Alec Kriebel

NP-completeness of the initial odd-ear prefix

Read the paper (PDF, 6 pages) · Complete source and exact verification archive (ZIP) · Zenodo record · CC BY 4.0

Plain-English explanation · Read on this page · Collaboration


Original abstract

For a finite 2-vertex-connected graph G, let β(G) be the largest number of consecutive odd ears at the beginning of an open ear decomposition, counting its initial circuit and its one-edge ears. We prove that deciding β(G)≥ t, with t part of the input, is NP-complete even for simple graphs with a universal vertex whose deletion leaves a connected bipartite graph. This answers the variable-threshold complexity question posed by Sebő and by Benchetrit–Sebő. The reduction is from explicitly encoded Set Cover. It uses the classical factor-critical formulation of β, a cone identity, and a block-loss inequality that permits arbitrary partial blocks. In particular, it does not assume an optimal subgraph is a union of whole blocks.

The idea in plain English

A graph can be built by starting with a cycle and then adding paths whose endpoints are already present and whose internal vertices are new. This is an open ear decomposition. The question here asks how many odd-length ears can occur consecutively at the start, before the first even-length ear.

The initial cycle and paths consisting of a single edge both count. The paper proves that deciding whether this initial odd prefix reaches a requested threshold is NP-complete, even for a restricted family of simple graphs formed by adding a vertex adjacent to every vertex of a connected bipartite graph.

NP-completeness means that a proposed positive answer has a polynomially checkable certificate and that the problem captures the difficulty of all problems in NP through polynomial reductions. An exact polynomial-time algorithm would imply P=NP. The threshold is supplied as part of the input.

The precise result

Let G be a finite loopless undirected 2-vertex-connected graph with at least three vertices; explicitly listed parallel edges are allowed. An open ear decomposition partitions all edges into an initial circuit and subsequent paths with distinct old endpoints and new internal vertices. One-edge ears are allowed, and oddness means an odd number of edges.

Let β(G) be the largest number of consecutive odd ears at the beginning of such a decomposition, counting the initial circuit and one-edge ears. The first even ear ends the prefix. If no odd initial circuit exists, β(G)=0.

Given G and an integer t in binary, deciding whether β(G)≥t is NP-complete. Hardness holds even when G is simple and has a specified universal vertex z such that G−z is connected and bipartite. Every nonpositive threshold is immediate on valid graph inputs; invalid graph inputs can be rejected in polynomial time.

The proof gives a polynomial reduction from explicitly encoded Set Cover and a polynomial certificate for positive instances. It applies to arbitrary input sizes and arbitrary partial selections of the construction blocks.

Why the result holds

The proof uses the classical Lovász characterization of 2-connected factor-critical graphs by all-odd open ear decompositions, in the precise form stated by Benchetrit–Sebő. A graph is factor-critical when deleting any vertex leaves a perfect matching. Each ear contributes one to cycle rank, so β can be expressed by the ranks of suitable factor-critical subgraphs.

For the cone z*B over a connected bipartite graph B with at least two vertices, this becomes β(z*B)=max |E(B[S])| over nonempty S for which B[S] is connected and has a perfect matching. The score counts all induced edges, so both connectedness and the choice of objective matter.

The Set Cover construction uses one complete bipartite block Kₖ,ₖ for the root and each universe element, together with a connector for each set. With T=Σⱼ(1+|Aⱼ|), k=T+n+3, M=(m+1)k² and threshold F=M−ks, a minimal cover supplies distinct private elements and an explicit perfect matching with enough induced edges.

The converse permits arbitrary partial blocks. If aᵢ and bᵢ count omitted vertices on the two sides, the lost internal edges satisfy dᵢ−k(aᵢ−bᵢ)=bᵢ(2k−aᵢ)≥0. Matching balance and the strict threshold gaps force the connector budget; connectedness forces coverage. The argument does not require a selection to be a union of whole blocks.

A positive-instance certificate consists of a suitable subgraph and a perfect matching after each of its vertex deletions. All required conditions can be checked in polynomial time. The accompanying exact finite diagnostics supplement the written universal proof.

Scope and limitations

The threshold varies with the input. The result does not establish hardness for any fixed threshold and is consistent with the earlier fixed-parameter result of Benchetrit–Sebő. It does not assert P≠NP.

The parameter concerns the initial consecutive odd open ears, with the initial circuit and one-edge ears counted. Closed ears and the total number of odd ears throughout a decomposition are different parameters.

The stated restricted hardness class consists of simple universal-vertex cones over connected bipartite bases. This manuscript does not establish bounded-degree, high-girth or approximation results.

The factor-critical characterization and Set Cover NP-completeness are classical inputs. The exact Lovász statement is used through Benchetrit–Sebő; the original 1972 proof was not obtained. The comparison with Weighted Connected Matchings uses the 2022 arXiv v1 body; its final 2025 journal body was not obtained.

Alper Ferudun’s related preprint gives a different route to the same general classification, using 3-SAT, edge multiplicities and odd subdivisions, with different sparse and high-girth graph restrictions. It is credited as related work; no certification of all its external claims or support is implied.

The observed Zenodo creation dates are 22 October 2026 for this record and 23 October 2026 for Ferudun’s related record. These two bibliographic facts do not establish worldwide firstness or the relationship between the works. Historical priority and present worldwide openness are not certified.

Status and AI use

Unrefereed preprint, version 1.0.0. Developed with extensive AI assistance; no external human peer review or formal proof certification.

AI tools were used extensively in mathematical research, proof development and reconstruction, literature work, computation, verification and writing. This is an unrefereed preprint without external human peer review, journal acceptance or formal proof-assistant certification. Alec Kriebel is the named human author and is responsible for the work. Finite diagnostics supplement the written proof.

Manuscript and declared publication date: 22 October 2026; version 1.0.0.

Read the paper

Read the complete 6-page deposited paper in the on-page PDF reader.

Open PDF in a full window · Download PDF · Download the original from Zenodo

Supporting materials

The archive includes the identical paper PDF, editable TeX, exact diagnostic code, results and original license notices. Authored prose is CC BY 4.0; authored code is MIT.

Collaborate on this result

Independent proof checks, counterexamples and precise comparisons with prior work are welcome. Developing a journal submission would require substantive research contributions. Authorship and acknowledgments would reflect actual contributions and mutual agreement; citations imply no participation or endorsement.

Authorship and responsibilities would be discussed in relation to the substantive contributions made.

Contact Alec

Cite this paper

Alec Kriebel. (2026). NP-completeness of the initial odd-ear prefix (Version 1.0.0) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.23576280

Download BibTeX

Background and prior work