Eisenbud–Goto regularity conjecture

About 16 years old · traced to

Let X⊆PrX\subseteq\mathbb P^r be a nondegenerate projective variety. Its Castelnuovo–Mumford regularity is denoted by reg⁡X\operatorname{reg} X, its degree by deg⁡X\deg X, and its codimension by codim⁡X\operatorname{codim} X. Eisenbud–Goto regularity conjecture. Then

reg⁡X≤deg⁡X−codim⁡X+1.\operatorname{reg} X\leq\deg X-\operatorname{codim} X+1.

This conjecture proposes a bound for the homological complexity of a projective variety in terms of its degree and codimension. It is refuted: counterexamples were found by McCullough and Peeva.

References

Primary source

Junho Choe, “Castelnuovo-Mumford regularity of unprojections and the Eisenbud-Goto regularity conjecture”, arXiv:2206.06151 (2024).

Additional references

6 papers in this index state this conjecture (2010–2022). The statement above is taken from the most recent of them; the others are arXiv:2106.12667, arXiv:1406.7404, arXiv:1402.1906, arXiv:1110.2124, arXiv:1012.5329.

Progress summary

Refreshed
Claimed solved

The conjecture is false: McCullough and Peeva constructed varieties whose regularity is far larger than the proposed degree bound, while important special cases remain valid.

Proposed in 1984, the conjecture asserts the bound reg⁡X≤deg⁡X−codim⁡X+1\operatorname{reg}X\leq\deg X-\operatorname{codim}X+1 for every nondegenerate projective variety XX. McCullough and Peeva’s constructions refute this general statement.

Known results

  • The bound holds for projective curves.
  • It holds for arithmetically Cohen–Macaulay varieties.
  • It holds for projective toric varieties of codimension 22.
  • It is known for smooth surfaces and smooth threefolds in P5\mathbb{P}^{5}.

2018 counterexamples and June 2022 quantitative refinements

McCullough and Peeva produced nondegenerate prime ideals whose regularity is not bounded by any polynomial in degree; one example has deg⁡X=31\deg X=31 and reg⁡X=38\operatorname{reg}X=38. A later construction gives, for fixed dimension n≥3n\geq3 and codimension e≥2e\geq2, families with reg⁡X=Ω((deg⁡X)k)\operatorname{reg}X=\Omega((\deg X)^k), where k=⌊(n+1)/2⌋k=\lfloor(n+1)/2\rfloor.

Current status (as of September 2026): The general conjecture is settled negatively by the reported McCullough–Peeva counterexamples, while the sharp boundary of valid special cases remains open.

Sources

Solutions 0

No solutions have been posted yet.