Euler characteristic

Intuition [Beginner]

The Euler characteristic is a single integer that captures the shape of a space at the coarsest level. For a polyhedron, you count vertices, subtract edges, add faces, and the answer for any convex polyhedron is two. A cube gives $8-12+6=2$. A tetrahedron gives $4-6+4=2$. The number does not depend on which polyhedron you chose, only on the topology of the underlying surface — in this case, the sphere.

The reason the answer is the same for every convex polyhedron is that any two of them can be deformed into one another without tearing or gluing. The Euler characteristic ignores everything that survives such deformation. It sees only the shape itself, not the way you happened to chop it up.

For a doughnut surface (a torus), the same recipe gives zero. For a two-holed doughnut, it gives minus two. Each new hole subtracts two from the count. The Euler characteristic measures, in this informal sense, the failure of a surface to be a sphere.

Visual [Beginner]

A schematic showing three closed surfaces side by side: a sphere with a triangulation labelled $V-E+F=2$, a torus with a triangulation labelled $V-E+F=0$, and a genus-two surface labelled $V-E+F=-2$. Below each picture, a small caption indicates the genus and the resulting Euler characteristic.

The picture conveys the essential message: counting cells with alternating signs gives a number that depends only on the surface, not on the triangulation. Each handle attached to a sphere reduces the count by two.

Worked example [Beginner]

Compute the Euler characteristic of the surface of a cube and of the surface of a doughnut, using the cellular recipe.

Step 1. The surface of a cube has six square faces. Count the pieces: 8 vertices (corners), 12 edges, 6 faces. Apply the recipe: $8-12+6=2$. The Euler characteristic of the cube surface is 2.

Step 2. Now subdivide each square face by drawing a diagonal. The number of vertices stays at 8 (no new corners on the original cube), but each face becomes two triangles, doubling the face count to 12. Each diagonal is a new edge, adding 6 to the edge count, giving 18. Apply the recipe: $8-18+12=2$. The answer is the same.

Step 3. The surface of a doughnut, drawn as a square with opposite sides glued, has 1 vertex (all four corners identify), 2 edges (the two pairs of glued sides), 1 face (the square itself). Apply the recipe: $1-2+1=0$.

Step 4. Now add a vertex in the middle of the doughnut's square face and connect it to the corner with two new edges, splitting the face into two triangles. New count: 2 vertices, 4 edges, 2 faces. The recipe: $2-4+2=0$. Same answer.

What this tells us: the Euler characteristic depends on the surface, not on the way you cut it up. A cube and a sphere give 2 because they are the same surface up to deformation. A doughnut gives 0 because it is genuinely a different shape.

Check your understanding [Beginner]

Formal definition [Intermediate+]

Let $X$ be a finite CW complex with $c_n$ cells in dimension $n$, where $c_n=0$ for all but finitely many $n$. The cellular Euler characteristic is $$ \chi(X) = \sum_{n \geq 0} (-1)^n c_n. $$

Let $X$ be a topological space whose integer singular homology groups $H_n(X;\mathbb{Z})$ are finitely generated abelian groups, vanishing for all but finitely many $n$. The homological Euler characteristic is $$ \chi(X) = \sum_{n \geq 0} (-1)^n \mathrm{rank}, H_n(X; \mathbb{Z}) = \sum_{n \geq 0} (-1)^n \dim_{\mathbb{Q}} H_n(X; \mathbb{Q}), $$ where $\mathrm{rank}A$ denotes the free rank of a finitely generated abelian group $A$ — the dimension of $A{\otimes }_{\mathbb{Z}}\mathbb{Q}$ as a rational vector space.

The two definitions agree on every finite CW complex, and the agreement implies that $\chi (X)$ is independent of the cellular structure chosen on $X$ — it is a topological invariant of the underlying space.

For a topological pair $(X,A)$ with $A\subseteq X$ a sub-CW-complex of a finite CW complex, the relative Euler characteristic is $\chi (X,A)=\chi (X)-\chi (A)$, and equivalently $\chi (X,A)={\sum }_n(-1{)}^n\mathrm{rank}{H}_n(X,A)$.

