04.09.01 · algebraic-geometry / hodge

Hodge decomposition

shipped3 tiersLean: partial

Anchor (Master): Voisin *Hodge Theory* Vol I; Griffiths-Harris; Hodge 1941 *Theory and Applications of Harmonic Integrals*

Intuition [Beginner]

The Hodge decomposition is one of the deepest theorems in geometry: on a compact Kähler manifold, the cohomology splits into pieces indexed by holomorphic type. A degree- $n$ cohomology class breaks into pieces $H^{p, q}$ with $p + q = n$ , where $H^{p, q}$ records " $p$ -holomorphic and $q$ -antiholomorphic" content — like decomposing a polynomial in $z$ and $\overset{z}{ˉ}$ by its bidegree.

W. V. D. Hodge proved this in his 1941 monograph The Theory and Applications of Harmonic Integrals. His method: every cohomology class on a compact Kähler manifold has a unique harmonic representative — a form annihilated by both the Laplacian $Δ$ and the complex Laplacians. Harmonic forms decompose by holomorphic type, hence so does cohomology.

The Hodge decomposition is the bridge between topology (singular cohomology of the underlying real manifold) and complex algebraic geometry (cohomology of holomorphic differential forms). For a smooth complex projective variety, it produces the Hodge diamond, a triangular array of integers $h^{p, q}$ that captures finer information than the Betti numbers alone.

Visual [Beginner]

A compact Kähler manifold with cohomology classes labeled by bidegree $(p, q)$ , organised into the Hodge diamond.

Worked example [Beginner]

For an elliptic curve $E$ (a torus, real dimension 2, complex dimension 1):

The Betti numbers are $b_{0} = 1$ , $b_{1} = 2$ , $b_{2} = 1$ .

The Hodge decomposition refines this: $H^{1} (E; C) = H^{1, 0} (E) \oplus H^{0, 1} (E)$ with $h^{1, 0} = h^{0, 1} = 1$ — one holomorphic 1-form and one anti-holomorphic 1-form, summing to $b_{1} = 2$ .

The Hodge diamond looks like:

h^{1, 0} = 1 h^{0, 0} = 1 h^{1, 1} = 1 h^{0, 1} = 1

For a K3 surface (compact Kähler, complex dimension 2): the Hodge diamond is

h^{2, 0} = 1 h^{1, 0} = 0 h^{2, 1} = 0 h^{0, 0} = 1 h^{1, 1} = 20 h^{2, 2} = 1 h^{0, 1} = 0 h^{1, 2} = 0 h^{0, 2} = 1

with Betti numbers $b_{0} = 1, b_{1} = 0, b_{2} = 22, b_{3} = 0, b_{4} = 1$ .

The diamond carries strictly more information than the Betti numbers — it remembers complex structure.

Check your understanding [Beginner]

Formal definition [Intermediate+]

Let $X$ be a compact Kähler manifold of complex dimension $n$ — a compact complex manifold equipped with a Hermitian metric $h$ whose associated real $(1, 1)$ -form $ω$ is closed: $d ω = 0$ . (In particular every smooth complex projective variety is Kähler, with $ω$ the Fubini-Study form pulled back from a projective embedding.)

Theorem (Hodge decomposition). For a compact Kähler manifold $X$ of complex dimension $n$ , there is a canonical direct-sum decomposition

H^{k} (X; C) = p + q = k ⨁ H^{p, q} (X), H^{p, q} (X) := H^{q} (X; Ω_{X}^{p}),

where $Ω_{X}^{p}$ is the sheaf of holomorphic $p$ -forms. The decomposition satisfies:

(H1) Hodge symmetry: $\overline{H^{p, q}} = H^{q, p}$ , hence $h^{p, q} = h^{q, p}$ .

(H2) Serre symmetry: $H^{p, q} \cong H^{n - p, n - q}^\vee$ , hence $h^{p, q} = h^{n - p, n - q}$ .

(H3) Combined Hodge-diamond symmetry: $h^{p, q} = h^{q, p} = h^{n - p, n - q} = h^{n - q, n - p}$ .

(H4) Compatibility with Betti numbers: $b_{k} (X) = \sum_{p + q = k} h^{p, q}$ .

