On generic 3-rigidity of graphs

We give a necessary condition of generic 3 -rigidity of graphs relying on partitioning the edges into 3 subsets; such that each subset-pair gives a generically 2-rigid graph, either by themselves or after an appropriate edge-deletion. Notably, as pointed out by Dewar and Gallet, the condition is still not sufficient.

💡 Research Summary

The paper investigates the combinatorial structure underlying generic three‑dimensional rigidity of graphs. After recalling the standard bar‑joint framework model, the authors define a graph G(V,E) to be generically d‑rigid if a placement of its vertices in ℝ^d in generic position yields a rigid framework; minimal d‑rigidity means the graph has exactly d|V|–(d+1 choose 2) edges and no non‑trivial self‑stress. While a complete combinatorial characterisation exists for d = 1 and d = 2 (Laman’s theorem and the Lovász‑Yemini theorem), the case d ≥ 3 remains open.

Motivated by the Lovász‑Yemini theorem, which states that a minimally 2‑rigid graph can be decomposed into two edge‑disjoint spanning trees after duplicating any edge, the authors seek a three‑dimensional analogue. They propose a necessary condition: for every edge e of a minimally 3‑rigid graph, the edge set E can be partitioned into three subsets S₁, S₂, S₃ with |S_i| = |V| − i such that the unions S₁∪S₂, S₁∪S₃/e, and S₂∪S₃/e are each minimally 2‑rigid (the notation “/e” denotes contraction of e).

The technical development proceeds through a detailed analysis of rigidity matrices. For a graph G, M(G,d) is the d‑dimensional rigidity matrix; after deleting the first column of each coordinate block the authors obtain a square matrix N(G,d). Full rank of N(G,d) is equivalent to minimal d‑rigidity. By expanding det N(G,3) along appropriate rows and columns, they isolate two invertible blocks: one corresponding to a spanning tree F and the other to the rigidity matrix of the graph obtained from G by removing F and contracting a chosen edge e. This yields Lemma 2 and Lemma 3, which together guarantee the existence of the required partitions for any edge e.

Lemma 2 shows that given a spanning tree F and an edge e∈F, the vertex set V\F can be split into two parts R₁ and R₂ such that the graphs (V,F∪R₁) and (V,F∪R₂)/e are minimally 2‑rigid. Lemma 3 proves the converse: for any edge e there exists a spanning tree F containing e such that the graph (V,E\F)/e is minimally 2‑rigid. The main result, Theorem 2, combines these lemmas to establish the three‑subset decomposition condition for every edge of a minimally 3‑rigid graph.

Importantly, the authors acknowledge that this condition is not sufficient. Sean Dewar (2026) and Matteo Gallet independently produced counter‑examples: graphs that satisfy the three‑subset decomposition yet fail to be 3‑rigid. Consequently, the condition remains a necessary but not characterising criterion.

The paper also discusses self‑stress analysis. By selecting a spanning tree and expressing the equilibrium equations in terms of a signed incidence matrix B, the authors parametrize internal forces using a set of free variables associated with the non‑tree edges. Directional constraints are encoded via diagonal matrices Dᵢⱼ, and redundancies are eliminated by fixing index choices. Lemma 1 establishes that in a given coordinate plane, if all directional constraints hold except for one edge, the missing constraint is automatically satisfied, a fact used repeatedly in the proofs.

Overall, the work contributes a rigorous linear‑algebraic framework for probing 3‑dimensional rigidity, clarifies the relationship between 3‑rigidity and collections of 2‑rigid substructures, and delineates the limits of a natural combinatorial condition. While the proposed condition does not resolve the long‑standing open problem of a full combinatorial characterisation for d ≥ 3, it provides a valuable necessary test and deepens understanding of the structural complexity inherent in three‑dimensional rigidity.