Second Vanishing Theorem

• Second Vanishing Theorem is a local cohomology result that characterizes when the next-to-top local cohomology H_I^(d-1)(R) vanishes based on dimensionality and connectedness of the punctured spectrum.
• It employs techniques such as the Huneke–Lyubeznik graph and Frobenius-theoretic methods to relate geometric connectedness with vanishing properties in both prime and mixed characteristic settings.
• Extensions include combinatorial and quadratic analogues, with applications in Stanley–Reisner rings and Chow–Witt intersection theory, underscoring its broad relevance in commutative algebra and algebraic geometry.

Second Vanishing Theorem is a label used for several vanishing phenomena. In commutative algebra, the standard usage concerns local cohomology: for a regular local ring, it characterizes vanishing of the next-to-top local cohomology, or equivalently the bound $\operatorname{cd}(A,I)\le \dim A-2$, by combining a dimension condition with connectedness of the punctured spectrum after passage to $A^\dagger$. The same phrase also appears by analogy in quadratic intersection theory, where a supported Chow–Witt intersection product is shown to vanish, and in combinatorial-topological analogues for Stanley–Reisner rings (Scheffelin, 25 Jun 2026, Feld, 2021, Bhattacharyya, 2021).

1. Classical local-cohomological form

For a regular local ring $(A,\mathfrak m)$ of dimension $d$, the classical formulation adopted in recent work is the equivalence

$\operatorname{cd}(A,I)\le \dim A-2$

if and only if

$$\dim A/I \ge 2, \qquad \operatorname{Spec}(A^\dagger/IA^\dagger)\setminus\{\mathfrak m\} \text{ is connected},$$

where

$A^\dagger:=\widehat{(\widehat A)^{\mathrm{sh}}}.$

In the parallel formulation used for a $d$-dimensional noetherian local ring $(R,\mathfrak m)$, the Second Vanishing Theorem holds for $R$ if, for each ideal $A^\dagger$0, the following are equivalent: $A^\dagger$1 for all $A^\dagger$2 and all $A^\dagger$3-modules $A^\dagger$4; and $A^\dagger$5 together with connectedness of the punctured spectrum of $A^\dagger$6, where $A^\dagger$7 is the completion of the strict henselization of the completion of $A^\dagger$8 (Scheffelin, 25 Jun 2026, Zhang, 2021).

For regular local rings, the universal quantifier over all modules can be reduced to the ring itself: $A^\dagger$9 for all $(A,\mathfrak m)$0 and all $(A,\mathfrak m)$1-modules $(A,\mathfrak m)$2 if and only if $(A,\mathfrak m)$3 for all $(A,\mathfrak m)$4. In the $(A,\mathfrak m)$5-dimensional setting, the relevant vanishing is therefore

$(A,\mathfrak m)$6

Since $(A,\mathfrak m)$7 is already controlled by Hartshorne–Lichtenbaum vanishing in the relevant range, the substantial content of the theorem is the vanishing of $(A,\mathfrak m)$8 (Zhang, 2021).

This places the theorem immediately after Hartshorne–Lichtenbaum in the hierarchy of top-degree vanishing statements. The “first” vanishing theorem governs the top local cohomology; the “second” governs the next degree and replaces a purely dimensional condition by connectedness of the punctured spectrum (Zhang, 2021, Bhattacharyya, 2020).

2. Connectedness, graphs, and proof mechanisms

A recurrent feature of the theorem is that the geometric hypothesis is encoded by connectedness of the punctured spectrum. In the complete local setting, the paper on Eisenstein extensions formulates this through the Huneke–Lyubeznik graph $(A,\mathfrak m)$9: if $d$0 are the minimal primes of a complete local ring $d$1, then $d$2 has vertices $d$3, with an edge $d$4 whenever $d$5 is not maximal-primary, and

$d$6

This graph-theoretic criterion is the standard device for comparing connected components across extensions and quotients (Bhattacharyya, 2020).

