Alexander Grothendieck
The visionary who rebuilt the foundations of algebraic geometry and reshaped the landscape of twentieth-century mathematics through the power of abstraction
23.1 Life & Intellectual Journey
Alexander Grothendieck (28 March 1928 – 13 November 2014) was perhaps the most profoundly original mathematician of the twentieth century. His life was marked by displacement, radical independence of thought, and an uncompromising moral vision that eventually led him to abandon the mathematical establishment at the height of his powers. No other figure has so thoroughly reshaped an entire branch of mathematics: algebraic geometry, before and after Grothendieck, are almost unrecognizable as the same subject.
Timeline of Grothendieck's Life
- 1928 — Born in Berlin to Alexander (Sascha) Schapiro, a Russian-Jewish anarchist, and Hanka Grothendieck, a German journalist. His father had lost an arm fighting in the Russian Revolution.
- 1933–1939 — Parents flee Nazi Germany. Alexander is left with a foster family in Hamburg while his parents join the Spanish Civil War. He is effectively stateless, holding a Nansen passport for displaced persons.
- 1940–1944 — Interned with his mother at the Rieucros camp in Lozère, then at various camps in Vichy France. His father perishes at Auschwitz in 1942. Despite the conditions, Alexander attends the local lycée in Mende.
- 1945–1948 — Studies at the University of Montpellier. Dissatisfied with the instruction, he independently reconstructs the Lebesgue theory of measure and integration — unaware that it already existed. This episode reveals the hallmarks of his mathematical personality: the drive to rebuild from scratch rather than learn from textbooks.
- 1949–1953 — Arrives in Paris. On Cartan and Weil’s advice, moves to Nancy to work under Laurent Schwartz and Jean Dieudonné. Completes his doctoral thesis on topological tensor products and nuclear spaces — a work of extraordinary depth that solves in one stroke the major open problems Schwartz had proposed.
- 1955–1958 — First stay in Kansas, then visiting positions. Begins the revolution in algebraic geometry with his theory of schemes. Delivers the legendary talk at the 1958 ICM (Edinburgh) on the Riemann–Roch theorem.
- 1958–1970 — The golden era at the IHES (Institut des Hautes Études Scientifiques) in Bures-sur-Yvette. With an extraordinary group of collaborators, produces the monumental EGA and SGA seminars, refounding algebraic geometry.
- 1966 — Awarded the Fields Medal at the Moscow ICM, but refuses to travel to the Soviet Union in protest against the political repression. Léon Motchane collects the medal on his behalf.
- 1970 — Discovers that IHES receives partial funding from the French military (DRET). Unable to reconcile this with his pacifist convictions, he resigns, effectively ending his most productive period.
- 1970–1973 — Founds the ecological and anti-militarist group “Survivre et Vivre.” Holds a temporary position at the Collège de France, teaching unconventional courses on ecology and society.
- 1973–1984 — Professor at the University of Montpellier. Continues mathematical work, producing among other things the magnificent “Pursuing Stacks” (1983).
- 1984 — Submits his Esquisse d’un Programme to the CNRS — a visionary document proposing a programme of research connecting Galois theory, topology, and the moduli of curves. Also writes the enormous autobiographical manuscript Récoltes et Semailles.
- 1991 — Withdraws from all mathematical and social contact, retreating to the village of Lasserre in the Ariège Pyrenees.
- 2014 — Dies on 13 November at the Saint-Girons hospital, aged 86.
“The introduction of the cipher 0 or the group concept was general nonsense too, and mathematics was more or less stagnating for thousands of years because nobody was around to take such childish steps.”
— Alexander Grothendieck
Grothendieck’s mathematical style was unlike any other. Where most mathematicians attack specific problems with clever tricks, Grothendieck would patiently build vast theoretical frameworks in which the problems would dissolve naturally — what he called the “rising sea” approach. He famously described it with the metaphor of a nut: rather than cracking it with a hammer and chisel, you immerse it in a softening liquid and wait until it opens of its own accord. This philosophy pervades everything he created.
His output during the IHES years was staggering. The Éléments de Géométrie Algébrique (EGA), written with Dieudonné, eventually ran to thousands of pages across four published volumes (of a planned thirteen). The Séminaire de Géométrie Algébrique (SGA) comprised seven multi-volume seminar notes. Together, these works constitute perhaps the most ambitious and sustained effort of mathematical writing in history.
23.2 Nuclear Spaces & Tensor Products
Grothendieck’s doctoral thesis, Produits tensoriels topologiques et espaces nucléaires (1955), was his first masterpiece. Laurent Schwartz had posed a list of open problems in the theory of topological vector spaces; the young Grothendieck solved virtually all of them, and in doing so introduced an entirely new framework for understanding the relationship between different topological tensor products.
Given two locally convex topological vector spaces E and F, there are many natural ways to topologize the algebraic tensor product \(E \otimes F\). Grothendieck identified two extreme cases — the projective and injective tensor products — and showed that the spaces for which these two extremes coincide form a remarkable class: the nuclear spaces.
