Dean's conjecture on cycles divisible by the minimum-degree bound

Less than 1 year old · traced to

All graphs under consideration are finite and simple. For a graph GG and a vertex vv, let dG(v)d_G(v) denote the degree of vv. Dean's conjecture. For every integer k≥3k \ge 3, every graph with minimum degree at least kk contains a cycle of length divisible by kk. The conjecture is known to be true for all k≠5k\neq 5, so the case k=5k=5 remains open.

References

Primary source

Ilkyoo Choi, Hojin Chu, Ringi Kim and Boram Park, “Existence of cycles of length divisible by 3 or 4”, arXiv:2605.02731 (2026).

Additional references

2 papers in this index state this conjecture (2026). The statement above is taken from the most recent of them; the others are arXiv:2601.13552.

Progress summary

Refreshed
Claimed progress

A reader-submitted proof claims to settle the remaining five-dimensional case, but no independent verification has been found.

Dean’s conjecture, attributed to Nathaniel Dean in 1988, says that minimum degree at least kk forces a cycle whose length is divisible by kk. All cases except k=5k=5 are recorded as settled; the submission concerns precisely this remaining case.

Known results

  • k=3k=3: proved by Chen and Saito.
  • k=4k=4: proved by Dean, Lesniak, and Saito.
  • k≥6k\geq 6: Liu, Ma, and Zhao proved a stronger modular-cycle theorem in 2026.
  • The literature explicitly leaves k=5k=5 for separate arguments.

Community submission (unverified), August 31, 2026

A submitted proof claims that a counterexample with minimum degree at least 55 can be reduced to a zero-free 55-weak graph, then eliminated through bipartite, triangle-containing, and nonbipartite triangle-free cases. The supplied text is only a summary, so the claimed proof has not been independently checked.

Current status (as of August 2026): Cases k=3,4k=3,4 and k≥6k\geq 6 are settled, while the k=5k=5 case has only an unverified community-submitted proof claim.

Sources

Solutions 1

ProofThis solution needs a summarySee full solutionHide full solution

You can find the full proof (v1.0.1 as of Aug 31) here: https://zenodo.org/records/22182448

A brief summary of the proof is as follows:

Suppose for contradiction that GG is a finite simple graph with δ(G)≥5\delta(G)\ge 5 but contains no cycle whose length is divisible by 55. Call a cycle CC with

∣C∣≡0(mod5)|C|\equiv 0\pmod 5

a zero-cycle, and call a graph with no zero-cycle zero-free.

The proof first reduces an arbitrary zero-free counterexample to a much more rigid class of graphs called 55-weak graphs. Starting from GG, one passes to a 22-connected subgraph JJ in which all but possibly one vertex still have degree at least 55. The structure of the 22-cuts of such a zero-free graph is then analyzed. This forces JJ to have one of two forms:

  • Type I: JJ is 33-connected and all but possibly one vertex have degree at least 55.
  • Type II: JJ has a unique degree-22 vertex θ\theta with neighbors u,vu,v, and replacing the path uθvu\theta v by the edge uvuv produces a 33-connected graph of minimum degree at least 55.

Thus, if a counterexample exists, there is a zero-free 55-weak graph JJ.

The rest of the proof establishes that no such graph can exist. Every 55-weak graph falls into one of the three exhaustive cases

bipartitecontains a trianglenonbipartite and triangle-free.\boxed{\text{bipartite}} \qquad \boxed{\text{contains a triangle}} \qquad \boxed{\text{nonbipartite and triangle-free}}.

1. Bipartite case

Suppose JJ is bipartite and zero-free. Since all cycles are even, the goal is to force an even cycle whose length is 0(mod5)0\pmod 5.

The argument chooses a maximal structured subgraph built from 44-cycles, called a tetragonal core, and studies how the rest of the graph can attach to it. Degree and connectivity conditions force many exterior attachments, while zero-freeness strongly restricts the possible lengths of paths through those attachments.

The possible tetragonal cores are eventually reduced to three small parameter cases. Each of these configurations is ruled out by a detailed structural analysis, so a zero-free bipartite 55-weak graph cannot exist. Hence every bipartite 55-weak graph contains a zero-cycle.

2. Graphs containing a triangle

Now suppose JJ contains a triangle. Starting from a triangle, the proof grows a maximal trigonal core by repeatedly adding vertices along boundary edges.

A key feature of such a core is that a trigonal graph on tt vertices contains cycles of every length

3,4,…,t.3,4,\ldots,t.

Consequently, a sufficiently large trigonal core would automatically contain five consecutive cycle lengths, one of which must be divisible by 55. Therefore, in a zero-free graph, a maximal trigonal core must be very small.

The remaining possibilities are a triangle itself or a 44-vertex core, abstractly of the form K4−eK_4-e or K4K_4. The high-degree and connectivity assumptions force enough paths outside the core to create several consecutive cycle lengths. Again, one of these lengths is 0(mod5)0\pmod 5, contradicting zero-freeness.

Thus every 55-weak graph containing a triangle contains a zero-cycle.

3. Nonbipartite, triangle-free case

This is the most involved case. Choose a shortest odd cycle OO. Because OO is shortest, it is induced, and triangle-freeness severely restricts how vertices outside OO may attach to it.

The proof studies the components outside OO, their attachment points on OO, and the collection of paths they create between vertices of OO. Combining these exterior paths with the two possible routes around OO produces cycles of many different residues modulo 55. The degree and 33-connectivity conditions guarantee that there are enough such attachments to force a zero-cycle.

The argument is carried out separately for the two forms of 55-weak graphs:

  • In Type I, the 33-connected structure around a shortest odd cycle directly forces the required collection of cycle lengths.
  • In Type II, write H=J−θH=J-\theta. If HH is nonbipartite, the shortest-odd-cycle analysis applies again. If HH is bipartite, a separate structural argument produces a simple uu--vv path PP with
∣P∣≡3(mod5).|P|\equiv 3\pmod 5.

Adding the path uθvu\theta v, which has length 22, gives a cycle of length

∣P∣+2≡3+2≡0(mod5).|P|+2\equiv 3+2\equiv0\pmod5.

Hence every nonbipartite triangle-free 55-weak graph also contains a zero-cycle.

Combining the three cases proves that every 55-weak graph contains a zero-cycle. But the initial reduction says that any zero-free graph GG with δ(G)≥5\delta(G)\ge5 would contain a zero-free 55-weak graph JJ. This is a contradiction.

Therefore every finite simple graph GG satisfying

δ(G)≥5\delta(G)\ge5

contains a cycle CC such that

∣C∣≡0(mod5).|C|\equiv0\pmod5.

NOTE: This proof was developed with substantial use of LLMs, chiefly GPT 5.6 Sol, with Opus 5 and GLM 5.3 Flash as ancillary support. This proof has undergone multiple full-length adversarial audits, but has not yet been reviewed by an independent mathematician. This proof was authored by Elias Botsford.