1 Foundations and Definitions
1.1 Graded modules and shifts
In commutative algebra, a finitely generated graded module over a standard graded polynomial ring \[ S=k[x_0,\dots,x_n],\quad \deg x_i=1, \] decomposes as \(M=\bigoplus_{d\in \mathbb Z} M_d\). A shift by an integer \(t\), written \(M(t)\), reindexes degrees via \(M(t)_d=M_{d+t}\). Shifts are the basic mechanism by which regularity translates between algebraic data (degrees in a resolution) and geometric data (twisting a sheaf by a line bundle).
1.2 Coherent sheaves on projective space
Let \(\mathbb P^n=\operatorname{Proj}(S)\). Graded \(S\)-modules modulo torsion correspond to coherent sheaves on \(\mathbb P^n\). Concretely, a graded module \(M\) determines a sheaf \(\widetilde M\), and twisting by \(\mathcal O_{\mathbb P^n}(1)\) corresponds to shifting the grading: \[ \widetilde{M(t)}\cong \widetilde M \otimes \mathcal O_{\mathbb P^n}(t). \] This dictionary makes it possible to define regularity either for graded modules or for coherent sheaves.
1.3 The cohomological definition of regularity
For a coherent sheaf \(\mathcal F\) on \(\mathbb P^n\), one defines its Castelnuovo–Mumford regularity as the smallest integer \(m\) such that certain higher cohomology groups vanish after twisting: \[ H^i(\mathbb P^n,\mathcal F(m-i))=0\quad \text{for all } i>0. \] This integer is denoted \(\operatorname{reg}(\mathcal F)\).
1.3.1 Vanishing criteria via twists
The vanishing condition above captures a stabilization phenomenon: once the twists are sufficiently positive, higher cohomology disappears in a predictable range. The choice of \(m\) is calibrated so that the twist by \(\mathcal O_{\mathbb P^n}(m-i)\) simultaneously handles all cohomological degrees \(i\).
1.3.2 Relation to global sections and higher cohomology
When \(\mathcal F\) is \(m\)-regular, global generation and surjectivity properties become accessible. For instance, for an \(m\)-regular sheaf, twisting by \(\mathcal O_{\mathbb P^n}(1)\) improves regularity behavior, and higher cohomology vanishing controls how sections lift across hyperplane restrictions. Thus regularity links the complexity of \(\mathcal F\) to concrete section-theoretic data.
1.4 The resolution-theoretic viewpoint
Regularity can also be read from minimal free resolutions of the associated graded module. Let \(M\) be a finitely generated graded \(S\)-module and consider a minimal graded free resolution: \[ \cdots \to \bigoplus_j S(-a_{i,j}) \to \cdots \to \bigoplus_j S(-a_{0,j}) \to M \to 0. \] The integers \(a_{i,j}\) record degrees of generators appearing in homological degree \(i\).
1.4.1 Minimal free resolutions and degree shifts
In a minimal resolution, the shifts \(a_{i,j}\) measure how far the module is from being generated by low-degree forms. Larger shifts indicate higher “complexity” in the syzygies. Regularity aggregates these shifts across all homological levels with a correction by homological degree.
1.4.2 Betti numbers and regularity bounds
Let \(\beta_{i,d}\) denote graded Betti numbers: \[ \beta_{i,d}=\dim_k \operatorname{Tor}^S_i(M,k)_d. \] A standard characterization states that \(\operatorname{reg}(M)\) is the smallest integer \(m\) such that \[ \beta_{i,d}(M)=0\quad \text{whenever } d>m+i. \] Equivalently, \(m\) bounds the degrees where \(\operatorname{Tor}\) can occur in each homological degree. This formulation makes regularity a numerical invariant derived from the graded structure of syzygies.
1.5 Equivalence of definitions
The two perspectives—cohomological vanishing and resolution degree bounds—agree for coherent sheaves on projective space and for graded modules.
1.5.1 Translating between module and sheaf language
Under the correspondence between \(M\) and \(\widetilde M\), the cohomology vanishing conditions for \(\widetilde M\) translate into constraints on \(\operatorname{Tor}\) and the twists in a minimal resolution. The resulting integer is consistent across the language switch, allowing one to use whichever viewpoint is more convenient for a given task.
