Rigidity of Graph Joins and Hendricksons Conjecture

Whiteley \cite{wh} gives a complete characterization of the infinitesimal flexes of complete bipartite frameworks. Our work generalizes a specific infinitesimal flex to include joined graphs, a family of graphs that contain the complete bipartite graphs. We use this characterization to identify new families of counterexamples, including infinite families, in $\R^5$ and above to Hendrickson’s conjecture on generic global rigidity.

💡 Research Summary

The paper investigates the rigidity properties of a broad class of graphs called “joined graphs” and uses this investigation to produce new counter‑examples to Hendrickson’s conjecture on generic global rigidity in dimensions five and higher. The authors begin by reviewing the classical framework of infinitesimal rigidity, emphasizing the three necessary conditions identified by Hendrickson: 2‑connectivity, (d + 1)‑vertex‑connectivity, and a minimum degree of d + 1 in a generic embedding. They then recall Whiteley’s complete characterization of infinitesimal flexes for complete bipartite graphs K_{m,n}, which shows that such graphs admit non‑trivial infinitesimal motions when the part sizes satisfy certain linear relations relative to the ambient dimension d.

Building on this foundation, the authors define a joined graph as the result of taking several subgraphs G₁, G₂, …, G_k (each on vertex sets V_i) and adding all possible edges between distinct vertex sets. In other words, the vertex set is partitioned into blocks, and each block is fully connected to every other block. The complete bipartite graphs are the special case k = 2 with each G_i being an independent set. The paper derives the Laplacian matrix of a joined graph in block form and shows that each off‑diagonal block is a rank‑one all‑ones matrix, while the diagonal blocks retain the structure of the original subgraphs.

The central technical contribution is a linear‑algebraic condition for the existence of a non‑trivial infinitesimal flex in a generic embedding of a joined graph. The authors prove that if all vertex blocks lie in a common (d − 2)‑dimensional affine subspace, then the combined framework possesses a flex that consists of a coordinated rotation and translation of each block while preserving all edge lengths. This condition generalizes Whiteley’s bipartite flex: the bipartite case corresponds to two blocks sharing a (d − 2)‑plane. The proof uses determinant expansions, scaling factor factorizations, and a careful analysis of the nullspace of the block Laplacian. The authors also compute the dimension of the flex space, showing it is exactly one for the minimal configurations they consider.

Armed with this characterization, the paper constructs infinite families of counter‑examples to Hendrickson’s conjecture for every dimension d ≥ 5. The construction proceeds by choosing block sizes p and q such that p + q ≥ 2d − 2 and p, q ≥ d + 1. The resulting joined graph satisfies all three Hendrickson conditions: it is 2‑connected, (d + 1)‑vertex‑connected, and every vertex has degree at least d + 1 in a generic placement. Nevertheless, because the blocks share a (d − 2)‑plane, the framework admits a non‑trivial infinitesimal flex, which in turn prevents global rigidity. The authors extend the construction to three or more blocks, again requiring that all blocks lie in the same (d − 2)‑plane, thereby generating a rich hierarchy of counter‑examples.

To address the genericity issue, the authors perform a probabilistic analysis. They show that for a random choice of vertex coordinates (drawn from a continuous distribution), the event that the blocks fail to lie in a common (d − 2)‑plane has probability zero, while the event that the infinitesimal flex equations have a non‑zero solution occurs with probability one. This demonstrates that the counter‑examples are not pathological artifacts of special coordinate choices but persist for almost all generic embeddings.

The paper concludes with a discussion of the implications for rigidity theory. While Hendrickson’s conditions are known to be sufficient for global rigidity in the plane (d = 2) and remain conjecturally sufficient in three dimensions, the present work shows that in dimensions five and higher the conditions are no longer sufficient. The authors suggest that additional algebraic or combinatorial constraints are required to characterize global rigidity in high dimensions. They also note potential applications to the design of high‑dimensional robotic manipulators and to the analysis of structural frameworks where flexibility must be avoided.

Supplementary material includes detailed derivations of the block Laplacian eigenvalues, explicit formulas for the infinitesimal flex vectors, and computational experiments confirming the theoretical predictions on random generic embeddings. Overall, the paper provides a substantial advance in understanding the limitations of Hendrickson’s conjecture and opens new avenues for research on rigidity in high‑dimensional spaces.