Counterexamples to common slips

• The Euler characteristic is not a homotopy invariant of arbitrary spaces — it is an invariant of spaces with finitely-generated homology in finitely many degrees. For an infinite-dimensional space such as a $K(\mathbb{Z},2)=\mathbb{C}{P}^{\infty }$ the alternating sum diverges and $\chi$ is undefined.
• Torsion in integer homology contributes nothing to $\chi$. The space $\mathbb{R}{P}^2$ has $H_0=\mathbb{Z}$, $H_1=\mathbb{Z}/2$, $H_2=0$; the ranks are $1,0,0$, giving $\chi (\mathbb{R}{P}^2)=1$. The torsion class in $H_1$ is invisible to $\chi$.
• Multiplicativity $\chi (X\times Y)=\chi (X)\chi (Y)$ holds without any orientability hypothesis, but the fibre-bundle generalisation $\chi (E)=\chi (F)\chi (B)$ requires the action of ${\pi }_1(B)$ on $H_{\ast }(F;\mathbb{Q})$ to be the identity, which is automatic when the base is simply connected.
• The vanishing $\chi (M)=0$ for closed odd-dimensional orientable manifolds is a Poincaré-duality consequence and holds for the manifold itself, not for arbitrary odd-dimensional CW complexes. The odd-sphere result $\chi (S^{2k+1})=0$ is a special case.

Key theorem with proof [Intermediate+]

Theorem (cellular equals homological; Hatcher Theorem 2.44). Let $X$ be a finite CW complex with $c_n$ cells in dimension $n$. Then $$ \sum_{n \geq 0} (-1)^n c_n = \sum_{n \geq 0} (-1)^n \mathrm{rank}, H_n(X; \mathbb{Z}). $$ In particular, the cellular Euler characteristic is a topological invariant: any two CW structures on the same space yield the same alternating cell sum.

Proof. Let $C_{\bullet }^{\mathrm{cell}}(X)$ denote the cellular chain complex of $X$, with $C_n^{\mathrm{cell}}={\mathbb{Z}}^{c_n}$ a free abelian group of rank $c_n$ on the set of $n$-cells. The cellular boundary ${\partial }_n:C_n^{\mathrm{cell}}\to C_{n-1}^{\mathrm{cell}}$ has image $B_{n-1}$ and kernel $Z_n$; the homology is $H_n=Z_n/B_n$.

Since each $C_n^{\mathrm{cell}}$ is free of rank $c_n$, the rank-nullity identity over $\mathbb{Z}$ (which holds for finitely generated abelian groups via tensoring with $\mathbb{Q}$ to reduce to vector spaces) gives $$ c_n = \mathrm{rank}, Z_n + \mathrm{rank}, B_{n-1}. $$ Multiply by $(-1{)}^n$ and sum over $n$: $$ \sum_n (-1)^n c_n = \sum_n (-1)^n \mathrm{rank}, Z_n + \sum_n (-1)^n \mathrm{rank}, B_{n-1}. $$ Re-index the second sum by $m=n-1$, so ${\sum }_n(-1{)}^n\mathrm{rank}{B}_{n-1}=-{\sum }_m(-1{)}^m\mathrm{rank}{B}_m$. Therefore $$ \sum_n (-1)^n c_n = \sum_n (-1)^n (\mathrm{rank}, Z_n - \mathrm{rank}, B_n). $$ The short exact sequence $0\to B_n\to Z_n\to H_n\to 0$ of finitely generated abelian groups yields, after tensoring with $\mathbb{Q}$, a short exact sequence of rational vector spaces, so $\mathrm{rank}{Z}_n-\mathrm{rank}{B}_n=\mathrm{rank}{H}_n$. The identity follows.

The agreement of the two sides is independent of the cellular structure used to compute the left side, since the right side is an invariant of $X$ alone. So the cellular sum $\sum (-1{)}^n{c}_n$ is a topological invariant. $\square$

Theorem (multiplicativity; Hatcher Theorem 3B.6). Let $X,Y$ be finite CW complexes. Then $\chi (X\times Y)=\chi (X)\cdot \chi (Y)$.