Definition: Projective Tensor Product
Let \(E\) and \(F\) be locally convex spaces with continuous seminorms \(p\) and \(q\) respectively. The projective tensor norm (or π-norm) on \(E \otimes F\) is defined by:
The infimum is taken over all finite representations of the element \(u\). This is the largest reasonable cross-norm. The completion with respect to \(\pi\) is denoted \(E \hat{\otimes}_\pi F\).
The projective tensor product has the universal property that bilinear maps from \(E \times F\) to any locally convex space \(G\) correspond bijectively to linear maps from \(E \hat{\otimes}_\pi F\) to \(G\). In categorical language, it is the coproduct in the category of locally convex spaces with bilinear maps.
Definition: Injective Tensor Product
The injective tensor norm (or ε-norm) is defined by testing against dual functionals:
where \(B_{E^*}\) and \(B_{F^*}\) are the unit balls in the strong duals. This is the smallest reasonable cross-norm. The completion is denoted \(E \hat{\otimes}_\varepsilon F\).
For any element \(u \in E \otimes F\), one always has the inequality \(\varepsilon(u) \le \pi(u)\). The question of when these two norms coincide is at the heart of Grothendieck’s thesis.
Definition: Nuclear Operator
A continuous linear operator \(T: E \to F\) between Banach spaces is called nuclear if there exist sequences \((x_n^*) \subset E^*\) and \((y_n) \subset F\) such that:
The nuclear norm of \(T\) is the infimum of \(\sum \|x_n^*\|\,\|y_n\|\) over all such representations.
Definition: Nuclear Space
A locally convex space \(E\) is nuclear if, for every continuous seminorm \(p\) on \(E\), there exists a continuous seminorm \(q \ge p\) such that the canonical map \(\hat{E}_q \to \hat{E}_p\) (between the completions of the quotient spaces \(E/\ker p\) and \(E/\ker q\)) is a nuclear operator.
Theorem (Grothendieck): Characterization of Nuclear Spaces
A locally convex space \(E\) is nuclear if and only if, for every locally convex space \(F\), the natural map:
is an isomorphism. That is, a space is nuclear precisely when the projective and injective tensor products coincide for all partners.
Key Examples of Nuclear Spaces
- Schwartz space \(\mathcal{S}(\mathbb{R}^n)\) of rapidly decreasing functions is nuclear.
- Space of smooth functions \(C^\infty(M)\) on a compact manifold is nuclear.
- Space of distributions \(\mathcal{D}'(\Omega)\) is nuclear.
- Infinite-dimensional Banach spaces are never nuclear — this is the Dvoretzky–Rogers theorem.
Insight: The Thesis as Proof of Genius
Schwartz later recalled: “There were 14 open problems that I had asked at the end of my book on distributions. Grothendieck solved them all during his thesis — and in the process, he created an entirely new theory.” The thesis also introduced the Grothendieck inequality (later rediscovered in theoretical computer science), which states that there exists a universal constant \(K_G\) such that for any \(n \times n\) matrix \((a_{ij})\) and unit vectors, a fundamental inequality relating bilinear forms on \(\ell^\infty\) to those on Hilbert space holds.
23.3 The Tôhoku Paper & Abelian Categories
In 1957, Grothendieck published “Sur quelques points d’algèbre homologique” in the Tôhoku Mathematical Journal — universally known as the “Tôhoku paper.” This single article revolutionized homological algebra by recasting it entirely in categorical language. Where Cartan and Eilenberg had worked with modules, Grothendieck worked with arbitrary abelian categories, achieving a vast generalization that would prove essential for algebraic geometry.
Definition: Abelian Category
An abelian category \(\mathcal{A}\) is an additive category satisfying:
- It has a zero object \(0\) (both initial and terminal).
- Every pair of objects has a biproduct (direct sum) \(A \oplus B\).
- Every morphism has a kernel and a cokernel.
- Every monomorphism is the kernel of its cokernel: \(\ker(\text{coker}\,f) = f\).
- Every epimorphism is the cokernel of its kernel: \(\text{coker}(\ker\,f) = f\).
The key insight is that conditions (4) and (5) ensure that the first isomorphism theorem holds: for any morphism \(f: A \to B\), we have a canonical isomorphism \(\text{coim}(f) \cong \text{im}(f)\). This is what makes homological algebra “work” in an abelian category.
Definition: Grothendieck's AB Axioms
Grothendieck introduced a hierarchy of additional axioms for abelian categories:
- AB3: Arbitrary coproducts (direct sums) exist. Equivalently, the category is cocomplete.
- AB4: AB3 holds and coproducts of monomorphisms are monomorphisms.
- AB5: AB3 holds and filtered colimits of exact sequences are exact. This is the crucial axiom: it says that direct limits commute with finite limits.
- AB3*–AB5*: The dual axioms involving products and inverse limits.
Theorem (Grothendieck): Enough Injectives
If \(\mathcal{A}\) is an AB5 abelian category with a generator (an object \(U\) such that \(\text{Hom}(U, -)\) is faithful), then \(\mathcal{A}\) has enough injectives: every object embeds into an injective object.