Hodge numbers and Hodge diamond. Define $h^{p, q} (X) := dim_{C} H^{p, q} (X)$ . The array of Hodge numbers $(h^{p, q})_{0 \leq p, q \leq n}$ is the Hodge diamond — a $(n + 1) \times (n + 1)$ table of non-negative integers with the symmetries (H3).

Examples.

(E1) Projective space $P^{n}$ . $h^{p, q} = 1$ if $p = q$ and $0 \leq p \leq n$ ; $h^{p, q} = 0$ otherwise. The Hodge diamond is a single column of $1$ s on the diagonal. Betti numbers $b_{2 k} = 1, b_{2 k + 1} = 0$ .

(E2) Smooth projective curve of genus $g$ . $h^{0, 0} = h^{1, 1} = 1$ , $h^{1, 0} = h^{0, 1} = g$ . Betti numbers $b_{0} = b_{2} = 1, b_{1} = 2 g$ .

(E3) K3 surface. $h^{0, 0} = h^{2, 2} = 1$ , $h^{2, 0} = h^{0, 2} = 1$ , $h^{1, 1} = 20$ , others $0$ . Total: $b_{0} = b_{4} = 1$ , $b_{1} = b_{3} = 0$ , $b_{2} = 22$ .

(E4) Quintic threefold (Calabi-Yau). $h^{0, 0} = h^{3, 3} = 1$ , $h^{3, 0} = h^{0, 3} = 1$ , $h^{2, 1} = h^{1, 2} = 101$ , $h^{1, 1} = h^{2, 2} = 1$ , others $0$ . The mirror quintic swaps $h^{2, 1} \leftrightarrow h^{1, 1}$ to $1$ and $101$ — mirror symmetry.

Construction via harmonic forms. Choose a Kähler metric on $X$ . The space of harmonic $(p, q)$ -forms

H^{p, q} (X) := {α \in Ω^{p, q} (X) : Δ_{\overset{ˉ}{\partial}} α = 0}

(where $Δ_{\overset{ˉ}{\partial}} = \overset{ˉ}{\partial} \overset{ˉ}{\partial}^{*} + \overset{ˉ}{\partial}^{*} \overset{ˉ}{\partial}$ ) provides a canonical representative for each Dolbeault cohomology class: $H^{p, q} (X) ≅ H^{p, q} (X)$ . The Kähler condition implies $Δ_{d} = 2 Δ_{\overset{ˉ}{\partial}}$ , so harmonic forms for the de Rham Laplacian and the Dolbeault Laplacian coincide. A harmonic $k$ -form decomposes uniquely by bidegree:

H^{k} (X; C) = p + q = k ⨁ H^{p, q} (X) .

This bidegree decomposition descends to cohomology, giving the Hodge decomposition.

Hodge filtration. The decomposition can be repackaged as a decreasing filtration

F^{p} H^{k} (X; C) := p^{'} \geq p ⨁ H^{p^{'}, k - p^{'}} (X),

with $F^{0} = H^{k} (C), F^{k + 1} = 0$ , and $gr_{F}^{p} H^{k} = H^{p, k - p}$ . The pair $(H^{k}, F^{∙})$ together with complex conjugation forms a Hodge structure of weight $k$ — a foundational object in modern Hodge theory.

Algebraic statement: Hodge-de Rham degeneration. For a smooth proper variety $X$ over a field $k$ of characteristic 0, the Hodge spectral sequence $E_{1}^{p, q} = H^{q} (X; Ω_{X}^{p}) \Rightarrow H_{dR}^{p + q} (X / k)$ degenerates at $E_{1}$ . This is an algebraic statement — true over any characteristic- $0$ base. Deligne-Illusie 1987 proved the algebraic version via reduction mod $p$ and lifting.

Key theorem with proof [Intermediate+]

Theorem (Hodge decomposition). Let $X$ be a compact Kähler manifold of complex dimension $n$ . Then for each $0 \leq k \leq 2 n$ :

H^{k} (X; C) = p + q = k ⨁ H^{q} (X; Ω_{X}^{p}) .

Proof outline. Step 1 — Hodge theorem on harmonic forms. On a compact Riemannian manifold $(X, g)$ , the Laplace-Beltrami operator $Δ = d d^{*} + d^{*} d$ is self-adjoint and elliptic, and its kernel — the space of harmonic forms $H^{k} (X)$ — provides a canonical representative for each de Rham cohomology class:

H^{k} (X) ≅ H_{dR}^{k} (X; R) .

