Alexander Grothendieck

Alexander Grothendieck (1928–2014) was a German-born, stateless-then-French mathematician who reshaped algebraic geometry by founding the theory of schemes, topoi, and cohomological methods. Key result in words: he recast geometry as the study of commutative rings and their functorial spectra, providing the framework (étale cohomology, motives, the standard conjectures) within which the Weil conjectures were proved. Where it applies: algebraic and arithmetic geometry, homological algebra, category theory, number theory. Who/when: principally at the IHÉS, Paris, 1958–1970, via the monumental EGA and SGA seminars; Fields Medal 1966.

Alexander Grothendieck was the central architect of mid-20th-century algebraic geometry, transforming a classical subject of polynomial equations into a vast structural edifice built on category theory and sheaf cohomology. His central innovation, the scheme, generalizes the notion of an algebraic variety by identifying geometric objects with arbitrary commutative rings, allowing nilpotents, mixed characteristic, and arithmetic to enter geometry on equal footing. The importance of this work lies in its provision of a language — relative, functorial, cohomological — flexible enough that deep arithmetic statements (the Weil conjectures, finiteness theorems, duality) could be formulated and proved. A surprising fact: Grothendieck, arguably the most influential mathematician of his century, largely abandoned mathematics after 1970 for political activism and eventually total reclusion, dying in 2014 in a Pyrenean village having forbidden the reproduction of his later writings.

The Core Insight

Geometry is the dual of commutative algebra, done relatively and functorially. Instead of studying a space directly, study the category of all sheaves, rings, and morphisms over it, and let universal properties — representable functors, adjoint functors, derived functors — do the geometric work. A scheme

X

is not primarily a set of points but a functor

T \mapsto Hom (T, X)

on the category of rings; cohomology is not a gadget attached to a space but a derived functor on a category of sheaves. This "relativization over a base" and "everything is a functor" philosophy is Grothendieck's signature contribution and the source of the field's modern power.

Formal Statement

The mathematical core of Grothendieck's program can be stated as a chain of definitions building the scheme and its cohomology.

Affine Scheme

Let $A$ be a commutative ring with unit. The prime spectrum $Spec A$ is the topological space whose points are the prime ideals $p \subset A$ , with the Zariski topology generated by closed sets

V (I) = {p \in Spec A : I \subseteq p}, I \subseteq A an ideal .

⟳closed sets must be closed under arbitrary intersection

V (\sum I_{α}) = ⋂ V (I_{α})

and finite union

V (I) \cup V (J) = V (I \cap J)

; check these from the prime-ideal definition.

Equipped with the structure sheaf $O_{Spec A}$ , whose sections on the basic open $D (f) = {p : f \in / p}$ are the localization $A_{f}$ , the pair $(Spec A, O_{Spec A})$ is an affine scheme. A scheme is a locally ringed space locally isomorphic to affine schemes.

Functor of Points

Every scheme $X$ over a base $S$ determines a contravariant functor on $S$ -schemes,

h_{X} (T) = Hom_{S} (T, X),

⟳this is the Yoneda embedding

X \mapsto h_{X}

restricted to schemes; functoriality is composition of morphisms.

and by the Yoneda lemma $X$ is determined up to unique isomorphism by $h_{X}$ . Representable functors are thus identified with their representing schemes, the basis of moduli theory.

Derived Functor Cohomology

For a sheaf $F$ of abelian groups on a scheme $X$ , the sheaf cohomology groups are the right derived functors of the global-sections functor $Γ (X, -)$ :

H^{i} (X, F) = R^{i} Γ (X, F) = H^{i} (Γ (X, I^{∙})),

⟳embed

F

in an injective resolution

0 \to F \to I^{∙}

in the abelian category of sheaves (which has enough injectives), apply

Γ

, and take cohomology; independence of resolution follows from the comparison theorem for derived functors.

These satisfy the long exact sequence, are computed for quasi-coherent $F$ on affine $X$ to vanish in positive degree ( $H^{i > 0} (Spec A, M) = 0$ ), and underlie Serre duality, flat/étale cohomology, and the Grothendieck–Riemann–Roch theorem.