This theorem is the cornerstone of Grothendieck’s approach to sheaf cohomology: the category of sheaves of abelian groups on a topological space satisfies AB5 and has a generator, hence has enough injectives, hence derived functors can be defined.
With enough injectives in hand, Grothendieck could define derived functors in full generality. Given a left exact functor \(F: \mathcal{A} \to \mathcal{B}\) between abelian categories (where \(\mathcal{A}\) has enough injectives), the right derived functors are constructed as follows:
Definition: Ext and Tor Functors
The two most fundamental derived functors in algebra are:
where \(0 \to N \to I^0 \to I^1 \to \cdots\) is an injective resolution of \(N\). Dually:
where \(\cdots \to P_1 \to P_0 \to M \to 0\) is a projective resolution of \(M\).
Insight: The Long Exact Sequence
Given a short exact sequence \(0 \to A \to B \to C \to 0\) in \(\mathcal{A}\) and a left exact functor \(F\), the derived functors produce the fundamental long exact sequence:
This is the single most important computational tool in homological algebra. In Grothendieck’s framework, it follows formally from the axioms of abelian categories.
23.4 Schemes & the Functor of Points
Grothendieck’s most consequential contribution was the theory of schemes, which replaced classical algebraic varieties with a vastly more general and flexible concept. The key idea is to take commutative algebra seriously: an algebraic variety should be determined by its ring of functions, and the “space” should be recovered from this ring by a canonical construction.
Definition: The Spectrum of a Ring
For a commutative ring \(R\), the spectrum is the set:
equipped with the Zariski topology, whose closed sets are:
for ideals \(I \subseteq R\). The basic open sets are \(D(f) = \text{Spec}(R) \setminus V(f)\) for \(f \in R\).
Definition: Structure Sheaf
The structure sheaf \(\mathcal{O}_{\text{Spec}(R)}\) is defined on basic open sets by:
The stalk at a prime \(\mathfrak{p}\) is the local ring \(\mathcal{O}_{\text{Spec}(R), \mathfrak{p}} = R_\mathfrak{p}\). This makes \((\text{Spec}(R), \mathcal{O}_{\text{Spec}(R)})\) a locally ringed space.
Definition: Scheme
A scheme is a locally ringed space \((X, \mathcal{O}_X)\) that is locally isomorphic to an affine scheme: every point \(x \in X\) has an open neighbourhood \(U\) such that \((U, \mathcal{O}_X|_U) \cong (\text{Spec}(R), \mathcal{O}_{\text{Spec}(R)})\) for some ring \(R\).
The revolutionary aspect of this definition is that \(R\) can be any commutative ring — it need not be a finitely generated algebra over a field. This means that schemes naturally incorporate arithmetic objects: \(\text{Spec}(\mathbb{Z})\) is a perfectly good scheme, and number rings like \(\text{Spec}(\mathbb{Z}[\sqrt{-5}])\) sit inside it. This unification of geometry and arithmetic is one of Grothendieck’s deepest insights.
Example: Affine Line over Z
The scheme \(\mathbb{A}^1_\mathbb{Z} = \text{Spec}(\mathbb{Z}[x])\) contains points of many different flavors: the generic point \((0)\), “horizontal” primes like \((x-n)\) for \(n \in \mathbb{Z}\), “vertical” primes like \((p)\) for prime numbers \(p\), and closed points like \((p, x-a)\) corresponding to points of \(\mathbb{A}^1\) over finite fields.
Definition: Functor of Points
Every scheme \(X\) over a base scheme \(S\) defines a contravariant functor:
The set \(X(T)\) is called the set of \(T\)-valued points of \(X\). For affine schemes, the Yoneda lemma gives:
when \(T = \text{Spec}(B)\). This is the profound duality between geometry and algebra at the heart of scheme theory.
Theorem: Yoneda Lemma
For any functor \(F: \mathcal{C}^{op} \to \mathbf{Set}\) and any object \(X \in \mathcal{C}\), there is a natural bijection:
given by \(\eta \mapsto \eta_X(\text{id}_X)\). As a consequence, the functor \(X \mapsto h_X\) is a fully faithful embedding of \(\mathcal{C}\) into the presheaf category \(\hat{\mathcal{C}} = \text{Fun}(\mathcal{C}^{op}, \mathbf{Set})\). A scheme is therefore completely determined by its functor of points.
Definition: Fiber Product of Schemes
Given morphisms \(f: X \to S\) and \(g: Y \to S\), the fiber product \(X \times_S Y\) is the scheme representing the functor:
For affine schemes, this corresponds to tensor product of rings: \(\text{Spec}(A) \times_{\text{Spec}(R)} \text{Spec}(B) \cong \text{Spec}(A \otimes_R B)\). The fiber product exists in the category of all schemes — a non-trivial theorem that relies on gluing.
23.5 The Cohomology Machine
One of Grothendieck’s greatest achievements was the systematic development of sheaf cohomology as a universal tool in algebraic geometry. His framework — based on derived functors in abelian categories — replaced the ad hoc constructions of earlier algebraic geometers with a uniform, computable theory.