The easy direction of the theorem is the implication from vanishing to connectedness. If the punctured spectrum is disconnected, one decomposes the defining ideal up to radical as an intersection $d$7 with $d$8 maximal-primary; the Mayer–Vietoris sequence then forces nonvanishing of the next-to-top local cohomology. This argument appears in both the classical mixed-characteristic treatments and the later ramified work (Bhattacharyya, 2020, Zhang, 2021).

The difficult direction is the converse. One major reduction is to the case of prime ideals $d$9 with $\operatorname{cd}(A,I)\le \dim A-2$0. Zhang proves a reduction theorem showing that, in a complete regular local ring with separably closed residue field, vanishing for all connected $\operatorname{cd}(A,I)\le \dim A-2$1 and $\operatorname{cd}(A,I)\le \dim A-2$2 follows once one knows vanishing for height-$\operatorname{cd}(A,I)\le \dim A-2$3 prime ideals (Zhang, 2021).

In characteristic $\operatorname{cd}(A,I)\le \dim A-2$4, this codimension-two case acquires a Frobenius-theoretic interpretation. For a regular local ring $\operatorname{cd}(A,I)\le \dim A-2$5 of characteristic $\operatorname{cd}(A,I)\le \dim A-2$6, with $\operatorname{cd}(A,I)\le \dim A-2$7, Lyubeznik’s criterion states

$\operatorname{cd}(A,I)\le \dim A-2$8

Zhang combines this with finite-length properties of $\operatorname{cd}(A,I)\le \dim A-2$9 for $$\dim A/I \ge 2, \qquad \operatorname{Spec}(A^\dagger/IA^\dagger)\setminus\{\mathfrak m\} \text{ is connected},$$0-dimensional local domains and Hartshorne–Speiser stable-part arguments to give a new proof of the prime-characteristic theorem (Zhang, 2021).

3. Mixed characteristic: unramified, partial ramified, and full ramified forms

The mixed-characteristic story is stratified by ramification. For complete unramified regular local rings of mixed characteristic with separably closed residue field, Zhang proves the full theorem: for every ideal $$\dim A/I \ge 2, \qquad \operatorname{Spec}(A^\dagger/IA^\dagger)\setminus\{\mathfrak m\} \text{ is connected},$$1,

$$\dim A/I \ge 2, \qquad \operatorname{Spec}(A^\dagger/IA^\dagger)\setminus\{\mathfrak m\} \text{ is connected},$$2

if and only if

$$\dim A/I \ge 2, \qquad \operatorname{Spec}(A^\dagger/IA^\dagger)\setminus\{\mathfrak m\} \text{ is connected},$$3

This settles Hartshorne’s problem in the unramified mixed-characteristic class and also extends the topological characterization of the highest Lyubeznik number to mixed characteristic (Zhang, 2021).

A first ramified extension was obtained only for extended ideals through Eisenstein extensions. If $$\dim A/I \ge 2, \qquad \operatorname{Spec}(A^\dagger/IA^\dagger)\setminus\{\mathfrak m\} \text{ is connected},$$4 is a $$\dim A/I \ge 2, \qquad \operatorname{Spec}(A^\dagger/IA^\dagger)\setminus\{\mathfrak m\} \text{ is connected},$$5-dimensional complete ramified regular local ring of mixed characteristic, $$\dim A/I \ge 2, \qquad \operatorname{Spec}(A^\dagger/IA^\dagger)\setminus\{\mathfrak m\} \text{ is connected},$$6 is an Eisenstein extension of a complete unramified regular local ring $$\dim A/I \ge 2, \qquad \operatorname{Spec}(A^\dagger/IA^\dagger)\setminus\{\mathfrak m\} \text{ is connected},$$7, and $$\dim A/I \ge 2, \qquad \operatorname{Spec}(A^\dagger/IA^\dagger)\setminus\{\mathfrak m\} \text{ is connected},$$8 is extended from $$\dim A/I \ge 2, \qquad \operatorname{Spec}(A^\dagger/IA^\dagger)\setminus\{\mathfrak m\} \text{ is connected},$$9, then for $A^\dagger:=\widehat{(\widehat A)^{\mathrm{sh}}}.$0,