This is the Hodge theorem in its real form (Hodge 1941; rigorous PDE proof by Weyl, simplified by Kodaira-Spencer). The proof uses spectral theory of elliptic operators on compact manifolds: $Δ$ is self-adjoint Fredholm, with finite-dimensional kernel and a Hodge orthogonal decomposition $Ω^{k} (X) = H^{k} (X) \oplus im (d) \oplus im (d^{*})$ .

Step 2 — Kähler identities. On a Kähler manifold, the complex structure $J$ and the Kähler form $ω$ satisfy the Kähler identities:

[Λ, \overset{ˉ}{\partial}] = - i \partial^{*}, [Λ, \partial] = i \overset{ˉ}{\partial}^{*},

where $Λ$ is contraction with $ω$ and $\partial^{*}, \overset{ˉ}{\partial}^{*}$ are the formal adjoints. Combining gives:

Δ_{d} = 2 Δ_{\partial} = 2 Δ_{\overset{ˉ}{\partial}} .

So a form is $d$ -harmonic iff $\partial$ -harmonic iff $\overset{ˉ}{\partial}$ -harmonic. The triple coincidence is the defining property of Kähler.

Step 3 — Bidegree splitting. Since $Δ_{\overset{ˉ}{\partial}}$ commutes with the bidegree operators (the projection onto $Ω^{p, q}$ ), the space $H^{k} (X) = ⨁_{p + q = k} H^{p, q} (X)$ decomposes into bidegrees. Each summand $H^{p, q} (X)$ is the space of harmonic $(p, q)$ -forms.

Step 4 — Identification with Dolbeault. The Dolbeault theorem (analogous to de Rham): $H^{q} (X; Ω_{X}^{p}) ≅ H_{\overset{ˉ}{\partial}}^{p, q} (X) ≅ H^{p, q} (X)$ . So $H^{p, q} (X) ≅ H^{q} (X; Ω_{X}^{p}) = H^{p, q} (X)$ .

Step 5 — Combining. From Steps 1–4: $H^{k} (X; C) ≅ H^{k} (X) = ⨁_{p + q = k} H^{p, q} (X) ≅ ⨁_{p + q = k} H^{q} (X; Ω_{X}^{p})$ .

The decomposition is canonical (independent of the choice of Kähler metric): the Hodge components $H^{p, q} (X)$ are determined by the complex structure alone, not the metric. Different Kähler metrics give different harmonic representatives, but the same cohomology subspaces. $□$

The Kähler identities are the load-bearing input. They reduce three Laplacians to one, forcing the bidegree splitting on harmonic forms — which is what the cohomology decomposition is.

Bridge. The construction here builds toward later units of the strand, where the same pattern is taken up at higher structure. 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 5 (medium, conceptual).

Why does Hodge theory require the Kähler condition rather than just compact complex structure?

Hint

The Kähler identities $Δ_{d} = 2 Δ_{\overset{ˉ}{\partial}}$ are essential for the harmonic-form decomposition.

Answer

The Hodge decomposition relies on the Kähler identities: the relations $Δ_{d} = 2 Δ_{\overset{ˉ}{\partial}} = 2 Δ_{\partial}$ between the Riemannian, Dolbeault, and conjugate-Dolbeault Laplacians.

These identities require the Kähler form $ω$ to be closed ( $d ω = 0$ ). On a compact complex manifold without the Kähler condition, the three Laplacians can be different operators — a $(p, q)$ -form can be $d$ -harmonic without being $\overset{ˉ}{\partial}$ -harmonic, breaking the bidegree splitting.

Counterexamples.

Hopf surface $S^{1} \times S^{3}$ with complex structure: compact complex 2-manifold but not Kähler. $h^{1, 0} (S^{1} \times S^{3}) = 0$ but $h^{0, 1} (S^{1} \times S^{3}) = 1$ . So $h^{1, 0} \neq = h^{0, 1}$ — Hodge symmetry fails.
Iwasawa manifold (quotient of nilpotent Heisenberg group): compact complex 3-manifold, not Kähler. $b_{1} = 4$ but $h^{1, 0} + h^{0, 1} = 5 \neq = 4$ . Not even compatible with Betti.

The Kähler condition forces enough rigidity (the closed (1,1)-form is a Riemannian-Hermitian compatibility) to make all three Laplacians coincide. Without it, Hodge theory collapses.