Definition: Sheaf Cohomology
Let \((X, \mathcal{O}_X)\) be a ringed space and \(\mathcal{F}\) a sheaf of \(\mathcal{O}_X\)-modules. The global sections functor:
is left exact. Its right derived functors define sheaf cohomology:
Concretely, embed \(\mathcal{F}\) into an injective resolution \(0 \to \mathcal{F} \to \mathcal{I}^0 \to \mathcal{I}^1 \to \cdots\), apply \(\Gamma(X,-)\), and take cohomology of the resulting complex.
In practice, injective resolutions are unwieldy. Grothendieck showed that one can often compute sheaf cohomology using more accessible resolutions. One approach is Čech cohomology:
Definition: Cech Cohomology
Given an open cover \(\mathfrak{U} = \{U_i\}_{i \in I}\) of \(X\), the Čech complex is:
with coboundary maps defined by the alternating sum formula:
The Čech cohomology \(\check{H}^n(\mathfrak{U}, \mathcal{F})\) is the cohomology of this complex. For a Leray cover (where all higher cohomology of intersections vanishes), Čech cohomology agrees with derived functor cohomology.
Theorem: Leray Spectral Sequence
Let \(f: X \to Y\) be a continuous map of topological spaces and \(\mathcal{F}\) a sheaf on \(X\). There is a spectral sequence:
where \(R^q f_*\) are the higher direct image sheaves. This spectral sequence is the fundamental tool for computing cohomology via a map: it decomposes the cohomology of \(X\) in terms of the cohomology of \(Y\) with coefficients in the fibers of \(f\).
Theorem: Grothendieck Spectral Sequence
Let \(F: \mathcal{A} \to \mathcal{B}\) and \(G: \mathcal{B} \to \mathcal{C}\) be left exact functors between abelian categories with enough injectives, and suppose \(F\) sends injective objects to \(G\)-acyclic objects. Then there is a spectral sequence:
The Leray spectral sequence is the special case where \(F = f_*\) (direct image) and \(G = \Gamma(Y, -)\) (global sections), using the identity:
Definition: Coherent Sheaf
A sheaf of \(\mathcal{O}_X\)-modules \(\mathcal{F}\) on a scheme \(X\) is coherent if:
- \(\mathcal{F}\) is of finite type: locally generated by finitely many sections.
- For every open \(U \subset X\) and every morphism \(\mathcal{O}_X^n|_U \to \mathcal{F}|_U\), the kernel is of finite type.
On a Noetherian scheme, coherent sheaves are the same as finitely generated \(\mathcal{O}_X\)-modules. On \(\text{Spec}(R)\) with \(R\) Noetherian, they correspond to finitely generated \(R\)-modules via the \(\widetilde{M}\) construction.
Theorem: Finiteness of Cohomology (Grothendieck)
Let \(X\) be a proper scheme over a Noetherian ring \(A\), and \(\mathcal{F}\) a coherent sheaf on \(X\). Then for all \(n \ge 0\), the cohomology groups \(H^n(X, \mathcal{F})\) are finitely generated \(A\)-modules.
Insight: Vanishing Theorems
On a Noetherian scheme of dimension \(d\), \(H^n(X, \mathcal{F}) = 0\) for all \(n > d\) and all sheaves \(\mathcal{F}\) (Grothendieck’s vanishing theorem). For an affine scheme \(X = \text{Spec}(R)\), all higher cohomology vanishes: \(H^n(X, \widetilde{M}) = 0\) for \(n \ge 1\). This is one reason affine schemes are computationally tractable.
23.6 Grothendieck–Riemann–Roch
The Riemann–Roch theorem, in its original form, is one of the gems of nineteenth-century mathematics. Grothendieck’s generalization — announced at the 1958 Edinburgh ICM but published only through the write-up by Borel and Serre — transformed it from a formula about line bundles on curves into a deep statement about the behavior of coherent sheaves under proper morphisms. This was the theorem that first made the mathematical world take notice of Grothendieck’s new methods.
Definition: Grothendieck Group K_0
For a scheme \(X\), the Grothendieck group \(K_0(X)\) is the free abelian group generated by isomorphism classes of coherent sheaves \([\mathcal{F}]\), modulo the relation:
Similarly, \(K^0(X)\) is defined using locally free sheaves (vector bundles). On a smooth variety, \(K^0(X) \cong K_0(X)\) since every coherent sheaf has a finite resolution by locally free sheaves.
A proper morphism \(f: X \to Y\) induces a pushforward on K-groups:
This is the derived pushforward in K-theory. The key question is: how does the Chern character interact with this pushforward?
Definition: Chern Character
For a vector bundle \(\mathcal{E}\) of rank \(r\) with Chern roots \(\alpha_1, \ldots, \alpha_r\) (the formal roots of the Chern polynomial), the Chern character is:
where \(c_i(\mathcal{E})\) are the Chern classes. The Chern character is a ring homomorphism \(\text{ch}: K^0(X) \to A^*(X)_\mathbb{Q}\) from K-theory to the Chow ring tensored with \(\mathbb{Q}\).