$A^\dagger:=\widehat{(\widehat A)^{\mathrm{sh}}}.$1

Under the additional assumption that $A^\dagger:=\widehat{(\widehat A)^{\mathrm{sh}}}.$2 is normal for every minimal prime $A^\dagger:=\widehat{(\widehat A)^{\mathrm{sh}}}.$3 of $A^\dagger:=\widehat{(\widehat A)^{\mathrm{sh}}}.$4, the punctured spectra of $A^\dagger:=\widehat{(\widehat A)^{\mathrm{sh}}}.$5 and $A^\dagger:=\widehat{(\widehat A)^{\mathrm{sh}}}.$6 have the same number of connected components (Bhattacharyya, 2020).

A different partial ramified theorem was later proved by using surjectivity of multiplication on local cohomology. In a $A^\dagger:=\widehat{(\widehat A)^{\mathrm{sh}}}.$7-dimensional complete regular local ring of mixed characteristic with separably closed residue field, if $A^\dagger:=\widehat{(\widehat A)^{\mathrm{sh}}}.$8 is a proper ideal such that every $A^\dagger:=\widehat{(\widehat A)^{\mathrm{sh}}}.$9 satisfies $d$0, together with additional finite-length hypotheses on local cohomology over $d$1, then

$d$2

The key technical input is a characteristic-$d$3 theorem showing that, under a finite-length hypothesis, $d$4 is divisible, hence multiplication by a nonzero element is surjective on $d$5 (Asgharzadeh et al., 2021).

The remaining mixed-characteristic case was then completed for ramified regular local rings. For a regular local ring $d$6 of ramified mixed characteristic and an ideal $d$7, the theorem now holds in full: $d$8 if and only if

$d$9

The proof uses a geometric formal-schemes reformulation, reductions to a normal surface singularity, and a resolution embedded in projective space; two decisive ingredients are vanishing of

$(R,\mathfrak m)$0

on the exceptional fiber, and a ramified mixed-characteristic global-generation input due to Gabber (Scheffelin, 25 Jun 2026).

One application of the full ramified theorem is a connected-components formula. If $(R,\mathfrak m)$1 is an $(R,\mathfrak m)$2-dimensional regular local ring with separably closed residue field and the punctured spectrum of $(R,\mathfrak m)$3 has $(R,\mathfrak m)$4 connected components, then

$(R,\mathfrak m)$5

This identifies the next-to-top local cohomology with the failure of punctured-spectrum connectedness (Scheffelin, 25 Jun 2026).

4. Stanley–Reisner analogues and topological interpretation

The phrase also has a combinatorial-topological analogue for Stanley–Reisner rings. Let $(R,\mathfrak m)$6 be simplicial complexes, let

$(R,\mathfrak m)$7

and write $(R,\mathfrak m)$8. The analogue studied in this setting concerns $(R,\mathfrak m)$9, but it is not a full if-and-only-if connectedness criterion. Instead, the paper proves a necessary condition and a stronger sufficient condition (Bhattacharyya, 2021).

The necessary condition is: $R$0 The sufficient condition requires more:

1. every $R$1-face of $R$2 contains a vertex of $R$3;
2. every $R$4-face of $R$5 is a $R$6-face of $R$7;
3. every $R$8-facet of $R$9 contains one vertex of $A^\dagger$00.

Under these three hypotheses,

$A^\dagger$01

An explicit example shows that the necessary condition alone is not sufficient, so this is only a partial analogue of the classical theorem (Bhattacharyya, 2021).

The topological interpretation comes from the Reiner–Welker–Yanagawa multigraded formula. For $A^\dagger$02, with

$A^\dagger$03

one has

$A^\dagger$04

In the proof of the necessary direction, a $A^\dagger$05-face $A^\dagger$06 disjoint from $A^\dagger$07 produces

$A^\dagger$08

forcing nonvanishing. In the sufficient direction, barycentric subdivision and elementary collapses remove all $A^\dagger$09- and $A^\dagger$10-faces relevant to the degree-$A^\dagger$11 relative cohomology group. The “second vanishing” language therefore survives, but the controlling invariant is no longer punctured-spectrum connectedness; it is relative simplicial topology (Bhattacharyya, 2021).

