The Realizable Extension Problem and the Weighted Graph $(K_{3,3},l)$

This note outlines the realizable extension problem for weighted graphs and provides results of a detailed analysis of this problem for the weighted graph $(K_{3,3},l)$. This analysis is then utilized to provide a result relating to the connectedness of the moduli space of planar realizations of $(K_{3,3},l)$. The note culminates with two examples which show that in general, realizability and connectedness results relating to the moduli spaces of weighted cycles which are contained in a larger weighted graph cannot be extended to similar results regarding the moduli space of the larger weighted graph.

💡 Research Summary

The paper introduces and studies a new problem in the theory of weighted graphs called the “Realizable Extension Problem.” A weighted graph (G,l) consists of a simple graph G together with a weight function l that assigns a positive real number to each edge. A “realization” of (G,l) is an embedding of the vertices as points in the Euclidean plane such that the Euclidean distance between any two adjacent vertices equals the prescribed edge weight. While the realizability of simple structures such as trees, cycles, and complete graphs has been completely characterized by classical distance‑inequality conditions (triangle inequality, ladder inequalities, etc.), much less is known for more complex, non‑planar graphs.

The Realizable Extension Problem asks: given a subgraph H of G together with a realizable weight assignment l_H on H, can one extend l_H to a weight function l on the whole graph G so that (G,l) remains realizable? In other words, the problem seeks a consistent extension of a distance matrix defined on a subset of vertices to a full distance matrix that satisfies all geometric constraints imposed by the additional edges. This formulation is equivalent to extending a system of linear and nonlinear distance constraints while preserving feasibility.

To illustrate the problem, the authors focus on the complete bipartite graph K₃,₃, a canonical non‑planar graph consisting of two vertex sets A={a₁,a₂,a₃} and B={b₁,b₂,b₃} with every possible edge between the two sets (nine edges in total). They begin by selecting a spanning tree T of K₃,₃ (five edges) and assume an arbitrary realizable weight assignment l_T on T. Since any tree can be realized in the plane, the central difficulty lies in adding the remaining four edges without violating geometric constraints.

For each new edge the authors systematically check the triangle inequalities and, more importantly, a family of “ladder inequalities.” A ladder inequality arises when two distinct paths connect the same pair of vertices; the sum of the edge lengths along each path must be equal, otherwise the two paths would prescribe incompatible distances for the same pair of points. In the context of K₃,₃ this yields conditions such as

 l(a_i,b_j) = l(a_i,a_k) + l(a_k,b_j) = l(a_i,b_ℓ) + l(b_ℓ,b_j)

for any distinct indices i,k,ℓ,j. These equalities force the weight function to define a consistent metric on the six vertices. The authors prove that if all ladder and triangle inequalities are satisfied, then a planar realization of the full weighted K₃,₃ exists.

Having established a concrete set of sufficient conditions, the paper turns to the topology of the associated moduli space M(G,l). The moduli space is defined as the set of all planar realizations of (G,l) modulo Euclidean motions (translations, rotations, reflections). For the weighted K₃,₃ that meets the extension criteria, the authors show that M(K₃,₃,l) is not connected: it consists of exactly two connected components. These components correspond to the two possible “handedness” (clockwise vs. counter‑clockwise ordering) of the vertex configuration. No continuous deformation within the space of realizations can change the handedness without violating one of the distance constraints, so the components are topologically isolated.

The paper then presents two counter‑examples that demonstrate the limits of extending realizability and connectivity results from subgraphs to the whole graph. The first example involves a 4‑cycle C₄ with a weight assignment that is realizable and whose moduli space is connected. When C₄ is embedded as a subgraph of K₃,₃ and the same edge weights are retained, the ladder inequalities for the additional edges become unsatisfiable, leading either to non‑realizability or to a disconnected moduli space for the larger graph. The second example uses a spanning tree (which always yields a connected moduli space) and shows that adding the missing edges of K₃,₃ introduces non‑linear constraints that split the moduli space into multiple components. These examples underline that properties verified on cycles or trees cannot be naively extrapolated to more complex graphs.

In conclusion, the authors provide a rigorous definition of the Realizable Extension Problem, develop a complete solution for the weighted K₃,₃, and analyze the resulting moduli space’s topology. Their work bridges distance geometry, graph theory, and topological moduli theory, offering a framework that can be applied to other non‑planar graphs and higher‑dimensional realization problems. Future research directions suggested include algorithmic approaches for testing realizability of extensions, classification of moduli‑space connectivity for broader families of graphs, and exploration of analogous problems in three‑dimensional Euclidean space or other metric settings.