Definition: Todd Class
The Todd class of a vector bundle \(\mathcal{E}\) with Chern roots \(\alpha_1, \ldots, \alpha_r\) is:
The Todd class arises naturally from the relationship between the Euler characteristic and topological invariants. The first few terms are computed by expanding the generating function \(\frac{x}{1-e^{-x}} = 1 + \frac{x}{2} + \frac{x^2}{12} - \frac{x^4}{720} + \cdots\) and expressing the symmetric functions of \(\alpha_i\) in terms of Chern classes.
Theorem: Grothendieck-Riemann-Roch (GRR)
Let \(f: X \to Y\) be a proper morphism of smooth varieties, and \(\mathcal{F}\) a coherent sheaf on \(X\). Let \(T_f = T_X - f^*T_Y\) denote the relative tangent bundle. Then in the Chow ring of \(Y\):
In words: the Chern character of the K-theoretic pushforward equals the Chow-theoretic pushforward of the Chern character corrected by the Todd class of the relative tangent bundle. The Todd class is the “correction factor” for the failure of the Chern character to commute with pushforward.
Derivation: Classical Riemann-Roch from GRR
We derive the classical Riemann–Roch theorem for a smooth projective curve as a special case of GRR.
Setup: Let \(X\) be a smooth projective curve of genus \(g\) over an algebraically closed field, and \(\mathcal{L}\) a line bundle of degree \(d\). Take \(Y = \text{Spec}(k)\) (a point) and \(f: X \to Y\) the structure morphism.
Step 1: Left-hand side.
The K-theoretic pushforward is:
since \(H^i = 0\) for \(i \ge 2\) on a curve. Taking the Chern character (which on a point is just the rank):
Step 2: Right-hand side — Chern character.
Since \(\mathcal{L}\) is a line bundle with \(c_1(\mathcal{L}) = d \cdot [\text{pt}]\):
(higher powers vanish since \(\dim X = 1\)).
Step 3: Right-hand side — Todd class.
The relative tangent bundle is \(T_f = T_X\) (since \(T_Y = 0\)). The tangent bundle of a curve has \(c_1(T_X) = (2-2g)\cdot[\text{pt}]\) by definition of genus. Therefore:
Step 4: Multiply and push forward.
Now \(f_*: A^*(X) \to A^*(\text{pt}) \cong \mathbb{Z}\) sends \([\text{pt}] \mapsto 1\) and kills \(A^0(X)\). Therefore:
Step 5: Equate.
GRR gives us:
This is the classical Riemann–Roch theorem for curves. Combined with Serre duality \(h^1(X, \mathcal{L}) = h^0(X, K_X \otimes \mathcal{L}^{-1})\), it gives the full statement \(h^0(\mathcal{L}) - h^0(K_X \otimes \mathcal{L}^{-1}) = d - g + 1\).
23.7 Étale Cohomology & the Weil Conjectures
The Weil conjectures (1949) proposed a deep analogy between the topology of algebraic varieties over \(\mathbb{C}\) and the arithmetic of varieties over finite fields \(\mathbb{F}_q\). Weil himself proved them for curves and abelian varieties, but the general case seemed to require a cohomology theory for varieties over finite fields that would play the role of singular cohomology over \(\mathbb{C}\). Grothendieck created such a theory: étale cohomology.
Definition: Weil Zeta Function
Let \(X\) be a smooth projective variety of dimension \(n\) over \(\mathbb{F}_q\). The Weil zeta function is the generating function for point counts:
where \(|X(\mathbb{F}_{q^m})|\) counts the number of \(\mathbb{F}_{q^m}\)-rational points. This encodes all the arithmetic information about \(X\).
The Weil Conjectures (1949)
For a smooth projective variety \(X/\mathbb{F}_q\) of dimension \(n\):
- Rationality: \(Z(X,t)\) is a rational function of \(t\):$$Z(X,t) \;=\; \frac{P_1(t) \cdot P_3(t) \cdots P_{2n-1}(t)}{P_0(t) \cdot P_2(t) \cdots P_{2n}(t)}$$with \(P_0(t) = 1-t\), \(P_{2n}(t) = 1-q^nt\), and each \(P_i(t) \in \mathbb{Z}[t]\).
- Functional equation: \(Z(X, 1/q^nt) = \pm q^{n\chi/2}\,t^\chi\,Z(X,t)\), where \(\chi\) is the Euler characteristic.
- Riemann hypothesis: The roots of \(P_i(t)\) have absolute value \(q^{-i/2}\). Equivalently, writing \(P_i(t) = \prod(1-\alpha_{ij}t)\), we have \(|\alpha_{ij}| = q^{i/2}\).
- Betti numbers: \(\deg P_i\) equals the \(i\)-th Betti number of the corresponding complex variety.
Grothendieck’s strategy was to build a cohomology theory that would make these conjectures follow from a Lefschetz fixed-point theorem, just as classical point-counting on manifolds follows from the Lefschetz theorem in algebraic topology.