5. Quadratic and intersection-theoretic analogue

In quadratic intersection theory, the expression is used analogically rather than as a theorem about local cohomology. The starting point is Serre’s classical vanishing theorem for intersection multiplicities: if $A^\dagger$12 is a regular local ring of dimension $A^\dagger$13, $A^\dagger$14 are finite $A^\dagger$15-modules with $A^\dagger$16 of finite length, and

$A^\dagger$17

then

$A^\dagger$18

The Chow–Witt analogue does not define a Grothendieck–Witt-valued Tor formula. Instead, it proves vanishing of the actual supported intersection product class (Feld, 2021).

Let $A^\dagger$19 be a Milnor–Witt cycle algebra over a fixed perfect field $A^\dagger$20, let $A^\dagger$21 be a regular local scheme of dimension $A^\dagger$22 over $A^\dagger$23 with closed point $A^\dagger$24, let $A^\dagger$25 be a virtual vector bundle, and let $A^\dagger$26 be closed subsets such that

$A^\dagger$27

Then for all integers $A^\dagger$28, the supported intersection product

$A^\dagger$29

is the zero map. In particular, for every $A^\dagger$30 and $A^\dagger$31,

$A^\dagger$32

This is stronger than vanishing of a numerical invariant, because the entire quadratic/cohomological class vanishes (Feld, 2021).

The theorem is local, assumes a perfect base field, allows coefficients in an arbitrary Milnor–Witt cycle algebra, and includes Chow–Witt groups as the case where $A^\dagger$33 is Milnor–Witt $A^\dagger$34-theory. Through the equivalence

$A^\dagger$35

it also applies to coefficients arising from homotopy modules or ring spectra. The paper presents the result explicitly as a support-theoretic analogue of Serre’s “second vanishing theorem” language, but also stresses that it is not a full quadratic reformulation of Serre’s numerical Tor-length theorem (Feld, 2021).

6. Terminological ambiguity in adjacent literatures

Outside local cohomology and the Serre-style intersection context, the phrase is frequently ambiguous. In several vanishing-theorem papers there is no result explicitly titled “Second Vanishing Theorem,” and the expression can only refer to a second principal statement or to an interpretive analogy.

In “A vanishing theorem” (Werner et al., 2012), there is no theorem explicitly titled that way; the best match is Theorem $A^\dagger$36, a hook-Schur reformulation equivalent to the main theorem on the vanishing of

$A^\dagger$37

In Fujino’s “Vanishing theorems” (Fujino, 2012), the most plausible match is Theorem $A^\dagger$38(ii), the vanishing half of the main torsion-free/vanishing package for simple normal crossing pairs. In Keum’s paper on fake projective planes (Keum, 2014), there is likewise no formal first/second nomenclature, but Theorem $A^\dagger$39 is the second principal vanishing statement. Wu’s note on Bogomolov vanishing explicitly says that the theorem proved there is not called “Second Vanishing Theorem”; it is a Bogomolov-type vanishing for pseudoeffective line bundles on compact Kähler manifolds (Wu, 2020). The 2025 paper on surfaces also states that it does not use the phrase anywhere, even though it introduces a new formulation of a surface vanishing theorem derived from relative Kawamata–Viehweg vanishing (Fujino et al., 24 Nov 2025).

Accordingly, “Second Vanishing Theorem” is best treated as a term with a stable local-cohomological meaning and a broader analogical afterlife. In local cohomology it names a precise connectedness criterion for the vanishing of $A^\dagger$40; in other areas it may designate a second principal vanishing statement, or an analogue of the same vanishing philosophy, without carrying a universally fixed formal definition (Zhang, 2021, Scheffelin, 25 Jun 2026).