Proof. Use rational coefficients, where Künneth becomes a tensor product of vector spaces. The Künneth formula gives $$ H_n(X \times Y; \mathbb{Q}) \cong \bigoplus_{p + q = n} H_p(X; \mathbb{Q}) \otimes_{\mathbb{Q}} H_q(Y; \mathbb{Q}). $$ Take dimensions: $$ \dim H_n(X \times Y; \mathbb{Q}) = \sum_{p + q = n} \dim H_p(X; \mathbb{Q}) \cdot \dim H_q(Y; \mathbb{Q}). $$ Form the alternating sum, with $b_p^X=\dim H_p(X;\mathbb{Q})$ and similarly for $Y$: $$ \chi(X \times Y) = \sum_n (-1)^n \sum_{p + q = n} b_p^X b_q^Y = \sum_{p, q} (-1)^{p + q} b_p^X b_q^Y = \left(\sum_p (-1)^p b_p^X\right) \left(\sum_q (-1)^q b_q^Y\right) = \chi(X) \chi(Y). \quad \square $$

Bridge. This establishes the Euler characteristic as a stable, computable invariant: cellular and homological definitions coincide, products multiply, and the answer depends only on the homotopy type of the space (provided the homology is finitely generated in finitely many degrees). The same alternating-sum mechanism builds toward two further appearances. First, on a closed orientable manifold of odd dimension Poincaré duality forces $\chi =0$, recovering $\chi (S^{2k+1})=0$ as the simplest instance — this is the central insight that animates 03.12.16 (Poincaré duality) at the level of integer-valued invariants. Second, the multiplicative formula generalises: for a fibre bundle $F\to E\to B$ with simply connected base and finite-type fibre, the same Künneth-style analysis applied to the Leray-Serre spectral sequence yields $\chi (E)=\chi (F)\chi (B)$, and the alternating-sum trick appears again in the proof of the Lefschetz fixed-point theorem, which generalises the Euler characteristic of the identity to a trace sum over arbitrary self-maps. Putting these together, the Euler characteristic is the foundational integer-valued shadow of all higher cohomological invariants, and the same telescoping-rank argument that proves cellular equals homological is the bridge between combinatorial counting and the differential-geometric incarnations that appear again in the Master section.

Exercises [Intermediate+]

Lean formalization [Intermediate+]

Mathlib has the cellular and singular homology infrastructure but does not yet ship the Euler characteristic as a named integer-valued invariant. The intended formalisation would read schematically:

[object Promise]

The proof gap is concrete. Mathlib needs a CWComplex.cellCount accessor, the alternating-sum identity for finite chain complexes of free abelian groups, the Künneth isomorphism over $\mathbb{Q}$ packaged for the cohomology of a product CW complex, and a formal statement of Poincaré duality for closed orientable manifolds with $\mathbb{Q}$ coefficients (currently present only for de Rham cohomology of compact manifolds, not the singular-homology form). Each piece is formalisable from existing infrastructure but has not been packaged. The bridge to differential geometry — Gauss-Bonnet and Poincaré-Hopf — is further out and requires Riemannian curvature theory and vector-field-index theory that Mathlib does not yet carry.

Advanced results [Master]

Theorem (Euler-Poincaré formula for fibre bundles). Let $F\to E\to B$ be a fibre bundle in which $F$ and $B$ are finite CW complexes and the action of ${\pi }_1(B)$ on $H_(F; \mathbb{Q})$istheidentity—automaticwhen$B$issimplyconnected.Then$\chi(E) = \chi(F) \cdot \chi(B)$.*

The proof uses the Leray-Serre spectral sequence with rational coefficients, $E_2^{p,q}=H^p(B;H^q(F;\mathbb{Q}))=H^p(B;\mathbb{Q})\otimes H^q(F;\mathbb{Q})$ under the triviality hypothesis on the local system, and the fact that the Euler characteristic of a graded vector space is invariant under taking the homology of any differential. The alternating sum on $E_{\infty }$ equals the alternating sum on $E_2$, which factorises as $\chi (B)\chi (F)$.

Theorem (vanishing on closed odd-dimensional orientable manifolds). Let $M$ be a closed orientable manifold of odd dimension $n$. Then $\chi (M)=0$.

Poincaré duality with rational coefficients gives $\dim H^k(M;\mathbb{Q})=\dim H^{n-k}(M;\mathbb{Q})$. The alternating sum ${\sum }_k(-1{)}^k\dim H^k$ pairs $k$ with $n-k$. For $n$ odd, $(-1{)}^k$ and $(-1{)}^{n-k}$ have opposite signs, so the pairing cancels each pair; the unpaired middle index $k=n/2$ does not occur. Therefore $\chi (M)=0$.

Theorem (Lefschetz fixed-point theorem; Hatcher §2.C). Let $X$ be a finite CW complex and $f:X\to X$ a continuous map. The Lefschetz number $$ L(f) = \sum_{n \geq 0} (-1)^n \mathrm{tr}\left(f_* : H_n(X; \mathbb{Q}) \to H_n(X; \mathbb{Q})\right) $$ is an integer that depends only on the homotopy class of $f$. If $L(f)\ne 0$, then $f$ has a fixed point.

For $f={\mathrm{id}}_X$, every trace equals $\dim H_n(X;\mathbb{Q})$, recovering $L(\mathrm{id})=\chi (X)$. So the Lefschetz number generalises the Euler characteristic of the identity to a trace sum over arbitrary self-maps. The contrapositive gives the existence statement: a fixed-point-free self-map forces $L(f)=0$. Hatcher proves the theorem by a simplicial-approximation argument plus a chain-level identification of $L(f)$ with the alternating trace on the cellular chain complex.

Theorem (Gauss-Bonnet; do Carmo Ch. 4). Let $(M,g)$ be a closed oriented Riemannian 2-manifold with Gaussian curvature $K$ and area form $dA$. Then $$ \int_M K, dA = 2\pi, \chi(M). $$

The classical proof on a triangulated surface localises the curvature integral to the angular defect at each vertex plus the geodesic curvature along each edge. Summing the angular defects and edges by the combinatorics of the triangulation produces the Euler-characteristic formula on the right.

Theorem (Chern-Gauss-Bonnet; Chern 1944). Let $(M,g)$ be a closed oriented Riemannian manifold of even dimension $n=2k$. Let $\Omega$ denote the curvature 2-form of the Levi-Civita connection on $TM$, an $\mathfrak{so}(n)$-valued 2-form. Let $\mathrm{Pf}(\Omega )$ denote the Pfaffian of the curvature, an $n$-form on $M$. Then $$ \int_M \mathrm{Pf}\left(\frac{\Omega}{2\pi}\right) = \chi(M). $$ For $n$ odd, $\chi (M)=0$ as above and there is no curvature integral to write.

The Pfaffian of an antisymmetric $n\times n$ matrix is the square root of its determinant, well-defined up to sign and stabilised by the orientation of $M$. The proof goes through Chern-Weil theory: the form $\mathrm{Pf}(\Omega /2\pi )$ is closed, gauge-invariant, and represents the Euler class $e(TM)\in H^n(M;\mathbb{R})$; the integral is the pairing of $e(TM)$ with the fundamental class $[M]$, which equals $\chi (M)$ by the Poincaré-Hopf theorem.

Theorem (Poincaré-Hopf; Milnor §6). Let $M$ be a closed oriented smooth manifold and $V$ a smooth vector field on $M$ with isolated zeros. Then $$ \sum_{p : V(p) = 0} \mathrm{ind}_p(V) = \chi(M), $$ where ${\mathrm{ind}}_p(V)$ is the local degree of $V/|V|$ on a small sphere around $p$.

The proof identifies the index sum with the pairing $⟨e(TM),[M]⟩$ via the Thom isomorphism for the tangent bundle and an excision argument that localises the Euler class to a neighbourhood of the zero set of $V$. The same identification, run in the other direction, gives the chain of equalities ${\sum }_p{\mathrm{ind}}_p(V)=⟨e(TM),[M]⟩=\chi (M)$ that links differential topology to homotopy invariants.

Theorem (Hopf and the Hopf index theorem). On a closed orientable surface ${\Sigma }_g$, every smooth vector field has zeros if and only if $g\ne 1$. For the torus $T^2={\Sigma }_1$, a non-vanishing vector field exists.

This is the Poincaré-Hopf theorem applied to surfaces: $\chi ({\Sigma }_g)=2-2g$ vanishes precisely for $g=1$, the torus. The non-vanishing torus vector field is a constant section in the trivialisation $T(T^n)=T^n\times {\mathbb{R}}^n$.

Synthesis. The Euler characteristic occupies a unique position in the architecture of topology: it is the simplest integer-valued homotopy invariant, and it is exactly the integer that mediates between every adjacent layer of geometric and topological structure. The central insight is that the alternating-sum mechanism $\sum (-1{)}^n\mathrm{rank}{H}_n$ is forced by the rank-additivity of short exact sequences of finitely generated abelian groups; this is exactly the same mechanism that makes the Euler characteristic of a graded vector space invariant under any differential, and the same mechanism that produces the Lefschetz number as the Euler characteristic of the identity generalised to a trace. Putting these together, $\chi$ identifies the cellular count with the homological rank, identifies the homological rank with the Pfaffian curvature integral via Chern-Gauss-Bonnet, and identifies the Pfaffian integral with the vector-field-index sum via Poincaré-Hopf. Each identification expresses the same integer in a different language: combinatorial (cells), algebraic (ranks), differential-geometric (curvature), and dynamical (vector-field zeros).

The bridge between these languages is the Euler class $e(TM)\in H^n(M;\mathbb{Z})$, the obstruction to a global non-vanishing section of the tangent bundle. Pairing the Euler class with the fundamental class $[M]$ produces the Euler characteristic in every setting; the four computations are four chart-level evaluations of the same cohomological pairing. Putting these together with the multiplicativity of $\chi$ under products and over fibre bundles, one obtains a complete dictionary: every topological operation on spaces — product, fibre bundle with simply connected base, suspension, wedge — has a corresponding arithmetic operation on Euler characteristics, and the dictionary is exact for finite CW complexes. This is the foundational reason that $\chi$ shows up in the Riemann-Roch theorem, in the Hirzebruch signature theorem, and in the Atiyah-Singer index theorem: each is a refinement of the same pairing in which the Pfaffian of curvature is replaced by a more elaborate characteristic class, and the Euler characteristic on the right is replaced by a more elaborate integer-valued invariant. The Euler characteristic is the simplest case of the general principle that index = topology, and the same alternating-rank cancellation that proves $\chi$ is well-defined is the bridge between the analytic index of an elliptic operator and the topological index of its symbol.

Full proof set [Master]

Theorem (cellular equals homological), proof. Already given in §I. The argument is rank-additivity over short exact sequences of finitely generated abelian groups, telescoped through the cellular chain complex. $\square$

Theorem (multiplicativity), proof. Already given in §I. The Künneth formula over $\mathbb{Q}$ produces a tensor-product factorisation of the Betti numbers of $X\times Y$, and the alternating sum factors as a product. $\square$

Theorem (Euler-Poincaré for fibre bundles), proof. Let $F\to E\to B$ be a fibre bundle satisfying the triviality hypothesis on the local system. The Leray-Serre spectral sequence has $E_2^{p,q}=H^p(B;\mathbb{Q}){\otimes }_{\mathbb{Q}}{H}^q(F;\mathbb{Q})$ and converges to $H^{p+q}(E;\mathbb{Q})$. The Euler characteristic of a graded vector space is invariant under taking the homology with respect to any differential, since rank-additivity over short exact sequences forces $\sum (-1{)}^{p+q}\dim E_r^{p,q}$ to be the same on every page $r\ge 2$. Hence $$ \chi(E) = \sum_{p, q} (-1)^{p + q} \dim E_\infty^{p, q} = \sum_{p, q} (-1)^{p + q} \dim E_2^{p, q} = \sum_{p, q} (-1)^{p + q} \dim H^p(B; \mathbb{Q}) \dim H^q(F; \mathbb{Q}) = \chi(B) \chi(F). \quad \square $$

Theorem (vanishing on closed odd-dimensional orientable manifolds), proof. Already given in §I (Exercise 5) and recapped above. Poincaré duality $H_p(M;\mathbb{Q})\cong H_{n-p}(M;\mathbb{Q})$ for $n$ odd produces $\chi (M)=-\chi (M)$ via parity of $(-1{)}^p$ versus $(-1{)}^{n-p}$, hence $\chi (M)=0$. $\square$

Theorem (Lefschetz fixed-point theorem), proof. Let $X$ be a finite CW complex and $f:X\to X$. The Lefschetz number $L(f)$ is integer-valued because the trace of $f_{\ast }$ on a finitely generated free abelian group is an integer, and $L(f)$ depends only on the homotopy class of $f$ because $f\simeq g$ induces $f_{\ast }=g_{\ast }$ on homology.

Suppose $f$ has no fixed point. By compactness of $X$ and the assumption that $f$ is fixed-point free, there exists a CW structure on $X$ (refined if necessary) and a cellular approximation $g:X\to X$ homotopic to $f$ such that $g(\bar{e})\cap \bar{e}=\varnothing$ for every cell $\bar{e}$ of $X$. (This uses the simplicial-approximation theorem applied to a fine subdivision.) Then on the cellular chain complex, $g_{\ast }:C_n^{\mathrm{cell}}(X)\to C_n^{\mathrm{cell}}(X)$ has zero diagonal entries in the basis of cells: the matrix coefficient of $g_{\ast }(e_{\alpha }^n)$ along $e_{\alpha }^n$ vanishes because the image of ${\bar{e}}_{\alpha }^n$ under $g$ is disjoint from ${\bar{e}}_{\alpha }^n$. Hence $\mathrm{tr}(g_{\ast })=0$ on every $C_n^{\mathrm{cell}}$. The alternating-trace identity (analogous to the cellular-equals-homological identity, with traces in place of dimensions) gives $$ L(f) = L(g) = \sum_n (-1)^n \mathrm{tr}(g_* : H_n \to H_n) = \sum_n (-1)^n \mathrm{tr}(g_* : C_n^{\mathrm{cell}} \to C_n^{\mathrm{cell}}) = 0. $$ The contrapositive: $L(f)\ne 0$ implies $f$ has a fixed point. $\square$

Theorem (Gauss-Bonnet), proof. Triangulate $(M,g)$ by geodesic triangles. The Gauss-Bonnet formula for a single geodesic triangle $T$ with interior angles ${\alpha }_1,{\alpha }_2,{\alpha }_3$ states $$ \int_T K, dA = (\alpha_1 + \alpha_2 + \alpha_3) - \pi. $$ This is the local Gauss-Bonnet formula, proven by integration of the connection form on the boundary using the structure equation $d\omega =-KdA$ and the fact that geodesic edges have zero geodesic curvature.

Sum over all triangles of the triangulation. The angles at a single vertex sum to $2\pi$ (the full angle around the vertex). The total angle sum is $2\pi V$ where $V$ is the number of vertices. The total $-\pi$ contribution is $-\pi F$ where $F$ is the number of triangles (faces). Each edge bounds two triangles. Standard counting on a triangulation of a closed surface gives $3F=2E$, hence $F=2E-2F$ rearranges to $V-E+F=V-F/2$. Substituting: $$ \int_M K, dA = 2\pi V - \pi F = \pi(2V - F) = 2\pi(V - E + F) = 2\pi \chi(M), $$ using $E=(3F)/2$ to convert. $\square$

Theorem (Chern-Gauss-Bonnet), proof. Sketch via Chern-Weil. The Levi-Civita connection on $TM$ has curvature 2-form $\Omega \in {\Omega }^2(M,\mathfrak{so}(n))$, and the Pfaffian polynomial $\mathrm{Pf}:\mathfrak{so}(n)\to \mathbb{R}$ extends by polynomial functoriality to a closed gauge-invariant differential form $\mathrm{Pf}(\Omega /2\pi )\in {\Omega }^n(M)$. The Chern-Weil theorem identifies its de Rham class with the image of the Euler class $e(TM)\in H^n(M;\mathbb{Z})$ under the change-of-coefficients map to $H^n(M;\mathbb{R})$. The pairing $⟨e(TM),[M]⟩$ equals the Euler characteristic by the Poincaré-Hopf theorem (next). Hence ${\int }_M\mathrm{Pf}(\Omega /2\pi )=⟨e(TM),[M]⟩=\chi (M)$. The full Chern 1944 proof uses an explicit transgression form on the unit-sphere bundle of $TM$. $\square$

Theorem (Poincaré-Hopf), proof. Let $V$ be a smooth vector field on $M$ with isolated zeros $p_1,\dots ,p_k$. Choose disjoint coordinate balls $B_i$ around each $p_i$ and let $M^{\prime }=M\setminus {⨆}_i{B}_i$. On $M^{\prime }$, $V$ is non-vanishing, so $V/|V|$ defines a section $s:M^{\prime }\to S(TM)$ of the unit-sphere bundle. The Thom isomorphism for $TM\to M$ relates the Euler class to the fibre integral of the Thom form, and the index ${\mathrm{ind}}_{p_i}(V)$ is by definition the degree of $V/|V|$ on $\partial B_i$. The compactness of $M$ and the orientation give an excision identity: $$ \langle e(TM), [M] \rangle = \sum_i \deg(V/|V| : \partial B_i \to S^{n-1}) = \sum_i \mathrm{ind}_{p_i}(V). $$ The pairing $⟨e(TM),[M]⟩$ equals $\chi (M)$ by the cohomological identification $e(TM)⌢[M]=\chi (M)[\mathrm{pt}]$, which is the original definition of the Euler characteristic via the Euler class evaluated on the fundamental class. The full proof in Milnor §6 uses Morse theory to construct a generic gradient field whose index sum is computed by the alternating count of Morse critical points, then identifies that count with the Euler characteristic. $\square$

Connections [Master]

• Cellular homology 03.12.13. The cellular chain complex of a finite CW complex provides the most direct computation of the Euler characteristic: $\chi (X)={\sum }_n(-1{)}^n{c}_n$ where $c_n$ is the number of $n$-cells. The cellular-equals-homological theorem is the statement that this counting agrees with the rank-based homological definition; the proof is the rank-additivity argument applied to the cellular boundary.

• Poincaré duality 03.12.16. Poincaré duality forces $\chi =0$ on closed orientable manifolds of odd dimension, by the parity-pairing argument. On even dimensions, Poincaré duality combined with Poincaré-Hopf identifies the Euler characteristic with the integral of the Pfaffian of the curvature 2-form. The Euler characteristic is therefore the simplest integer-valued shadow of Poincaré duality at the level of Betti numbers.

• Universal coefficient theorem 03.12.18. The universal coefficient theorem implies that the Euler characteristic computed with field coefficients ${\sum }_n(-1{)}^n{\dim }_{\mathbb{F}}{H}_n(X;\mathbb{F})$ is independent of the field $\mathbb{F}$, since the torsion contributions telescope. This is the source of the field-independence of $\chi$ and underlies the universal-coefficient calculation of $\chi (\mathbb{R}{P}^n)$ and of products of projective spaces.

• Künneth theorem 03.04.12. The Künneth formula over a field expresses the cohomology of a product as a tensor product, and dimension-counting yields $\chi (X\times Y)=\chi (X)\chi (Y)$. The multiplicativity of $\chi$ is therefore a Künneth corollary; the same Künneth identity, applied fibrewise to a Leray-Serre spectral sequence, gives the Euler-Poincaré formula for fibre bundles with simply connected base.