Definition: Etale Morphism
A morphism of schemes \(f: X \to Y\) is étale if it is:
- Flat: \(\mathcal{O}_{Y,f(x)} \to \mathcal{O}_{X,x}\) is flat for all \(x\).
- Unramified: For all \(x \in X\), the fiber \(\mathfrak{m}_{f(x)}\mathcal{O}_{X,x} = \mathfrak{m}_x\) and the residue field extension \(k(f(x)) \hookrightarrow k(x)\) is separable.
Equivalently, \(f\) is étale if and only if it is flat and the sheaf of relative differentials vanishes: \(\Omega_{X/Y} = 0\). Étale morphisms are the algebraic analogue of local homeomorphisms (or local diffeomorphisms in differential geometry).
The étale morphisms to a scheme \(X\) form a site (the étale site \(X_\text{ét}\)), and one can define sheaves and sheaf cohomology on this site. However, the resulting cohomology groups with, say, \(\mathbb{Z}\) coefficients are not well-behaved. Grothendieck’s solution was to use ℓ-adic coefficients.
Definition: l-adic Cohomology
Fix a prime \(\ell\) different from \(\text{char}(\mathbb{F}_q)\). The \(\ell\)-adic cohomology of a variety \(X\) over \(\mathbb{F}_q\) is defined as the inverse limit:
where \(X_{\bar{\mathbb{F}}_q} = X \times_{\mathbb{F}_q} \bar{\mathbb{F}}_q\) is the base change to the algebraic closure. These are finite-dimensional \(\mathbb{Q}_\ell\)-vector spaces carrying a continuous action of \(\text{Gal}(\bar{\mathbb{F}}_q/\mathbb{F}_q)\).
Theorem: Grothendieck-Lefschetz Trace Formula
Let \(X/\mathbb{F}_q\) be a smooth projective variety of dimension \(n\), and let \(F: X_{\bar{\mathbb{F}}_q} \to X_{\bar{\mathbb{F}}_q}\) be the geometric Frobenius. Then:
This is the arithmetic analogue of the classical Lefschetz fixed-point theorem. Combined with the rationality of \(\ell\)-adic cohomology and Poincaré duality, this proves the first two Weil conjectures (rationality and functional equation). Grothendieck proved this in SGA 4½.
Insight: The Polynomials P_i
The trace formula immediately gives the factorization of the zeta function. Setting \(P_i(t) = \det(1 - Ft \mid H^i_\text{ét})\), one computes:
The Riemann hypothesis (part 3 of the Weil conjectures) requires showing that the eigenvalues of Frobenius on \(H^i_\text{ét}\) have the correct absolute value. This was finally proved by Deligne in 1974, using Grothendieck’s machinery but with a different strategy than Grothendieck had envisioned (which relied on the unproven Standard Conjectures).
Example: Zeta Function of Projective Space
For \(\mathbb{P}^n_{\mathbb{F}_q}\), we have \(|\mathbb{P}^n(\mathbb{F}_{q^m})| = 1 + q^m + q^{2m} + \cdots + q^{nm}\). The zeta function is:
confirming that \(P_{2i}(t) = 1 - q^i t\) and \(P_{2i+1}(t) = 1\) (odd cohomology vanishes). The eigenvalue of Frobenius on \(H^{2i}_\text{ét}\) is \(q^i\), which has absolute value \(q^{2i/2} = q^i\), confirming the Riemann hypothesis.
23.8 Motives & the Standard Conjectures
Grothendieck was dissatisfied with the proliferation of cohomology theories in algebraic geometry — de Rham, Betti, étale, crystalline — and dreamed of a universal cohomology theory that would underlie all of them. The objects of this universal theory would be motives: the “atoms” of algebraic geometry, encoding the essential cohomological information of a variety in a way independent of any particular cohomology functor.
Definition: Category of Pure Motives (Chow Motives)
The category \(\mathcal{M}_\text{rat}(k)\) of Chow motives over a field \(k\) is constructed as follows:
- Objects: Triples \((X, p, m)\) where \(X\) is a smooth projective variety, \(p \in A^{\dim X}(X \times X)_\mathbb{Q}\) is an idempotent correspondence (\(p \circ p = p\)), and \(m \in \mathbb{Z}\) is a twist.
- Morphisms: Between \((X, p, m)\) and \((Y, q, n)\):$$\text{Hom}((X,p,m),\, (Y,q,n)) \;=\; q \circ A^{\dim X + n - m}(X \times Y)_\mathbb{Q} \circ p$$
- Composition: Given by composition of correspondences via the pushforward-pullback formula:$$(\beta \circ \alpha) \;=\; (\text{pr}_{13})_*\!\left(\text{pr}_{12}^*(\alpha) \cdot \text{pr}_{23}^*(\beta)\right)$$
Every smooth projective variety \(X\) gives a motive \(h(X) = (X, \Delta_X, 0)\) where \(\Delta_X\) is the diagonal. The simplest motives are the Lefschetz motive \(\mathbb{L} = (\text{Spec}(k), \text{id}, -1)\) and the Tate motive \(\mathbb{Q}(1) = \mathbb{L}\).