The algebraic counterpart: smooth projective varieties over $C$ are automatically Kähler (the Fubini-Study form pulls back), so Hodge theory always holds for them. Smooth proper non-projective complex varieties may not be Kähler — Moishezon manifolds with non-vanishing first Chern class are an example. $□$

Exercise 6 (hard, proof).

Outline Deligne-Illusie's algebraic proof of Hodge degeneration via reduction mod $p$ .

Hint

Lift mod $p^{2}$ , decompose $Ω^{∙}$ via Frobenius, and use Cartier isomorphism.

Answer

Deligne-Illusie 1987 (Invent. Math. 89, 247–270). For $X$ smooth proper over a field $k$ of characteristic 0, the Hodge spectral sequence $E_{1}^{p, q} = H^{q} (X; Ω^{p}) \Rightarrow H_{dR}^{p + q} (X / k)$ degenerates at $E_{1}$ .

Strategy.

Reduce to char $p$ . By spread-out, $X$ is the generic fibre of a smooth proper $X \to S$ with $S = Spec (R)$ , $R$ a finitely generated $Z$ -algebra. Hodge degeneration on $X$ follows from Hodge degeneration on geometric fibres $X_{s} = X_{s}$ over closed points $s$ of $S$ . Closed points of $Spec (R)$ have residue fields of finite characteristic.
Lift to char $p^{2}$ . For a smooth proper variety $X_{0}$ over a perfect field $k$ of characteristic $p$ with $dim X_{0} \leq p$ , suppose $X_{0}$ admits a flat lift $\tilde{X}$ to $W_{2} (k)$ (the second Witt vectors). Such lifts exist when $X_{0}$ is "geometrically nice" (e.g., projective with no obstructions).
Cartier isomorphism + Frobenius decomposition. The Frobenius $F : X_{0} \to X_{0}^{(p)}$ has the Cartier isomorphism $C^{- 1} : Ω_{X_{0}^{(p)} / k}^{∙} ≅ H^{∙} (F_{*} Ω_{X_{0} / k}^{∙})$ — the cohomology sheaves of the pushforward of the de Rham complex are the differential forms on the Frobenius twist.

The lift to $W_{2}$ allows constructing a splitting of the Frobenius pushforward as a complex: $τ_{\leq p} (F_{*} Ω_{X_{0} / k}^{∙}) ≃ ⨁_{i \leq p} Ω_{X_{0}^{(p)} / k}^{i} [- i]$ (in the derived category).

Degeneration consequence. This splitting forces the Hodge spectral sequence to degenerate: the $E_{1}$ differential $H^{q} (Ω^{p}) \to H^{q} (Ω^{p + 1})$ is identified with a Frobenius-twisted operator that vanishes by the splitting.
Lift back to char 0. By semicontinuity (constructibility of cohomology dimensions), Hodge degeneration on the geometric special fibre $X_{s}$ in char $p$ implies degeneration on $X = X_{η}$ in char 0.

The proof avoids Hodge theory entirely — no harmonic forms, no PDE, no analysis. It is purely algebraic / arithmetic, and works over any characteristic-0 base. The Kähler hypothesis (which is automatic for smooth projective $X$ ) is replaced by the lifting hypothesis, satisfied in essentially all cases of interest.

This proof had a major impact: it gave the first algebraic proof of Hodge degeneration, opened up Hodge theory in mixed characteristic, and inspired p-adic Hodge theory (Fontaine, Faltings, Tsuji). $□$

Exercise 7 (hard, conceptual).

Define a polarised Hodge structure of weight $k$ and explain its role in Hodge theory.

Hint

Polarisation = bilinear form satisfying Hodge-Riemann bilinear relations.

Answer

A Hodge structure of weight $k$ on a finite-dimensional $Q$ -vector space $V_{Q}$ is:

A direct-sum decomposition $V_{C} = V_{Q} \otimes_{Q} C = ⨁_{p + q = k} V^{p, q}$ with $\overline{V^{p, q}} = V^{q, p}$ .

Equivalently, a decreasing filtration $F^{∙}$ on $V_{C}$ with $V^{p, q} = F^{p} \cap \overline{F^{q}}$ .