Grothendieck–Riemann–Roch

For a proper morphism $f : X \to Y$ of smooth quasi-projective schemes and a coherent sheaf (or element of K-theory) $E$ on $X$ ,

ch (f_{!} E) \cdot td (T_{Y}) = f_{*} (ch (E) \cdot td (T_{X})),

⟳reduce to the case of a closed immersion and a projection

P^{n} \times Y \to Y

; the Todd class corrects the failure of the Chern character to commute with pushforward, the discrepancy being measured by the relative tangent complex.

where $ch$ is the Chern character, $td$ the Todd class, $f_{!}$ the K-theoretic pushforward (alternating sum of $R^{i} f_{*}$ ), and $f_{*}$ the homological pushforward on Chow groups. This relativizes Hirzebruch–Riemann–Roch.

Why It Holds / Derivation

The spectrum is a faithful functor — sketch

We sketch why $A \mapsto Spec A$ gives an anti-equivalence between commutative rings and affine schemes.

Points encode ideals. A ring homomorphism $φ : A \to B$ pulls back primes: $q \mapsto φ^{- 1} (q)$ is prime, giving a continuous map $Spec φ : Spec B \to Spec A$ .

Localization recovers structure sheaf. On the basis ${D (f)}$ one sets $O (D (f)) = A_{f}$ ; compatibility on overlaps $D (f) \cap D (g) = D (f g)$ follows from the universal property of localization, so the presheaf sheafifies canonically.

Global sections recover the ring. One proves $Γ (Spec A, O) = A$ , so the global-sections functor is a quasi-inverse on the affine level.

Morphisms correspond. Every morphism of locally ringed spaces $Spec B \to Spec A$ arises from a unique ring map $A \to B$ (locality of the morphism on stalks forces this). Hence

Hom_{Sch} (Spec B, Spec A) ≅ Hom_{Ring} (A, B) .

⟳the stalk at

p

is the local ring

A_{p}

; a morphism of locally ringed spaces must send the maximal ideal into the maximal ideal, which pins down the ring map uniquely.

This anti-equivalence is the rigorous form of "geometry = dual of algebra," and gluing affines yields the full category of schemes. $■$

Alexander Grothendieck | 1928–2014

Born 28 March 1928 in Berlin to anarchist parents (his father Alexander "Sascha" Shapiro perished at Auschwitz), Grothendieck grew up stateless and refugee. After early functional-analysis work (topological tensor products, nuclear spaces — his thesis under Laurent Schwartz and Jean Dieudonné) he turned to algebraic geometry at the newly founded IHÉS in 1958. There, with Dieudonné as scribe, he produced the Éléments de géométrie algébrique (EGA) and led the legendary Séminaire de géométrie algébrique (SGA), a collaborative output of tens of thousands of pages. He received the Fields Medal in 1966 (boycotting the Moscow ICM for political reasons). From 1970 he withdrew — first into ecological-pacifist activism (the Survivre et Vivre group), then into increasingly solitary mathematical-mystical writing (Récoltes et Semailles, La Clef des Songes). He resigned from organized mathematics, eventually living in seclusion in Lasserre, Ariège, where he died on 13 November 2014.

Building Intuition

Think of a commutative ring $A$ as the ring of "functions" on an as-yet-unseen space; Grothendieck's reversal is to define the space as whatever has $A$ as its functions. The points are prime ideals because primality is exactly the condition that makes the quotient an integral domain — a "function-like" object with no zero-divisors, i.e. a generic point. Crucially, schemes allow nilpotents ( $Spec k [ϵ] / ϵ^{2}$ is a point "with a tangent direction"), so infinitesimal and arithmetic phenomena become geometric. The second pillar is relativity: never study $X$ alone but always a morphism $X \to S$ , treating $X$ as a family of fibers over the base $S$ ; properties like properness, smoothness, and flatness are properties of the map. The third is functoriality: rather than asking "what is the curve?" ask "what does the curve represent?" — i.e. which functor of test schemes it classifies — which is how moduli spaces and representability arise naturally.