Insight: Decomposition of the Diagonal
If the Standard Conjectures held, the diagonal class of a smooth projective variety \(X\) of dimension \(n\) would decompose as:
where the \(\pi_i\) are mutually orthogonal idempotents projecting onto the \(i\)-th cohomological degree. This would yield acanonical decomposition \(h(X) = h^0(X) \oplus h^1(X) \oplus \cdots \oplus h^{2n}(X)\).
Standard Conjectures (Grothendieck, 1968)
Let \(X\) be a smooth projective variety over a field \(k\), with a fixed embedding into projective space defining a hyperplane class \(H\).
- Conjecture A (Lefschetz type): The inverse of the hard Lefschetz isomorphism \(L^{n-i}: H^i(X) \xrightarrow{\sim} H^{2n-i}(X)\) is induced by an algebraic correspondence.
- Conjecture B (Hodge type): The Lefschetz decomposition is compatible with the intersection pairing, and the “primitive” Lefschetz operator \(\Lambda\) is algebraic.
- Conjecture C (Künneth type): The Künneth components \(\pi_i\) of the diagonal are algebraic cycles.
- Conjecture D (Numerical ⇔ Homological): Numerical equivalence and homological equivalence coincide for algebraic cycles.
All four conjectures remain open in general (as of 2026). Conjecture B implies A; A and D together imply C. In characteristic zero, C is known for abelian varieties (Lieberman, Katz–Messing). Conjecture D would imply that the category of numerical motives is an abelian semisimple category — Jannsen proved this is equivalent to semisimplicity.
Grothendieck envisioned that the Standard Conjectures would provide the “right” proof of the Weil conjectures — a proof through the theory of motives rather than through Deligne’s detour via monodromy. He was reportedly disappointed when Deligne proved the Riemann hypothesis for varieties over finite fields without resolving the Standard Conjectures, seeing it as a missed opportunity for deeper understanding.
23.9 Toposes & Descent Theory
Grothendieck recognized early on that the notion of “topological space” was too rigid for many purposes in algebraic geometry. The Zariski topology, for instance, has too few open sets to support a good cohomology theory. His solution was radical: replace the notion of a space with the notion of a topos, defined purely in terms of the category of sheaves on it.
Definition: Grothendieck Topology
A Grothendieck topology \(J\) on a category \(\mathcal{C}\) assigns to each object \(U\) a collection \(J(U)\) of covering sieves (subfunctors of \(h_U\)) satisfying:
- Maximality: The maximal sieve \(h_U \in J(U)\).
- Stability: If \(S \in J(U)\) and \(f: V \to U\), then \(f^*(S) \in J(V)\).
- Transitivity: If \(S \in J(U)\) and \(R\) is a sieve on \(U\) such that \(f^*(R) \in J(V)\) for all \((f: V \to U) \in S\), then \(R \in J(U)\).
A category equipped with a Grothendieck topology is called a site. The pair \((\mathcal{C}, J)\) generalizes the notion of a topological space.
Definition: Grothendieck Topos
A Grothendieck topos is a category equivalent to the category of sheaves of sets on some site: \(\mathcal{E} \simeq \text{Sh}(\mathcal{C}, J)\).
Theorem (Giraud): Characterization of Grothendieck Toposes
A category \(\mathcal{E}\) is a Grothendieck topos if and only if it satisfies:
- \(\mathcal{E}\) has all small colimits.
- Colimits are universal (preserved by pullback).
- Coproducts are disjoint.
- Every equivalence relation is effective (every congruence has a quotient).
- \(\mathcal{E}\) has a set of generators.
This intrinsic characterization allows one to work with toposes without choosing a particular site presentation.
A geometric morphism \(f: \mathcal{E} \to \mathcal{F}\) between toposes consists of an adjoint pair \(f^* \dashv f_*\) where the inverse image \(f^*\) is left exact (preserves finite limits). This is the correct notion of “continuous map” between generalized spaces.
Definition: Descent Data
Let \(f: Y \to X\) be a morphism of schemes. A descent datum relative to \(f\) for a quasi-coherent sheaf is a quasi-coherent sheaf \(\mathcal{G}\) on \(Y\) together with an isomorphism (the gluing data):
satisfying the cocycle condition on \(Y \times_X Y \times_X Y\):
Theorem: Grothendieck's Faithfully Flat Descent
If \(f: Y \to X\) is faithfully flat and quasi-compact (fpqc), then the category of quasi-coherent sheaves on \(X\) is equivalent to the category of descent data on \(Y\) relative to \(f\):
In this case, descent is effective: every descent datum actually comes from a sheaf on \(X\). This is the algebraic-geometric analogue of the statement that a vector bundle on a manifold is determined by its transition functions.
Insight: The Cech-Alexander Complex
The descent condition can be packaged as exactness of the Čech–Alexander complex (or cosimplicial resolution). For an fpqc cover \(f: Y \to X\) and a quasi-coherent sheaf \(\mathcal{F}\) on \(X\):
is exact. This is a sheaf-theoretic analogue of the nerve of a cover, and it connects descent theory to simplicial methods that later became central in derived algebraic geometry.