A polarisation of a Hodge structure of weight $k$ is a $Q$ -bilinear form $Q : V_{Q} \times V_{Q} \to Q$ that is:

$(- 1)^{k}$ -symmetric (symmetric for $k$ even, antisymmetric for $k$ odd).
Satisfies the Hodge-Riemann bilinear relations:

(HR1) $Q (V^{p, q}, V^{p^{'}, q^{'}}) = 0$ if $(p, q) \neq = (k - p^{'}, k - q^{'})$ . (HR2) $i^{p - q} Q (α, \overset{α}{ˉ}) > 0$ for $α \in V^{p, q} ∖ 0$ .

The pair $(V_{Q}, Q)$ with the Hodge structure is a polarised Hodge structure of weight $k$ .

Examples.

Curve cohomology. For a smooth projective curve $C$ , $H^{1} (C; Q)$ has a polarised Hodge structure of weight 1 with polarisation given by the cup product $\int_{C} α \cup β$ .
Surface intersection form. $H^{2} (X; Q)$ on a smooth projective surface has a polarised Hodge structure of weight 2; the polarisation is the cup product followed by integration.

Role in Hodge theory.

Variation of Hodge structure. A family of smooth projective varieties $X \to S$ produces a variation of polarised Hodge structure over $S$ — a system of Hodge structures on $H^{k} (X_{s})$ varying with $s$ , satisfying Griffiths transversality. Period domains parametrise polarised Hodge structures; the period map $S \to period domain$ encodes the variation.
Mixed Hodge theory. For non-compact or singular varieties, Deligne (1971-74) introduced mixed Hodge structures — filtered Hodge structures with weight filtration and Hodge filtration interacting. Foundational for Hodge theory on non-projective varieties.
Hodge conjecture. Asks whether classes in $H^{p, p} (X) \cap H^{2 p} (X; Q)$ on a smooth projective variety are algebraic — i.e., $Q$ -linear combinations of fundamental classes of subvarieties of codimension $p$ . Open in general; one of the Millennium Prize Problems.
Modularity. Polarised Hodge structures of weight 1 with $h^{1, 0} = 1$ (elliptic-curve type) classify elliptic curves. The modularity theorem (Wiles-Taylor 1995-99) and its predecessors connect polarised Hodge structures to automorphic forms via Galois representations.

The polarised Hodge structure is the modern algebraic-geometric incarnation of the classical Riemann period relations on Riemann surfaces — Riemann's 1857 imposition that the period matrix of a curve must satisfy specific symmetry and positivity conditions. $□$

Lean formalization [Intermediate+]

lean_status: partial — Mathlib has differential forms, partial de Rham cohomology infrastructure, and complex-manifold structure; the full Hodge theorem for compact Kähler manifolds is not yet a named theorem.

[object Promise]

Advanced results [Master]

Hodge-Riemann bilinear relations. For a compact Kähler manifold $X$ of dimension $n$ with Kähler class $ω \in H^{1, 1} (X)$ , the primitive cohomology $P^{p, q} = ker (Λ) \cap H^{p, q}$ admits a positive-definite Hermitian form $H_{ω} (α, β) = i^{p - q} (- 1)^{(p + q) (p + q - 1) /2} \int_{X} ω^{n - p - q} \land α \land \overset{ˉ}{β}$ . This positivity is the source of the Lefschetz hyperplane theorem and the hard Lefschetz theorem.

Hard Lefschetz theorem. For a compact Kähler manifold of dimension $n$ with Kähler class $ω$ , the cup product map $ω^{k} \cup : H^{n - k} (X) \to H^{n + k} (X)$ is an isomorphism for $0 \leq k \leq n$ . This produces the Lefschetz $sl_{2}$ -decomposition of cohomology and the primitive components.

Hodge index theorem. On a smooth projective surface $X$ , the intersection form on $H^{2} (X; R)$ has signature $(1, h^{1, 1} - 1)$ on the (1, 1)-component. This is foundational for surface theory.

Mixed Hodge structures. Deligne (1971-74, Théorie de Hodge II, III) extended Hodge theory to all complex algebraic varieties — including non-compact, non-smooth — via mixed Hodge structures: a triple $(W_{∙}, F^{∙}, V_{Q})$ where $W_{∙}$ is the weight filtration and $F^{∙}$ the Hodge filtration, with $gr_{W}^{k}$ a pure Hodge structure of weight $k$ .