2 Basic Properties
2.1 Invariance under isomorphism
Regularity depends only on the isomorphism class of the object. If \(\mathcal F \cong \mathcal G\), then \(\operatorname{reg}(\mathcal F)=\operatorname{reg}(\mathcal G)\). Similarly, isomorphic graded modules have the same regularity.
2.2 Behavior under twisting
Twisting by \(\mathcal O_{\mathbb P^n}(t)\) shifts the regularity by \(t\): \[ \operatorname{reg}(\mathcal F(t))=\operatorname{reg}(\mathcal F)+t. \] This rule is immediate from the definition, since twisting by \(\mathcal O(1)\) modifies the cohomology groups in a controlled manner.
2.2.1 Regularity of line bundles
For the structure sheaf, one gets \(\operatorname{reg}(\mathcal O_{\mathbb P^n})=0\). More generally, for line bundles \(\mathcal O_{\mathbb P^n}(t)\), regularity equals \(t\). These examples calibrate the normalization of the invariant.
2.3 Monotonicity and exact sequences
Regularity behaves predictably with respect to morphisms and exact sequences, offering a mechanism for inductive control.
2.3.1 Regularity in short exact sequences
Given a short exact sequence \[ 0\to \mathcal A\to \mathcal B\to \mathcal C\to 0, \] regularity of the middle term can be bounded in terms of the other two. While sharp forms vary by convention, the general principle is that vanishing of cohomology for two terms forces vanishing patterns for the third, with possible shifts by small integers.
2.3.2 Consequences for kernels, cokernels, and images
Because kernels and cokernels appear naturally in exact sequences, one can propagate regularity bounds through constructions such as taking images of maps of sheaves. In practice, these inequalities make it possible to study complicated sheaves by decomposing them into simpler components.
2.4 Subadditivity and tensor products
Regularity interacts well with tensor products. If \(\mathcal F\) and \(\mathcal G\) are coherent sheaves, then \(\operatorname{reg}(\mathcal F\otimes \mathcal G)\) is bounded above by a function of \(\operatorname{reg}(\mathcal F)\) and \(\operatorname{reg}(\mathcal G)\). This “subadditive” behavior reflects the way syzygies combine under multiplication and the way cohomological vanishing patterns multiply under tensoring.
2.5 Direct sums and filtrations
Direct sums satisfy \[ \operatorname{reg}(\mathcal F\oplus \mathcal G)=\max\{\operatorname{reg}(\mathcal F),\operatorname{reg}(\mathcal G)\}. \] Filtrations similarly let one bound regularity from data on successive quotients, since vanishing and degree constraints pass through the steps of a filtration.
3 Computation and Examples
3.1 Regularity of standard objects
Some canonical sheaves and modules have easily described regularity.
3.1.1 Free modules and shifts
For a free graded module \(S(-t)\), the minimal resolution is trivial, and regularity equals \(t\). For a finite direct sum of such shifts, regularity is the maximum shift appearing among summands.
3.1.2 Structure sheaves of linear subspaces
If \(L\subset \mathbb P^n\) is a linear subspace of dimension \(r\), its ideal is generated by linear forms. The associated quotient has a predictable linear resolution, and the regularity corresponds to the expected value for a scheme cut out by linear equations. Such examples illustrate how geometric linearity translates into low regularity.
3.2 Hypersurfaces and complete intersections
Regularity is especially tractable for subschemes defined by homogeneous equations with controlled degrees.
3.2.1 Regularity for principal ideals
For a hypersurface defined by a single homogeneous polynomial of degree \(d\), the ideal is generated in degree \(d\) with a short resolution. The resulting regularity can be expressed directly in terms of \(d\), giving an explicit measure of how the degree of the defining equation controls cohomology vanishing.
3.2.2 Regularity bounds for complete intersections
For a complete intersection cut out by forms of degrees \(d_1,\dots,d_c\), the resolution has a structured form resembling a Koszul complex. Regularity becomes computable from the \(d_i\)’s through the degrees appearing in that resolution. This yields effective numerical bounds and explains why complete intersections often have “tame” syzygy behavior.
3.3 Monomial ideals and combinatorial methods
Monomial ideals admit descriptions through combinatorics, and regularity can be studied using graded pieces indexed by monomials.
3.3.1 Squarefree monomial ideals
For squarefree monomial ideals, there is a well-developed correspondence with simplicial complexes. Regularity can be read from combinatorial invariants related to reduced homology and induced subcomplexes, providing a bridge between algebraic syzygies and topological properties of complexes.
3.3.2 Using simplicial complexes (overview)
At a high level, the graded local cohomology of \(S/I\) for squarefree \(I\) decomposes according to multidegrees supported on faces of a simplicial complex. One then extracts a bound—or sometimes an exact value—for regularity from the homological degrees where these multigraded pieces are nonzero.
3.4 Small-dimensional examples
Working in small dimensions helps illustrate the mechanisms behind regularity.
3.4.1 Curves in projective space
For subschemes of \(\mathbb P^n\) of dimension one, regularity connects to the degrees where line bundles on the curve become globally generated and to the growth of syzygies along the embedding. In many cases, bounds can be expressed using degree and arithmetic invariants of the curve.
3.4.2 Zero-dimensional schemes
For finite subschemes (zero-dimensional schemes), regularity measures when their coordinate ring stabilizes in a strong sense. Cohomology vanishing for twists translates into statements about how many degrees are needed for the ideal to become saturated and for the Hilbert function to align with the Hilbert polynomial (which is constant in this setting).
4 Geometric Interpretation
4.1 Projective embedding and generation of ideals
Regularity provides effective criteria for when an ideal sheaf is generated by forms of bounded degree. If an ideal sheaf \(\mathcal I_Z\) is \(m\)-regular, then one can often conclude that the scheme \(Z\) is determined by equations of degree \(\le m\) under appropriate hypotheses. Thus regularity organizes the relationship between the embedding of a subscheme and the degrees of defining equations.
4.2 Castelnuovo-type bounds
Named bounds in algebraic geometry estimate the degrees and cohomological behavior of subschemes. Regularity can be viewed as a unifying framework for such estimates, since it packages vanishing and generation into a single integer that often scales similarly to classical Castelnuovo results.
4.3 Regularity and syzygy constraints
Syzygies describe relations among generators, relations among relations, and so on. Regularity bounds the degrees where these syzygies can occur, preventing “unexpected” high-degree jumps. This constraint can be interpreted geometrically as limitations on how a projective embedding fails to be cut out by equations of low degree.
4.4 Hyperplane restriction and stabilization
A key geometric operation is restricting a sheaf to a hyperplane. General hyperplanes often simplify cohomology and create inductive patterns.
4.4.1 Inductive strategies with general hyperplanes
| By comparing \(\mathcal F\) on \(\mathbb P^n\) with its restriction to a general hyperplane \(H\cong \mathbb P^{n-1}\), one can propagate vanishing results downward in dimension. Regularity inequalities from short exact sequences involving \(\mathcal F(-1)\), \(\mathcal F\), and \(\mathcal F | _H\) support this induction. |
|---|
4.5 Relation to Hilbert functions and Hilbert polynomials
Cohomology vanishing for twists implies stabilization properties of the Hilbert function. For sufficiently large degrees, the Hilbert function equals the Hilbert polynomial. Regularity supplies an explicit bound on when this equality begins, thereby connecting cohomological behavior to numerical invariants of graded rings.
5 Functorial Behavior and Operations
5.1 Restriction to linear subspaces
When restricting a coherent sheaf to a linear subspace, regularity typically does not increase dramatically. Under favorable hypotheses (such as using general linear subspaces and applying standard exact sequences), one obtains bounds that reflect how vanishing patterns descend to smaller projective spaces.
5.2 Tensoring and products of sheaves
Tensor operations correspond to products at the level of graded modules. Regularity bounds show that the tensor product’s cohomological complexity is controlled by the regularities of the factors. This is useful when constructing sheaves from known ones, such as those arising from multiplication maps of graded algebras.
5.3 Pullbacks and pushforwards (settings where defined)
For morphisms between projective schemes where pushforward and pullback preserve coherence, regularity can be studied via derived functors and projection formulas. Exact behavior depends on the morphism and on how twisting interacts with relative ampleness, but in many standard settings one obtains inequalities or asymptotic statements.
5.4 Symmetric powers and Rees-type constructions (overview)
Symmetric powers arise from operations on vector bundles and graded modules. Rees-type constructions encode blow-ups and filtrations; regularity can be used to control when associated graded objects acquire predictable cohomology. In these contexts, one often studies how regularity behaves as exponents grow.
5.5 Powers of ideals
Ideals define subschemes and their powers correspond to thickened structures. Regularity provides a way to measure how complicated the defining equations of these thickenings become.
5.5.1 Asymptotic regularity (overview)
For many classes of ideals, regularity of \(I^p\) grows at most linearly with \(p\) for large \(p\). Asymptotic results describe eventual stabilization of patterns in syzygies, giving quantitative control over families of subschemes defined by ideal powers.
6 Applications
6.1 Vanishing theorems and cohomology control
Since regularity is defined by cohomology vanishing of twists, it supplies direct vanishing statements. These are applied to prove that certain maps between global sections are surjective and that obstructions in higher cohomology disappear after twisting sufficiently far.
6.2 Bounds for generators and relations
Regularity translates into constraints on the degrees needed to generate an ideal and the degrees where relations appear in a minimal free resolution. This has practical consequences: it limits the search space in computational problems and guides theoretical constructions in projective geometry.
6.3 Effective results in projective geometry
Many existence and reconstruction statements in projective geometry become effective when accompanied by a numerical invariant. Regularity can quantify when linear systems associated to a sheaf behave well, when embeddings satisfy property-like criteria, and when projection and restriction techniques succeed.
6.4 Interactions with depth and depth-sensitive invariants
Regularity is often compared with other homological invariants such as depth and projective dimension. While each invariant captures different information, inequalities between them help interpret how the graded structure of a module reflects geometric complexity, including the depth of the corresponding sheaf or the singularity behavior of subschemes.
6.5 Implications for schemes defined by ideals
For a scheme \(Z\) with ideal sheaf \(\mathcal I_Z\), regularity governs the complexity of the coordinate ring and the scheme’s embedding. In particular, it provides guidance on when \(Z\) becomes scheme-theoretically determined by bounded degree data, and it supports inductive arguments that reduce problems to hyperplane sections and smaller-dimensional cases.
7 Advanced Topics (Optional Survey)
7.1 Regularity over different rings and base changes (overview)
When changing the base ring or performing a base change, regularity can behave subtly. Over flat families or under suitable conditions, one can study how regularity varies with parameters and whether bounds persist. Often, results relate regularity to cohomological upper semicontinuity and to the stability of graded Betti numbers in families.
7.2 Local cohomology characterization
Regularity can be characterized using local cohomology modules with respect to the irrelevant ideal. This viewpoint refines the cohomological vanishing definition by encoding the degrees where local cohomology becomes nonzero, thereby capturing regularity in terms of graded support and depth-like phenomena.
7.3 Boij–Söderberg perspective (high-level)
In the study of Betti tables, Boij–Söderberg theory describes cones generated by pure resolutions. Regularity corresponds to linear constraints on Betti tables, so it can be interpreted as bounding the region of possible Betti data. The perspective is useful for understanding which degree distributions are compatible with a given regularity.
7.4 Computational aspects and algorithms (high-level)
Computing regularity may involve constructing resolutions or analyzing cohomology. In monomial cases, combinatorial algorithms based on simplicial complexes can be used. For general ideals, one typically uses bounds, Gröbner basis techniques, or syzygy computations in computer algebra systems to estimate or determine regularity.
7.5 Comparisons with other invariants (e.g., depth, projective dimension)
Regularity is compared with invariants measuring the size of resolutions or the vanishing of certain Ext groups. While projective dimension records how long a resolution lasts, regularity records how degrees shift through that resolution. Together, they provide complementary information about the graded and homological structure of the module.
8 Notation, Conventions, and References
8.1 Common grading conventions
A standard convention sets \(\deg x_i=1\) for the polynomial ring and uses the twist \( \mathcal F(t)=\mathcal F\otimes \mathcal O_{\mathbb P^n}(t)\) for sheaves. Regularity depends on this normalization, so consistent conventions are important when comparing statements across sources.
8.2 Typical assumptions on the base field and projective space
Many results assume \(k\) is a field and \(\mathbb P^n\) is projective space over \(k\). Smoothness and characteristic conditions are sometimes relevant for stronger theorems, though the basic definition and many core properties of regularity are characteristic-free under standard hypotheses.
8.3 Key references and standard sources
Foundational treatments appear in classic texts on syzygies and projective geometry, where Castelnuovo–Mumford regularity is developed alongside cohomological criteria, examples, and applications to embeddings and generators. Surveys and advanced chapters often connect regularity to local cohomology, Betti tables, and modern computational approaches.