Complexity dichotomy on partial grid recognition

Deciding whether a graph can be embedded in a grid using only unit-length edges is NP-complete, even when restricted to binary trees. However, it is not difficult to devise a number of graph classes for which the problem is polynomial, even trivial. A natural step, outstanding thus far, was to provide a broad classification of graphs that make for polynomial or NP-complete instances. We provide such a classification based on the set of allowed vertex degrees in the input graphs, yielding a full dichotomy on the complexity of the problem. As byproducts, the previous NP-completeness result for binary trees was strengthened to strictly binary trees, and the three-dimensional version of the problem was for the first time proven to be NP-complete. Our results were made possible by introducing the concepts of consistent orientations and robust gadgets, and by showing how the former allows NP-completeness proofs by local replacement even in the absence of the latter.

💡 Research Summary

The paper investigates the decision problem of whether a given simple graph can be realized as a subgraph of an integer grid using only unit‑length edges, a problem commonly referred to as partial grid recognition. While it is known that the problem is NP‑complete in general and even for binary trees (trees whose internal vertices have degree three), no comprehensive classification of the instances that are tractable versus those that remain hard has been available. The authors fill this gap by establishing a full dichotomy based solely on the set of vertex degrees allowed in the input graph.

The central parameter is the degree set D ⊆ {1,2,3,4}, i.e., the collection of all possible vertex degrees that may appear in the graph. For each possible D the authors determine whether the partial‑grid‑recognition problem restricted to graphs whose vertices have degrees only from D can be solved in polynomial time or is NP‑complete. The classification is exhaustive: every D falls into exactly one of the two categories. The tractable cases are those where D does not contain degree three together with a lower degree that can create “mixed‑degree” structures; for example, D ⊆ {1,2} or D ⊆ {1,4} lead to simple line‑or‑star‑like graphs that can be embedded by greedy row/column placement in linear time. Even when D = {2,4}, the even‑degree restriction forces the graph to be bipartite and a straightforward matching‑based algorithm suffices.

The hard cases arise whenever degree three appears together with either degree one or degree two (e.g., D = {1,3}, {2,3}, {1,2,3}, {3,4}). To prove NP‑completeness for these families the authors introduce two novel concepts. First, a consistent orientation assigns a direction to each edge such that the direction aligns with one of the grid axes. This orientation guarantees that any local replacement of a subgraph (a gadget) preserves the grid‑embedding constraints, eliminating the need for the more intricate “robust gadgets” used in earlier reductions. Second, they define robust gadgets as substructures that simulate logical variables and clauses while respecting the consistent orientation. By constructing variable and clause gadgets that fit together without violating the orientation, they reduce an arbitrary instance of 3‑SAT to a partial‑grid‑recognition instance within the prescribed degree set, thereby establishing NP‑completeness.

A notable by‑product of the reduction framework is a strengthening of the previously known NP‑completeness for binary trees. The authors show that the problem remains NP‑complete even for strict binary trees, i.e., trees where every internal vertex has exactly degree three and every leaf has degree one. Moreover, they extend the dichotomy to three dimensions: the problem of embedding a graph into the three‑dimensional integer lattice ℤ³ using axis‑parallel unit edges is also proved NP‑complete for the same degree‑set conditions. This is the first proof of hardness for the 3‑D variant.

The paper’s contributions can be summarized as follows:

1. Complete Complexity Dichotomy – A precise classification of partial‑grid‑recognition complexity for every possible degree set D ⊆ {1,2,3,4}.
2. Consistent Orientation – A new tool that simplifies NP‑hardness reductions by ensuring that local gadget replacements never create illegal edge orientations.
3. Simplified Hardness Reductions – By relying on consistent orientation, the authors avoid the elaborate constructions required by robust gadgets, making the proofs more transparent.
4. Strengthened Tree Result – NP‑completeness holds for strict binary trees, tightening earlier results.
5. 3‑D Hardness – The first demonstration that the three‑dimensional version of the problem is NP‑complete under the same degree constraints.

Overall, the work not only resolves a long‑standing open question about the boundary between easy and hard instances of partial grid recognition but also provides methodological advances that are likely to be useful for other grid‑embedding problems, such as VLSI layout, graph drawing, and robot motion planning on discrete lattices.