Variations of Hodge structure. For a family of smooth projective varieties $π : X \to S$ over a smooth base, the cohomology $H^{k} (X_{s}; Q)$ varies in $s \in S$ as a variation of polarised Hodge structure — a local system equipped with a varying Hodge filtration satisfying Griffiths transversality. The period map $S \to period domain /Γ$ encodes the variation. Schmid's nilpotent-orbit and SL $_{2}$ -orbit theorems describe degenerations.

Hodge conjecture. Open since 1950: for a smooth projective variety $X$ , every class in $H^{p, p} (X) \cap H^{2 p} (X; Q)$ is algebraic (a $Q$ -linear combination of cohomology classes of codimension- $p$ subvarieties). One of the Clay Millennium Prize Problems. Special cases known: codimension 1 (Lefschetz 1+1+1, 1924), abelian varieties (Hodge for codim-1 follows from polarisation), uniruled threefolds (Conte-Murre).

Hodge theory in derived category. Saito's theory of mixed Hodge modules (1988-90) — a sheaf-theoretic refinement of mixed Hodge structures, with functors and a six-functor formalism. Foundational for modern Hodge-theoretic perspectives on motives and intersection cohomology.

Synthesis. This construction generalises the pattern fixed in 04.03.01 (sheaf cohomology), 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]

The Hodge decomposition is sketched in the formal-definition section. The Kähler identities ( $Δ_{d} = 2 Δ_{\overset{ˉ}{\partial}}$ ) are proved in Voisin Vol I §6.1 or Griffiths-Harris Ch. 0; they require the local Kähler model: at each point, choose holomorphic coordinates $z_{i}$ with $g_{i \overset{ˉ}{j}} = δ_{ij} + O (∣ z ∣^{2})$ — the Kähler condition forces this enhanced local structure.

Hodge degeneration in characteristic 0 (Deligne-Illusie 1987): proved via reduction mod $p$ + Frobenius decomposition, sketched in Exercise 6.

Hard Lefschetz theorem: proved via the Hodge-Riemann bilinear relations (positivity argument), Voisin Vol I §6.2.

Hodge conjecture: open in general; proved in special cases — see Voisin Vol II for current status.

Connections [Master]

Sheaf cohomology 04.03.01 — $H^{p, q} (X) = H^{q} (X; Ω_{X}^{p})$ is sheaf cohomology of differential forms.
De Rham cohomology 03.04.06 — Hodge decomposition refines de Rham cohomology by holomorphic type.
Sheaf of differentials 04.08.01 — $Ω^{p}$ are exterior powers of $Ω^{1}$ .
Canonical sheaf 04.08.02 — $ω_{X} = Ω^{n}$ is the top component, contributing $H^{n, q}$ to the Hodge decomposition.
Serre duality 04.08.03 — gives Hodge symmetry $H^{p, q} ≅ H^{n - p, n - q \lor}$ .
Kodaira vanishing 04.09.02 — uses Hodge decomposition + positivity.
Period matrix 06.06.02 — Hodge decomposition on a Riemann surface is the period matrix data.
Theta function 06.06.05 — sections of theta-bundles realise specific Hodge classes on abelian varieties.

Historical & philosophical context [Master]

W. V. D. Hodge's 1941 monograph The Theory and Applications of Harmonic Integrals (Cambridge University Press; 2nd ed. 1952) introduced the harmonic-integral approach to cohomology of complex manifolds. Hodge — born 1903 in Edinburgh, working at Cambridge — proved that on a compact orientable Riemannian manifold, every de Rham cohomology class has a unique harmonic representative; on a compact Kähler manifold, the harmonic representatives split by holomorphic-antiholomorphic bidegree, giving the Hodge decomposition.

The 1941 proof used a combination of variational methods and PDE theory. Hodge's approach: minimise the Dirichlet energy $\int_{X} ∣ α ∣^{2}$ over forms in a given cohomology class, finding a unique critical point — the harmonic representative. The minimisation was technically delicate (compactness of the underlying space, regularity of solutions) and Hodge's proof had gaps. Hermann Weyl rewrote the proof using rigorous Hilbert-space methods (1943); Kodaira-Spencer further simplified using elliptic-operator theory (1953); the modern presentation uses the spectral theory of self-adjoint Fredholm operators on compact manifolds.

