diff --git a/agglomeration_poisson/README.md b/agglomeration_poisson/README.md
index 39c0fbd9..4878ebab 100644
--- a/agglomeration_poisson/README.md
+++ b/agglomeration_poisson/README.md
@@ -82,7 +82,7 @@ The mesh skeleton is defined by
@f}
The mesh skeleton $\Gamma$ is decomposed into $(d-1)$–dimensional simplices $F$ representing the mesh faces, shared by at most two elements. These are distinct from elemental interfaces, which are defined as the simply connected components of the intersection between the boundary of an element and either a neighboring element or $\partial \Omega$. As such, an interface between two elements may consist of more than one face, separated by hanging nodes/edges shared by those two elements only. We denote by $\Gamma_{\mathrm{int}}$ the union of all interior faces, and by $\Gamma_{\mathrm D} := \Gamma \cap \partial\Omega$ the union of Dirichlet boundary faces.
-In practice, the polytopic mesh is obtained by agglomeration, so that each element $K \in T_h$ is the union of a collection of leaf cells. For each agglomerated element $K$, we associate an axis-aligned bounding box $B_K$. On $B_K$ we define the standard polynomial space $Q_p(B_K)$ spanned by tensor-product Lagrange polynomials of degree $p$ in each coordinate direction. Since $K \subset B_K$, the basis on $K$ is defined by restricting each basis function to $K$. In the implementation, this corresponds to using the deal.II finite element `FE_DGQ` on the bounding box and taking its restriction to the agglomerated element. The global discrete space $V_h$ is then obtained by assembling these local spaces in a discontinuous manner over all $K \in T_h$. For $u_h, v_h \in V_h$ we use the broken gradient $\nabla_h$ and the standard jump and average operators $[\![\cdot]\!]$ and $\{\!\!\{\cdot\}\!\!\}$ on faces.
+In practice, the polytopic mesh is obtained by agglomeration, so that each element $K \in T_h$ is the union of a collection of leaf cells. For each agglomerated element $K$, we associate an axis-aligned bounding box $B_K$. On $B_K$ we define the standard polynomial space $Q_p(B_K)$ spanned by tensor-product Lagrange polynomials of degree $p$ in each coordinate direction. Since $K \subset B_K$, the basis on $K$ is defined by restricting each basis function to $K$. In the implementation, this corresponds to using the deal.II finite element `FE_DGQ` on the bounding box and taking its restriction to the agglomerated element. The global discrete space $V_h$ is then obtained by assembling these local spaces in a discontinuous manner over all $K \in T_h$. For $u_h, v_h \in V_h$ we use the broken gradient $\nabla_h$ and the standard jump and average operators $[\![\cdot]\!]$ and $\{\!\!\{\cdot\}\!\!\}$ on faces.
The DG formulation reads: find $u_h \in V_h$ such that
@f{align*}{
@@ -190,14 +190,7 @@ with their enclosing MBRs, while the right image shows the associated tree
structure (leaf and internal nodes). This visual example helps explain how the
hierarchical grouping is later used to extract agglomerates.
-
-
-
-
-
(1) Examples of MBRs holding geometric data and their MBRs
-
(2) Corresponding R-tree data structure elements
-
-
-
-
-
-
(3) Original 8x8 mesh
-
(4) Bounding boxes and mesh elements
-
-
-
-
(5) Tree hierarchy: root node, internal nodes, and leaf nodes from top to bottom (only two child subtrees of the root are shown for clarity)
-
-
-
-
-
-
(6) Input mesh
-
(7) Fine level mesh obtained by five global refinements of the input mesh
-
Comparison of agglomeration strategies
-
-
-
-
-
(8) METIS, 91 agglomerates
-
(9) METIS, 364 agglomerates
-
-
-
-
-
-
(10) R-tree, 91 agglomerates
-
(11) R-tree, 364 agglomerates
-
-
-
-
-
(12) h-convergence for Q_p elements (p=1,2,3) with METIS and R-tree agglomeration
-
-
-
-
(13) Convergence under p-refinement for METIS and R-tree agglomeration (p = 1,2,3,4,5)
-
-
-
-
(14) Wall-clock time (seconds) for building polytopal grids with R-tree and METIS
-
-
-
-
-
(15) Numerical solution using the R-tree strategy
-
(16) Numerical solution using the METIS strategy
-