• Sphere bundle and Euler class 03.05.10 pending. The Euler class $e(E)\in H^r(B;\mathbb{Z})$ of an oriented rank-$r$ vector bundle is the obstruction to a global non-vanishing section. For $E=TM$ on a closed oriented manifold, the pairing $⟨e(TM),[M]⟩$ equals the Euler characteristic. The Pfaffian curvature representative of $e(TM)$ is the differential-geometric incarnation of the same integer, and the index sum of a generic vector field is its dynamical incarnation.

• Atiyah-Singer index theorem 03.09.10. The Atiyah-Singer index theorem expresses the analytic index of an elliptic operator as the integral of a characteristic class. For the de Rham complex on a closed oriented even-dimensional manifold, the analytic index is the Euler characteristic, the characteristic class is the Pfaffian of curvature, and the theorem reduces to Chern-Gauss-Bonnet. The Euler characteristic is therefore the simplest index = topology identity.

• Eilenberg-Steenrod axioms 03.12.15. The Euler characteristic is determined by the Eilenberg-Steenrod axioms applied to a finite CW complex: the dimension axiom fixes the value on a point, the additivity and Mayer-Vietoris axioms force multiplicativity over wedges and additivity over excisive triples, and the homotopy axiom makes $\chi$ a homotopy invariant. Any integer-valued function on finite CW complexes that satisfies these axiomatic properties agrees with the Euler characteristic.

Historical & philosophical context [Master]

Leonhard Euler observed in Elementa doctrinae solidorum (Novi Comm. Acad. Sci. Petrop. 4, 1758, pp. 109-140) ^[pending] that for every convex polyhedron the count $V-E+F$ equals two. The paper is the earliest record of the formula and contains a partial proof; a rigorous proof in modern hands was given by Cauchy in 1813 (Cauchy's proof of $V-E+F=2$ for convex polyhedra by stereographic projection onto the plane). Euler's contemporaries did not recognise the formula as a topological invariant — the very notion of topology had not yet been articulated — but the constancy of $V-E+F$ across different polyhedra of the same combinatorial type was understood to encode something intrinsic to the sphere.

The modern algebraic-topology framing was given by Henri Poincaré in the 1895 Analysis Situs (Journal de l'École Polytechnique (2) 1, pp. 1-121) ^[pending], where the Euler characteristic was identified with the alternating sum of Betti numbers (the ranks of homology in modern language) and shown to be a topological invariant of a manifold rather than a combinatorial accident of a polyhedral structure. Poincaré used the formula in his proof of what is now called Poincaré duality, observing that for a closed orientable manifold of odd dimension the alternating Betti sum vanishes by the duality pairing.

Solomon Lefschetz extended the Euler-characteristic framework in Intersections and transformations of complexes and manifolds (Trans. Amer. Math. Soc. 28, 1926, pp. 1-49) ^[pending], generalising the alternating Betti sum to the alternating trace of an arbitrary self-map and proving the Lefschetz fixed-point theorem. The Euler characteristic is recovered as the Lefschetz number of the identity map. The differential-geometric incarnation of $\chi$ via the Pfaffian of the curvature 2-form was given by Shiing-Shen Chern in A simple intrinsic proof of the Gauss-Bonnet formula for closed Riemannian manifolds (Ann. Math. 45, 1944, pp. 747-752) ^[pending], extending the classical Gauss-Bonnet theorem from surfaces to closed Riemannian manifolds of any even dimension. Chern's proof, using the global angular form on the unit-sphere bundle of $TM$, became the template for Chern-Weil theory and the modern theory of characteristic classes.

The vector-field formulation of $\chi$ as an index sum was given by Heinz Hopf and Henri Poincaré in the 1880s-1920s, with the modern smooth-manifold version codified by John Milnor in Topology from the Differentiable Viewpoint (Princeton 1965). The convergence of the cellular, homological, differential-geometric, and dynamical formulations of the Euler characteristic in the mid-20th century laid the groundwork for the Atiyah-Singer index theorem of 1963, which expresses the analytic index of an arbitrary elliptic operator as a topological pairing — generalising the Chern-Gauss-Bonnet theorem in exactly the direction Euler's polyhedral count first hinted at.

Bibliography [Master]

[object Promise]