23.10 Legacy & Open Problems
Grothendieck’s influence on mathematics is beyond measure. His ideas did not merely solve problems — they redefined what problems could be asked and what tools were available. Nearly every major development in algebraic geometry and number theory since 1960 builds on foundations he laid.
Downstream Triumphs
- Weil Conjectures (Deligne, 1974): Deligne’s proof of the Riemann hypothesis for varieties over finite fields used the full machinery of étale cohomology and \(\ell\)-adic sheaves that Grothendieck had built.
- Mordell Conjecture (Faltings, 1983): Faltings proved that a curve of genus \(g \ge 2\) over a number field has only finitely many rational points. His proof used Grothendieck’s theory of schemes and étale cohomology extensively.
- Fermat’s Last Theorem (Wiles, 1995): Wiles’s proof of the modularity conjecture for semistable elliptic curves rested on Galois representations in \(\ell\)-adic étale cohomology and deformation theory of Galois representations — all children of Grothendieck’s revolution.
- Langlands Programme: The geometric Langlands programme, studying automorphic forms via \(\mathcal{D}\)-modules and perverse sheaves on moduli stacks, is built entirely on Grothendieck’s infrastructure of schemes, sheaves, and functorial methods.
Grothendieck's Section Conjecture (Anabelian Geometry)
In his Esquisse d’un Programme and letter to Faltings, Grothendieck proposed that for a “sufficiently complicated” (hyperbolic) curve \(X\) over a number field \(k\), the rational points are determined by sections of the fundamental exact sequence:
Conjecture: The map \(X(k) \to \{\text{sections } \text{Gal}(\bar{k}/k) \to \pi_1^{\text{ét}}(X)\}/\text{conj.}\) is a bijection. This remains open, though partial results are known (e.g., for curves over\(p\)-adic fields by Mochizuki and others).
Homotopy Hypothesis (Grothendieck, 'Pursuing Stacks', 1983)
Grothendieck proposed that the category of \(\infty\)-groupoids (weak \(n\)-groupoids in the limit \(n \to \infty\)) should be equivalent to the homotopy category of topological spaces:
This conjecture has been established in various precise formulations (Joyal, Lurie) and is a foundational principle of modern higher category theory and derived algebraic geometry. The fundamental \(\infty\)-groupoid of a space encodes all its homotopy information.
Major Open Problems in Grothendieck's Legacy
- Standard Conjectures A–D: All remain open in characteristic \(p > 0\). They would imply the semisimplicity of the category of numerical motives and provide Grothendieck’s “motivic” proof of the Weil conjectures.
- Hodge Conjecture: The assertion that every Hodge class on a smooth projective complex variety is a rational linear combination of classes of algebraic cycles. This is a Clay Millennium Prize Problem and is closely related to the Standard Conjectures.
- Tate Conjecture: The \(\ell\)-adic analogue of the Hodge conjecture for varieties over finite fields: every Tate class is algebraic.
- Section Conjecture: As described above — open for number fields.
- Dessins d’enfants: Grothendieck’s observation that the absolute Galois group \(\text{Gal}(\bar{\mathbb{Q}}/\mathbb{Q})\) acts faithfully on combinatorial objects (bipartite graphs on surfaces) suggests deep connections between combinatorics and arithmetic that remain poorly understood.
- Motivic Galois Groups: Constructing the Tannakian formalism for mixed motives and understanding the motivic Galois group \(\mathcal{G}_{\text{mot}}\) in full generality.
- Derived Algebraic Geometry: Completing Grothendieck’s programme from “Pursuing Stacks” of reformulating algebraic geometry in the language of \(\infty\)-categories (advanced by Lurie, Toën–Vezzosi).
“Il est arrivé un moment, il y a des années de cela déjà, où le goût des mathématiques m’a quitté — comme ça, sans crier gare. [...] La découverte que j’ai faite alors [...] c’est que la créativité en science, comme en toute autre chose, et la connaissance de la vérité vivante, ne sont nullement l’apanage du “spécialiste”, du “chercheur de métier”.”
— Alexander Grothendieck, Récoltes et Semailles (1986)
In English: “There came a moment, years ago now, when the taste for mathematics left me — just like that, without warning. [...] The discovery I made then [...] is that creativity in science, as in everything else, and the knowledge of living truth, are by no means the prerogative of the ‘specialist’ or the ‘professional researcher.’”
Insight: The Rising Sea
Grothendieck’s mathematical legacy can be summarized in his own metaphor: rather than attacking a problem with force, build the right framework in which the problem dissolves. His schemes unified geometry and arithmetic. His toposes abstracted the notion of space. His motives sought to unify all cohomologies. At every level, the pattern is the same: patient construction of vast, general theories that reveal the deep simplicity hidden beneath apparent complexity. The “rising sea” continues to rise, and much of twenty-first century algebraic geometry — stacks, derived categories, motivic homotopy theory, perfectoid spaces — flows directly from the springs Grothendieck opened.