The Kähler condition is essential: Erich Kähler's 1933 paper on Hermitian metrics with $d ω = 0$ (Math. Z. 38, 472–500) introduced the metric structure that would bear his name. Hodge realised that this condition is exactly what makes the three Laplacians ( $d, \partial, \overset{ˉ}{\partial}$ ) coincide — Kähler identities — and hence what makes the bidegree splitting carry over from forms to cohomology.

The algebraic interpretation came later. Pierre Deligne's Théorie de Hodge II–III (Publ. Math. IHES 40 (1971), 5–57; 44 (1974), 5–77) extended Hodge theory to non-compact and singular complex algebraic varieties via mixed Hodge structures. Deligne combined the Hodge filtration with a weight filtration to produce a refined structure; the Hodge conjecture (Hodge's 1950 ICM problem) and the Standard Conjectures of Grothendieck became central problems in the new framework. Deligne-Illusie 1987 (Invent. Math. 89, 247–270) gave the first purely algebraic proof of Hodge degeneration via reduction mod $p$ — a proof that worked in characteristic 0 without any analysis, opening up p-adic Hodge theory (Fontaine, Faltings, Tsuji, Berger, Scholze).

The modern Hodge-theoretic landscape includes: variations of Hodge structure over moduli spaces (Griffiths 1968-72; Schmid 1973), mixed Hodge modules (Saito 1988-90), non-abelian Hodge theory (Hitchin, Simpson, 1980s-90s), and p-adic Hodge theory via Scholze's perfectoid spaces (2010s).

Hodge theory remains a defining framework of modern algebraic geometry. Riemann's 1857 period relations on Riemann surfaces, Klein and Poincaré's 1880s analytic Riemann-surface theory, Lefschetz's 1920s topological methods, and Hodge's 1941 harmonic-integral synthesis all converge in the modern theory of polarised Hodge structures and their variations.

Bibliography [Master]

Hodge, W. V. D., The Theory and Applications of Harmonic Integrals, Cambridge University Press 1941; 2nd ed. 1952.
Voisin, C., Hodge Theory and Complex Algebraic Geometry, I & II, Cambridge University Press 2002–03.
Griffiths, P. & Harris, J., Principles of Algebraic Geometry, Wiley 1978 — Chapter 0.
Wells, R. O., Differential Analysis on Complex Manifolds, Springer GTM 65, 1980.
Deligne, P., Théorie de Hodge I, II, III, ICM 1970; Publ. Math. IHES 40 (1971), 5–57; 44 (1974), 5–77.
Deligne, P. & Illusie, L., Relèvements modulo $p^{2}$ et décomposition du complexe de de Rham, Invent. Math. 89 (1987), 247–270.
Griffiths, P., Periods of integrals on algebraic manifolds I, II, III, Amer. J. Math. 90 (1968); Publ. Math. IHES 38 (1970).
Schmid, W., Variation of Hodge structure: the singularities of the period mapping, Invent. Math. 22 (1973), 211–319.
Saito, M., Mixed Hodge modules, Publ. RIMS Kyoto Univ. 26 (1990), 221–333.
Kähler, E., Über eine bemerkenswerte Hermitische Metrik, Abh. Math. Sem. Univ. Hamburg 9 (1933), 173–186.
Kodaira, K., On a differential-geometric method in the theory of analytic stacks, Proc. Nat. Acad. Sci. USA 39 (1953), 1268–1273.

Prerequisites

04.03.01
03.04.06

Tier anchors

beginner: Strogatz-style: cohomology splits into holomorphic and anti-holomorphic pieces
intermediate: Voisin Vol I Ch. 6 (Fast Track 3.27); Griffiths-Harris Ch. 0; Wells *Differential Analysis on Complex Manifolds*
master: Voisin *Hodge Theory* Vol I; Griffiths-Harris; Hodge 1941 *Theory and Applications of Harmonic Integrals*

References

TODO_REF
Hodge — The Theory and Applications of Harmonic Integrals · Cambridge 1941, 2nd ed. 1952
TODO_REF
Voisin — Hodge Theory and Complex Algebraic Geometry · Vol I Ch. 6, Cambridge 2002
TODO_REF
Griffiths-Harris — Principles of Algebraic Geometry · Ch. 0
TODO_REF
Wells — Differential Analysis on Complex Manifolds · Springer GTM 65

Lean module

Codex.AlgebraicGeometry.Hodge.HodgeDecomposition

Reviewer

TBD

Estimated time

beginner: 12m
intermediate: 40m
master: 70m