:     Poisson equation, $p$-robustness, $hp$-adaptive finite element method, convergence, 1-irregular quadrilateral meshes.

21: \end{abstract}

22: