04.01.02 · algebraic-geometry / sheaves

Stalk of a sheaf

shipped3 tiersLean: partial

Anchor (Master): Hartshorne §II; Bredon *Sheaf Theory*; Iversen *Cohomology of Sheaves*; Godement *Topologie algébrique et théorie des faisceaux*

Intuition [Beginner]

The stalk of a sheaf $F$ at a point $x$ is the data of $F$ "infinitely close to $x$ " — what you see if you zoom into $x$ ever more tightly. Concretely: stalk $F_{x}$ records all sections of $F$ defined on some open neighbourhood of $x$ , identified when they agree on a smaller neighbourhood. Each element of $F_{x}$ is called a germ at $x$ .

The germ of a function captures local behaviour at $x$ — the function's value, its rate of change, its higher derivatives, all packaged together. Two functions have the same germ at $x$ if they agree on some open neighbourhood (no matter how small) of $x$ . Stalks separate the local from the global: a section of $F$ over an open set $U$ has a germ at every point of $U$ , and the section is determined by its collection of germs (this is the locality axiom of a sheaf).

For algebraic varieties, the stalk of the structure sheaf $O_{X}$ at a point $x$ is the local ring $O_{X, x}$ — the ring of rational functions defined near $x$ , with the maximal ideal recording functions that vanish at $x$ . The local rings of $X$ encode the geometry of $X$ at every point: smoothness, singularity, dimension, multiplicity. Stalks are how sheaves carry pointwise information.

Visual [Beginner]

A point $x$ in a topological space, with a chain of nested open neighbourhoods shrinking towards $x$ ; sections of the sheaf on these neighbourhoods are identified in the colimit to form the stalk at $x$ .

Worked example [Beginner]

Consider the sheaf of continuous real-valued functions $C^{0}$ on $R$ . The stalk at the origin $C_{0}^{0}$ consists of germs of continuous functions at $0$ : pairs $(U, f)$ with $f$ continuous on an open neighbourhood $U$ of $0$ , modulo the equivalence $(U, f) \sim (V, g)$ iff $f = g$ on a smaller open neighbourhood of $0$ .

Two examples:

The germ of $f (x) = x$ at $0$ .
The germ of $g (x) = sin (x)$ at $0$ .

These are different germs at $0$ — the functions agree at $x = 0$ (both equal 0) but differ on every open neighbourhood (since $sin (x) \neq = x$ for $x \neq = 0$ small). So the germs are distinct elements of $C_{0}^{0}$ .

In contrast, $f (x) = x$ and $h (x) = x \cdot χ_{(- ϵ, ϵ)} (x)$ for any $ϵ > 0$ (where $χ$ is a continuous bump function equal to 1 near $0$ ) define the same germ at $0$ : they agree on $(- ϵ /2, ϵ /2)$ , so represent the same equivalence class in $C_{0}^{0}$ .

For the structure sheaf $O_{X}$ of a variety $X = A_{k}^{1} = Spec k [t]$ , the stalk at the origin is $O_{X, 0} = k [t]_{(t)}$ — the localisation of $k [t]$ at the prime ideal $(t)$ . This is the ring of rational functions $f / g$ with $g (0) \neq = 0$ . Its maximal ideal $m = (t) k [t]_{(t)}$ records the rational functions vanishing at $0$ . The quotient $O_{X, 0} / m = k$ is the residue field at $0$ .

Check your understanding [Beginner]

Formal definition [Intermediate+]

Let $X$ be a topological space, $F$ a presheaf (of sets, abelian groups, rings, modules, ...) on $X$ , and $x \in X$ . The stalk of $F$ at $x$ is the directed colimit

F_{x} := U ∋ x lim F (U),

where the colimit is taken over the directed system of open neighbourhoods $U$ of $x$ , ordered by reverse inclusion ( $U \leq V$ iff $V \subseteq U$ ), with restriction maps $F (U) \to F (V)$ for $V \subseteq U$ .

Concrete description. Elements of $F_{x}$ are equivalence classes of pairs $(U, s)$ with $U$ an open neighbourhood of $x$ and $s \in F (U)$ . The equivalence relation: $(U, s) \sim (V, t)$ iff there exists an open $W \subseteq U \cap V$ with $x \in W$ and $s ∣_{W} = t ∣_{W}$ . The equivalence class $[U, s] \in F_{x}$ is the germ of $s$ at $x$ , denoted $s_{x}$ .

Universal property. For each open $U ∋ x$ , the natural map $F (U) \to F_{x}$ , $s \mapsto s_{x}$ , is a morphism in the appropriate category (sets, groups, rings). For any other system $F (U) \to G$ compatible with the restriction maps, there is a unique factorisation through $F_{x}$ .

Key properties for sheaves.

(S1) Locality from stalks. If $F$ is a sheaf and $s, t \in F (U)$ have $s_{x} = t_{x}$ for every $x \in U$ , then $s = t$ . (For sheaves, sections are determined by their germs.)

(S2) Morphism is iso iff stalks are iso. A morphism $φ : F \to G$ of sheaves is an isomorphism iff $φ_{x} : F_{x} \to G_{x}$ is an isomorphism for every $x \in X$ . (For sheaves, isomorphism is detected stalkwise.)

(S3) Exactness from stalks. A sequence of sheaves $F^{'} \to F \to F^{''}$ is exact iff each stalk sequence $F_{x}^{'} \to F_{x} \to F_{x}^{''}$ is exact.

These properties are failures for general presheaves; they characterise sheaves.

Local rings of a scheme. For a scheme $X$ and a point $x \in X$ , the stalk $O_{X, x}$ of the structure sheaf is a local ring — a commutative ring with a unique maximal ideal $m_{x}$ . The residue field $κ (x) := O_{X, x} / m_{x}$ is the residue field at $x$ . For an affine scheme $X = Spec (R)$ and a prime $p \in X$ , $O_{X, p} = R_{p}$ is the localisation of $R$ at $p$ , with $m_{p} = p R_{p}$ and $κ (p) = R_{p} / p R_{p} = Frac (R / p)$ .

Skyscraper sheaf. For a point $x \in X$ and an abelian group (or ring, or module) $A$ , the skyscraper sheaf $i_{x, *} A$ is the sheaf with stalk $A$ at $x$ and zero elsewhere:

(i_{x, *} A) (U) = {A 0 x \in U x \in / U .

Skyscraper sheaves are the simplest stalk-supported sheaves. Quotients by skyscraper sheaves appear throughout the theory of divisors and curves.

Étale space. The étale space of $F$ is the topological space $\overset{ˊ}{E} t (F) = ⨆_{x \in X} F_{x}$ , equipped with a map $π : \overset{ˊ}{E} t (F) \to X$ sending germs at $x$ to $x$ , and topologised so that local sections $s \in F (U)$ define continuous sections $U \to \overset{ˊ}{E} t (F)$ , $x \mapsto s_{x}$ . The étale space recovers $F$ as its sheaf of continuous local sections; this is the "espace étalé" perspective of Cartan-Godement, where sheaves are étale spaces.

Key theorem with proof [Intermediate+]

Theorem (sheaves are determined by their stalks and restriction maps). Let $F, G$ be sheaves on $X$ , and $φ, ψ : F \to G$ two morphisms. The following are equivalent.

(i) $φ = ψ$ as morphisms of sheaves.

(ii) $φ (U) = ψ (U)$ for every open $U \subseteq X$ .

(iii) $φ_{x} = ψ_{x}$ for every $x \in X$ .

Proof. (i) $\Leftrightarrow$ (ii) is the definition of morphism equality (a morphism of sheaves is a natural transformation, equality is sectionwise).

(ii) $\Rightarrow$ (iii): The stalk map $φ_{x}$ is induced by passing to the colimit over $U ∋ x$ . If $φ (U) = ψ (U)$ for every $U$ , the colimit maps are also equal: $φ_{x} = ψ_{x}$ .

(iii) $\Rightarrow$ (ii): Assume $φ_{x} = ψ_{x}$ for every $x \in X$ . Let $U \subseteq X$ be open and $s \in F (U)$ . We must show $φ (U) (s) = ψ (U) (s)$ in $G (U)$ .

For each $x \in U$ , the stalks satisfy $φ_{x} (s_{x}) = ψ_{x} (s_{x})$ . The germ $φ_{x} (s_{x}) \in G_{x}$ is the germ of $φ (U) (s)$ at $x$ (since the stalk map is induced by $φ (U)$ ); similarly for $ψ$ . So $φ (U) (s)_{x} = ψ (U) (s)_{x}$ in $G_{x}$ for every $x \in U$ .

By the locality axiom (S1) for the sheaf $G$ : two sections of $G$ on $U$ that have the same germ at every point are equal. So $φ (U) (s) = ψ (U) (s)$ . $□$

This theorem is the foundational reason stalks matter: sheaves are determined by their pointwise (stalk) data plus the gluing structure. To check sheaf-theoretic statements, it suffices to check at every stalk. This pointwise / local perspective is what makes sheaves the right tool for local-to-global problems.

Bridge. The construction here builds toward 04.01.03 (sheafification), where the same data is upgraded, and the symmetry side is taken up in 04.01.04 (direct and inverse image of sheaves). The defining pattern appears again in those units in a sharpened form, where the local data is glued or quotiented. Putting these together, the foundational insight is that the data of this unit gives the structural signature that the rest of the strand reads off.

Exercises [Intermediate+]

Exercise 1 (easy, proof).

Show that for the structure sheaf of an affine scheme $X = Spec (R)$ , the stalk at a prime $p$ is the localisation $R_{p}$ .

Hint

Distinguished opens $D (f)$ form a basis; sections on $D (f)$ are $R [f^{- 1}]$ .

Answer

The distinguished opens $D (f) = {q \in Spec (R) : f \in / q}$ form a basis for the topology of $Spec (R)$ . The structure sheaf has $O_{X} (D (f)) = R [f^{- 1}]$ .

The stalk at $p$ is the colimit over open neighbourhoods, which we can compute over the basis of distinguished opens $D (f) ∋ p$ — i.e., over $f \in / p$ :

O_{X, p} = f \in / p lim R [f^{- 1}] .

The directed colimit of localisations $R [f^{- 1}]$ over $f \in / p$ is precisely the localisation $R_{p} = (R ∖ p)^{- 1} R$ : every element $r / g$ with $g \in / p$ lifts to a section in some $R [g^{- 1}]$ , and equality in the colimit is determined by equality after further localisation, which matches the definition of localisation. $□$

The stalk $O_{X, p} = R_{p}$ is a local ring with unique maximal ideal $p R_{p}$ — the algebraic incarnation of the geometric notion of "ring of germs of functions near $p$ ."

Exercise 3 (medium, proof).

Show that a morphism $φ : F \to G$ of sheaves is injective (in the category of sheaves) iff $φ_{x}$ is injective for every $x \in X$ .

Hint

Use the locality axiom: a section is zero iff every germ is zero.

Answer

$(\Rightarrow)$ Suppose $φ$ is injective on sheaves: for every $U$ , $φ (U) : F (U) \to G (U)$ is injective. The stalk map $φ_{x} : F_{x} \to G_{x}$ is the colimit of the maps $φ (U)$ over $U ∋ x$ . Filtered colimits of injections in $Set$ (or $Ab$ , or $R$ -modules) preserve injectivity: if every $φ (U)$ is injective, the colimit is injective.

$(\Leftarrow)$ Conversely, suppose every $φ_{x}$ is injective. Let $s \in F (U)$ with $φ (U) (s) = 0$ in $G (U)$ . Then for every $x \in U$ , the germ $φ_{x} (s_{x}) = φ (U) (s)_{x} = 0_{x} = 0$ in $G_{x}$ . By injectivity of $φ_{x}$ , $s_{x} = 0$ in $F_{x}$ for every $x \in U$ .

By the locality axiom for the sheaf $F$ : a section whose germs are all zero is zero. So $s = 0$ . Hence $φ (U)$ is injective for every $U$ , i.e., $φ$ is an injection of sheaves. $□$

The corresponding statement for surjectivity is more subtle: $φ$ is surjective on stalks iff the image presheaf is dense in $G$ , but the surjectivity of $φ (U)$ on each open $U$ is not equivalent — the difference is precisely the obstruction measured by sheaf cohomology $H^{1}$ .

Exercise 5 (medium, conceptual).

Why does the étale-space construction give an equivalence between sheaves and "local homeomorphisms" $π : E \to X$ ?

Hint

The étale space topologises the disjoint union of stalks so that local sections are continuous.

Answer

The étale space of a sheaf $F$ on $X$ is

\overset{ˊ}{E} t (F) = x \in X ⨆ F_{x},

with $π : \overset{ˊ}{E} t (F) \to X$ , $s_{x} \mapsto x$ . The topology on $\overset{ˊ}{E} t (F)$ is the smallest making continuous all sections $s : U \to \overset{ˊ}{E} t (F)$ , $x \mapsto s_{x}$ , where $s \in F (U)$ ranges over local sections.

In this topology, $π$ is a local homeomorphism (étale map): every $ξ \in \overset{ˊ}{E} t (F)$ has an open neighbourhood mapped homeomorphically onto an open subset of $X$ by $π$ .

Conversely, given a local homeomorphism $π : E \to X$ , the sheaf of continuous local sections $F (U) := {s : U \to E continuous, π \circ s = id_{U}}$ is a sheaf on $X$ , and $\overset{ˊ}{E} t (F) ≅ E$ .

This Espace Étalé / sheaf equivalence (Cartan, Godement) gives a topological description of sheaves. Stalks are precisely the fibres of $π$ , and morphisms of sheaves correspond to fibre-preserving continuous maps over $X$ .

The étale-space perspective is foundational: it explains why "stalk" is the right notion (fibre over a point = stalk at that point), why injectivity / surjectivity of sheaf maps is detected stalkwise (= fibrewise for the étale spaces), and why sheaf cohomology is a topological invariant of the étale space. Grothendieck's site generalisation (étale topology on schemes) extends this view to non-Hausdorff settings.

Exercise 6 (hard, proof).

Show that for sheaves of abelian groups, the kernel sheaf has stalks the kernels of stalks: $(ker φ)_{x} = ker (φ_{x})$ .

Hint

Filtered colimits commute with finite limits in $Ab$ .

Answer

The kernel sheaf $ker φ$ is the sheaf $U \mapsto ker (φ (U) : F (U) \to G (U))$ . (For sheaves of abelian groups, this is automatically a sheaf — kernels in $Sh (X; Ab)$ coincide with presheaf kernels.)

The stalk at $x$ is

(ker φ)_{x} = U ∋ x lim ker (φ (U)) .

We claim this equals $ker (φ_{x} : F_{x} \to G_{x})$ .

Forward containment. A germ $[(U, s)] \in (ker φ)_{x}$ has $s \in ker φ (U)$ , i.e., $φ (U) (s) = 0$ . Then $φ_{x} (s_{x}) = φ (U) (s)_{x} = 0_{x} = 0$ , so $s_{x} \in ker φ_{x}$ .

Reverse containment. Conversely, take $[s_{x}] \in ker φ_{x}$ with $φ_{x} (s_{x}) = 0$ in $G_{x}$ . By the colimit description, there exists an open neighbourhood $V \subseteq U$ of $x$ with $φ (V) (s ∣_{V}) = 0$ in $G (V)$ . So $s ∣_{V} \in ker φ (V)$ , and $[s_{x}] = [(V, s ∣_{V})] \in (ker φ)_{x}$ .

Equivalently: filtered colimits commute with finite limits in $Ab$ (more generally in any locally finitely presentable category). The kernel is a finite limit, the stalk is a filtered colimit, so they commute. $□$

This is the prototype of exactness preserved by stalks: kernels, cokernels, and (finite) products are computed stalkwise. This is why exact sequences of sheaves correspond exactly to exact sequences at every stalk — the foundational property organising sheaf cohomology and homological algebra of sheaves.

Exercise 7 (hard, conceptual).

Explain how the local rings of a scheme — the stalks of the structure sheaf — encode the geometric properties (smoothness, dimension, multiplicity) of the scheme.

Hint

A regular local ring of dimension $n$ corresponds to a smooth point of dimension $n$ .

Answer

For a scheme $X$ and a point $x \in X$ , the local ring $O_{X, x}$ encodes the local geometry at $x$ :

Dimension. $dim X$ at $x$ is the Krull dimension $dim O_{X, x}$ of the local ring (length of the longest chain of prime ideals).
Smoothness. $X$ is smooth (regular) at $x$ iff $O_{X, x}$ is a regular local ring: the maximal ideal is generated by exactly $dim O_{X, x}$ elements (equivalently, $dim_{κ (x)} m_{x} / m_{x}^{2} = dim O_{X, x}$ ).
Multiplicity. For a singular point, the multiplicity of $X$ at $x$ is the leading coefficient of the Hilbert function of the graded ring $gr_{m_{x}} O_{X, x} = ⨁_{n} m_{x}^{n} / m_{x}^{n + 1}$ , with appropriate normalisation.
Tangent space. The Zariski tangent space at $x$ is $T_{x} X := (m_{x} / m_{x}^{2})^{\lor}$ — the dual of the cotangent space, viewed as a vector space over $κ (x)$ . This is the algebraic incarnation of the tangent space of differential geometry.
Cotangent complex. For singular schemes, the cotangent complex $L_{X / k}$ at $x$ generalises the Zariski cotangent: $H^{0} (L) = Ω_{X / k}^{1} ∣_{x}$ , $H^{- 1} (L)$ records obstruction information for deformations.
Local-global principle. Many global properties of $X$ are detected stalkwise: $X$ is reduced iff every $O_{X, x}$ is reduced; $X$ is normal iff every $O_{X, x}$ is normal; $X$ is Cohen-Macaulay iff every $O_{X, x}$ is. Failure of these properties has a local origin at specific points.

The local rings of a scheme are the algebraic data sufficient to reconstruct the scheme as a locally ringed space (modulo the topology). Grothendieck's perspective: schemes are the locally-ringed-space pairs $(X, O_{X})$ where stalks are local rings, locally isomorphic to spectra of commutative rings. The geometry lives in the local rings.

Lean formalization [Intermediate+]

lean_status: partial — Mathlib has TopCat.Presheaf.stalk and the full stalk colimit construction.

[object Promise]

Advanced results [Master]

Stalkwise exactness. A sequence of sheaves of abelian groups $0 \to F^{'} \to F \to F^{''} \to 0$ is exact iff each stalk sequence $0 \to F_{x}^{'} \to F_{x} \to F_{x}^{''} \to 0$ is exact. This characterisation is the foundation of homological algebra of sheaves: the abelian category of sheaves of abelian groups has stalkwise-detected exactness.

Stalks vs. global sections. $H^{0} (X; F) = F (X)$ records global sections; the stalks $F_{x}$ record local sections. The local-to-global spectral sequence (Leray, Cartan-Eilenberg) computes $H^{p} (X; F)$ from local data via the étale space and Čech cohomology. The stalk vanishing of $F$ does not imply $H^{0} (X; F) = 0$ ; conversely, vanishing of $H^{0}$ does not imply stalk vanishing. This is the obstruction encoded by higher cohomology.

Skyscraper sheaves and torsion. Skyscraper sheaves $i_{x, *} A$ have the unique nonzero stalk $A$ at $x$ . They serve as "torsion" or "punctual" contributions in long exact sequences. For example, on a smooth curve $C$ and a point $p$ , the sequence

0 \to O_{C} \to O_{C} (p) \to i_{p, *} κ (p) \to 0

relates the structure sheaf, the line bundle of $p$ , and the skyscraper at $p$ — the inductive engine of the Riemann-Roch proof.

Henselian and strict henselian local rings. Beyond the stalk $O_{X, x}$ , deeper local invariants are the Henselisation $O_{X, x}^{h}$ (smallest henselian extension) and the strict Henselisation $O_{X, x}^{s h}$ (smallest henselian extension with separably closed residue field). These are the local rings of $X$ in the étale topology — the fibres of the étale stalk of a sheaf.

Stalks in the étale topos. For a scheme $X$ and the étale site $X_{\overset{e}{ˊ} t}$ , the stalk of a sheaf $F$ at a geometric point $\overset{x}{ˉ}$ (a morphism $Spec (\overset{κ}{ˉ} (x)) \to X$ ) is computed by the colimit over étale neighbourhoods. This is the foundation of étale cohomology and the Weil conjectures (Grothendieck-Deligne).

Stalks in derived algebraic geometry. Lurie's $\infty$ -categorical reformulation: a sheaf of $\infty$ -categories or sheaf of spectra has higher stalks $F_{x}$ that are themselves $\infty$ -categorical / spectral objects. The cotangent complex $L_{X / k}$ has stalks at $x$ that are derived objects controlling deformations of $X$ at $x$ . This extends classical stalks to a homotopical setting.

Microlocal sheaves. The microlocalisation (Kashiwara-Schapira) of a sheaf $F$ on a manifold passes to the cotangent bundle, recording not just stalks at points but stalks at codirections: $F_{(x, ξ)}$ for $(x, ξ) \in T^{*} X$ . This is the foundation of microlocal analysis and the geometry of D-modules.

Synthesis. This construction generalises the pattern fixed in 04.01.01 (sheaf), with the symmetric data replaced by its skew or twisted analogue. Read in the opposite direction, the construction is dual to the metric story: complements and orthogonality are taken with respect to the bilinear datum of this unit, not a metric, and the resulting category of subobjects is the one the rest of the strand classifies. The central insight is that this datum identifies algebra with geometry: functions become vector fields, subspaces become quotients, and invariants become cohomology classes — and that identification is the engine driving every theorem downstream.

Full proof set [Master]

Detailed proofs of: stalkwise exactness for sheaves of abelian groups, the equivalence between sheaves and étale spaces (Cartan-Godement), agreement of $H^{0}$ and global sections, and the local-to-global spectral sequence — these are deferred to companion units in the sheaf-cohomology and étale-cohomology strands. The basic stalk-determination theorem is proved in the formal-definition section.

Connections [Master]

Sheaf 04.01.01 — stalks are the fundamental local invariants of a sheaf; sheaves are determined by stalks plus gluing.
Sheafification 04.01.03 — sheafification preserves stalks: $F_{x} = (F^{+})_{x}$ for any presheaf $F$ .
Direct and inverse image 04.01.04 — the inverse image $f^{- 1} G$ at a point $x$ has stalk $G_{f (x)}$ ; this is the key stalk-functoriality property.
Scheme 04.02.01 — the local rings $O_{X, x}$ are the stalks of the structure sheaf and encode local geometry.
Affine scheme 04.02.02 — for $X = Spec (R)$ , the stalk at $p$ is the localisation $R_{p}$ .
Sheaf cohomology 04.03.01 — stalkwise exactness is the foundation of long exact sequences in sheaf cohomology.
Coherent sheaf 04.06.02 — a coherent sheaf has finitely generated stalks, the local-finiteness condition.

Historical & philosophical context [Master]

The notion of sheaf, with its stalks, was introduced by Jean Leray during his imprisonment as a French officer at Oflag XVII-A in Edelbach, Austria, from 1940 to 1945. Leray was a leading specialist in fluid dynamics and partial differential equations; to avoid being put to work for the German war effort, he disguised his expertise and presented himself as a topologist working on what he called harmless algebraic topology. While imprisoned, he gave lectures to fellow prisoners, eventually establishing a "university in captivity" (université en captivité) where he and other French academics taught one another. Out of this mathematical correspondence and these lectures came his foundational works of 1942–46, published initially in the Journal de Mathématiques Pures et Appliquées and the Comptes Rendus. His 1946 L'anneau d'homologie d'une représentation (Comptes Rendus 222) introduced sheaves (under the name faisceau), spectral sequences, and what is now called Leray cohomology — a machinery developed entirely under wartime captivity and offered as harmless algebraic topology while in fact constituting one of the most important mathematical innovations of the 20th century.

Leray's stalks were geometric and somewhat informal; he thought of a sheaf as a system of local data attached to each point of a topological space. The systematic treatment came from Henri Cartan's seminars at the École Normale Supérieure (1948–1954). Cartan's exposés introduced the precise definition of a sheaf as a presheaf satisfying the locality and gluing axioms, established the stalk as the colimit over open neighbourhoods, and developed the étale-space picture (sheaf = local homeomorphism) that gave a topological meaning to sheaves. Jean-Pierre Serre's 1955 Faisceaux Algébriques Cohérents (FAC) made the leap to algebraic geometry: the stalks of the structure sheaf $O_{X}$ on a variety are local rings, and the geometry of $X$ at a point is encoded in this local ring. Roger Godement's 1958 monograph Topologie algébrique et théorie des faisceaux gave the complete systematic treatment, including the comparison of Čech and derived-functor cohomology, and remains a standard reference.

Alexander Grothendieck's transformation in the late 1950s (Tôhoku 1957, EGA from 1960) made stalks the foundational pointwise invariants of sheaves on schemes. The local rings $O_{X, x}$ became the algebraic data sufficient to reconstruct a scheme as a locally ringed space. Grothendieck's later work (SGA 1-4, 1960s) extended the stalk concept to the étale topology, with geometric points $\overset{x}{ˉ}$ as the relevant base points and Henselian local rings as the relevant local invariants. The étale stalks are what enabled the proof of the Weil conjectures (Deligne, 1973) and continue to drive arithmetic geometry.

In contemporary mathematics, stalks appear in many forms: as fibres of derived $\infty$ -categorical sheaves (Lurie), as microlocal stalks at points of the cotangent bundle (Kashiwara-Schapira), as stalks in $p$ -adic / perfectoid settings (Scholze), and as stalks in homotopy type theory (univalent sheaf theory). The single concept — the local data of a sheaf at a point — has thus traversed an enormous landscape, from Leray's wartime fluid-dynamics-disguised-as-topology, through Cartan-Serre-Grothendieck's foundational reformulations, to the cutting edge of derived and arithmetic geometry. Leray's 1946 papers and Cartan's seminar exposés remain readable today and reward direct study; Godement's Topologie algébrique et théorie des faisceaux is the canonical mid-century reference.

Bibliography [Master]

Hartshorne, Algebraic Geometry — §II.1 covers sheaves and stalks systematically.
Vakil, The Rising Sea: Foundations of Algebraic Geometry — §2.4 (stalks and sheafification).
Leray, "L'anneau d'homologie d'une représentation," Comptes Rendus 222 (1946) — original introduction of sheaves and stalks.
Cartan, Séminaire Cartan (1948–1954) — exposés on sheaves and étale spaces.
Godement, Topologie algébrique et théorie des faisceaux (1958) — the systematic mid-century reference.
Bredon, Sheaf Theory — Ch. I, foundations of sheaves and stalks.
Iversen, Cohomology of Sheaves — Ch. II, sheaves and their stalks.
Grothendieck, Tôhoku paper (Sur quelques points d'algèbre homologique, 1957) — sheaves on sites and the categorical framework.
Kashiwara-Schapira, Sheaves on Manifolds — microlocal sheaves and stalks at codirections.
Lurie, Higher Topos Theory — $\infty$ -categorical sheaves and stalks.
Eckmann-Cartan-Eilenberg, Homological Algebra — stalks and derived functors.

Prerequisites

04.01.01

Used in

04.01.03
04.01.04

Tier anchors

beginner: Strogatz analogous level for the germ of a function at a point
intermediate: Hartshorne §II.1; Vakil §2.4; Bredon Ch. I
master: Hartshorne §II; Bredon *Sheaf Theory*; Iversen *Cohomology of Sheaves*; Godement *Topologie algébrique et théorie des faisceaux*

References

TODO_REF
Hartshorne — Algebraic Geometry · §II.1, sheaves and stalks
TODO_REF
Vakil — The Rising Sea · §2.4, stalks and sheafification
TODO_REF
Leray — L'anneau d'homologie d'une représentation · 1946, original introduction of sheaves in Oflag XVII-A
TODO_REF
Godement — Topologie algébrique et théorie des faisceaux · 1958, the systematic treatment of sheaves and stalks
TODO_REF
Bredon — Sheaf Theory · Ch. I, foundations of sheaves and stalks

Lean module

Codex.AlgebraicGeometry.Sheaves.Stalk

Reviewer

TBD

Estimated time

beginner: 10m
intermediate: 30m
master: 50m