Spec of a ring: generic and closed points, with the Zariski topology

Worked Examples

Cohomology of a line bundle on $\mathbb{P}^1$

Given:

X = P_{k}^{1}

over a field

k

, the line bundle

O (d)

d \in Z

. Steps: Use the standard Čech computation on the cover

U_{0} = D (x_{0}), U_{1} = D (x_{1})

. The result, a computation of

H^{0}

and

H^{1}

$H^{0} (P^{1}, O (d)) = {$ homogeneous polynomials of degree $d}$ , dimension $d + 1$ for $d \geq 0$ , else $0$ .
$H^{1} (P^{1}, O (d))$ has dimension $- d - 1$ for $d \leq - 2$ , else $0$ (by Serre duality $H^{1} (O (d)) ≅ H^{0} (O (- 2 - d))^{\lor}$ since $ω = O (- 2)$ ).

Check with Riemann–Roch:

χ = h^{0} - h^{1} = de g + 1 - g = d + 1 - 0 = d + 1

h^{0} (O (d)) = max (d + 1, 0), h^{1} = max (- d - 1, 0), χ = d + 1

GRR for a curve mapping to a point

Given:

f : X \to Spec k

X

a smooth projective curve of genus

g

, and

E = L

a line bundle of degree

d

. Here

Y = Spec k

is a point,

td (T_{Y}) = 1

. Steps: GRR equation [4] gives

ch (f_{!} L) = f_{*} (ch (L) \cdot td (T_{X}))

. On a curve

ch (L) = 1 + c_{1} (L)

and

td (T_{X}) = 1 + \frac{1}{2} c_{1} (T_{X}) = 1 - \frac{1}{2} K_{X}

(since

c_{1} (T_{X}) = - K_{X}

de g K_{X} = 2 g - 2

). The degree-

1

part pushed forward is the integral over

X

χ (L) = \int_{X} (c_{1} (L) - \frac{1}{2} K_{X}) = d - \frac{1}{2} (2 g - 2) = d + 1 - g .

Answer:

χ (X, L) = de g L + 1 - g = d + 1 - g

recovering the classical Riemann–Roch theorem as the GRR special case over a point.

Common Misconceptions

A scheme is not just "a variety with weird points." Three frequent errors: (1) Treating

Spec A

as merely the set of maximal ideals — the generic points (non-maximal primes) carry essential topological information and are not optional. (2) Believing the Zariski topology behaves like a manifold topology — it is not Hausdorff, is generally not even

T_{1}

, and is far too coarse for analytic/topological intuition (hence the need for the étale topology). (3) Thinking nilpotents are pathological; in fact

Spec k [ϵ] / (ϵ^{2})

is precisely the right object to encode tangent vectors and deformation theory, and discarding it ruins moduli theory. Zariski cohomology fails to be the "right" cohomology for arithmetic. One might hope ordinary Zariski sheaf cohomology with constant coefficients

\underline{Z}

would yield Betti-like numbers and prove the Weil conjectures. It does not: on an irreducible scheme the Zariski topology is irreducible, so the constant sheaf is flasque and

H^{i > 0} (X, \underline{Z}) = 0

— completely trivial, giving no topological information. This vanishing is exactly why Grothendieck had to invent the étale topology (a Grothendieck topology, not a topology on a point set), where covers are étale maps rather than open immersions, recovering

ℓ

-adic cohomology groups

H_{\overset{e}{ˊ} t}^{i} (X, Q_{ℓ})

with the expected Betti numbers and Galois action.

Connections

Grothendieck's framework is the trunk from which much modern mathematics branches. The étale topology generalizes to the abstract notion of a Grothendieck Topology and thence to Topos Theory, a setting unifying geometry and logic. His derived-functor viewpoint is the foundation of Homological Algebra and its modern refinement Derived Categories (introduced by his student Verdier). The Weil Conjectures were the guiding goal — proved via étale cohomology (Grothendieck, M. Artin) and the final Riemann-hypothesis analogue by Pierre Deligne. His vision of a universal cohomology theory crystallized as the still-conjectural theory of Motives, governed by the Standard Conjectures. The functor-of-points philosophy directly produces Moduli Spaces and Stacks, and his Esquisse d'un Programme launched Anabelian Geometry and the study of the absolute Galois Group via dessins d'enfants.

Real-World Applications

Though purely foundational, Grothendieck's machinery underpins much applicable mathematics:

Cryptography & coding theory — Algebraic curves over finite fields, counted via étale cohomology, give elliptic-curve cryptography (ECC) and Goppa/algebraic-geometry codes with optimal parameters.
The proof of Fermat's Last Theorem — Wiles and Taylor's modularity argument runs on schemes, Galois representations, and deformation rings, all Grothendieck-era technology.
Number-theoretic algorithms — Point-counting on curves (Schoof–Elkies–Atkin) and the structure of $L$ -functions rest on $ℓ$ -adic cohomology and the Weil conjectures.
Mathematical physics — Mirror symmetry, topological field theory, and the geometric Langlands program are formulated with derived categories of (quasi-)coherent sheaves on schemes and stacks.
Functional analysis / quantum theory — His early work on nuclear spaces and topological tensor products is standard in distribution theory and quantum field theory.
Optimization & inequalities — The Grothendieck inequality from his Banach-space period is central in theoretical computer science (semidefinite-programming relaxations, the SDP gap of Max-Cut-type problems).
Statistics / data — Tensor-norm and operator-ideal theory descending from his thesis informs random-matrix and high-dimensional probability estimates.

Historical Development

1928 — Born in Berlin to anarchist parents; childhood marked by statelessness and World War II internment camps.
1949–1953 — Doctoral work under Schwartz and Dieudonné on functional analysis: nuclear spaces, the Produits tensoriels topologiques memoir, and the Grothendieck inequality.
1957 — The Tôhoku paper (Sur quelques points d'algèbre homologique) recasts sheaf cohomology via abelian categories with enough injectives, founding modern homological algebra; the same year, the Grothendieck–Riemann–Roch theorem is announced.
1958 — Joins the new IHÉS; outlines the scheme program at the ICM in Edinburgh.
1960–1967 — EGA (Éléments de géométrie algébrique), with Dieudonné, lays scheme-theoretic foundations across four published parts.
1960–1969 — The SGA seminars (SGA 1–7) develop fundamental groups, étale and ℓ-adic cohomology, Grothendieck topologies, topoi, and intersection theory; M. Artin, Deligne, Verdier, Berthelot and others contribute.
1966 — Awarded the Fields Medal at the Moscow ICM, which he declines to attend in protest of Soviet policy.
1970 — Leaves the IHÉS over its partial military funding; founds the ecological group Survivre et Vivre; effectively departs mainstream mathematics.
1980–1991 — Writes La Longue Marche à travers la théorie de Galois, Esquisse d'un Programme, À la Poursuite des Champs (pursuing stacks), and the autobiographical Récoltes et Semailles; from 1991 lives in total seclusion.
2014 — Dies on 13 November in Saint-Lizier/Lasserre, Ariège, France.

Subtleties & Limits

The deepest unfinished part of Grothendieck's vision is the theory of motives and its governing Standard Conjectures (of Lefschetz and Hodge type). He proposed that all reasonable cohomology theories (Betti, de Rham, $ℓ$ -adic, crystalline) should factor through a universal abelian category of motives, with comparison isomorphisms encoding all of arithmetic geometry. Proving that the category of pure motives is abelian semisimple, and establishing the Standard Conjecture D (numerical equivalence = homological equivalence) and the algebraicity of the Künneth/Lefschetz operators, remains open in general; these would, among other things, give a motivic proof of the Weil conjectures independent of Deligne's. Related open territory includes the Section Conjecture of anabelian geometry, the precise formulation and existence of a category of mixed motives with the expected $t$ -structure, and a fully developed theory of Motivic Cohomology realizing his "pursuing stacks" higher-categorical program — only partially achieved through later work on higher topoi and homotopy theory.