nLab Donaldson invariant

Redirected from "Donaldson invariants".

Context

Differential geometry

synthetic differential geometry

Introductions

from point-set topology to differentiable manifolds

geometry of physics: coordinate systems, smooth spaces, manifolds, smooth homotopy types, supergeometry

Differentials

V-manifolds

smooth space

Tangency

The magic algebraic facts

Theorems

Axiomatics

cohesion

infinitesimal cohesion

tangent cohesion

differential cohesion

graded differential cohesion

singular cohesion

id ⊣ id ∨ ∨ fermionic ⇉ ⊣ ⇝ bosonic ⊥ ⊥ bosonic ⇝ ⊣ Rh rheonomic ∨ ∨ reduced ℜ ⊣ ℑ infinitesimal ⊥ ⊥ infinitesimal ℑ ⊣ & étale ∨ ∨ cohesive ʃ ⊣ ♭ discrete ⊥ ⊥ discrete ♭ ⊣ ♯ continuous ∨ ∨ ∅ ⊣ * \array{ && id &\dashv& id \\ && \vee && \vee \\ &\stackrel{fermionic}{}& \rightrightarrows &\dashv& \rightsquigarrow & \stackrel{bosonic}{} \\ && \bot && \bot \\ &\stackrel{bosonic}{} & \rightsquigarrow &\dashv& \mathrm{R}\!\!\mathrm{h} & \stackrel{rheonomic}{} \\ && \vee && \vee \\ &\stackrel{reduced}{} & \Re &\dashv& \Im & \stackrel{infinitesimal}{} \\ && \bot && \bot \\ &\stackrel{infinitesimal}{}& \Im &\dashv& \& & \stackrel{\text{étale}}{} \\ && \vee && \vee \\ &\stackrel{cohesive}{}& \esh &\dashv& \flat & \stackrel{discrete}{} \\ && \bot && \bot \\ &\stackrel{discrete}{}& \flat &\dashv& \sharp & \stackrel{continuous}{} \\ && \vee && \vee \\ && \emptyset &\dashv& \ast }

Models

Lie theory, ∞-Lie theory

differential equations, variational calculus

Chern-Weil theory, ∞-Chern-Weil theory

Cartan geometry (super, higher)

Topology

topology (point-set topology, point-free topology)

see also differential topology, algebraic topology, functional analysis and topological homotopy theory

Introduction

Basic concepts

Universal constructions

Extra stuff, structure, properties

Examples

Basic statements

Theorems

Analysis Theorems

topological homotopy theory

Contents

Idea

Donaldson invariants, due to Donaldson 1986, 1990, are invariants of the smooth structure of smooth 4-manifolds based on the methods behind Donaldson's theorem, due to Donaldson 1983, 1987. In particular, key informations are obtained from the Yang-Mills moduli space of the Yang-Mills equations on the 4-manifold, which as partial differential equations depend on the smooth structure. Donaldson invariants can distinguish these, hence are able to detect exotic smooth structures.

SU(2) Donaldson invariants

Let XX be an oriented closed smooth 4-manifold and P↠XP\twoheadrightarrow X be a principal SU(2)-bundle, then there exists a unique homotopy class represented by a classifying map τ:X→BSU(2)≅ℍP ∞\tau\colon X\rightarrow BSU(2)\cong\mathbb{H}P^\infty into the classifying space for SU(2) giving a pullback P=τ *ESU(2)=τ *S ∞P=\tau^*ESU(2)=\tau^*S^\infty. With the Kronecker pairing and the fundamental class [X]∈H 4(X,ℤ)[X]\in H_4(X,\mathbb{Z}) from the orientation, the second Chern class concretely gives a group isomorphism:

⟨c 2(−),[X]⟩:Prin SU(2)(X)→≅ℤ. \langle c_2(-),[X]\rangle\colon Prin_{SU(2)}(X)\xrightarrow\cong\mathbb{Z}.

Let 𝒜:=Ω Ad 1(P,𝔰𝔲(2))\mathcal{A}:=\Omega_{Ad}^1(P,\mathfrak{su}(2)) be the configuration space of connections on PP, which is an affine vector space not canonically isomorphic to Ω 1(X,Ad(P))\Omega^1(X,Ad(P)). Let 𝒢≔Aut SU(2)(P)\mathcal{G}\coloneqq Aut_{SU(2)}(P) be the gauge group of gauge transformations of PP, which acts on 𝒜\mathcal{A}. Let ℬ≔𝒜/𝒢\mathcal{B}\coloneqq\mathcal{A}/\mathcal{G} be the orbit space, (Don. 86, Eq. (2.1)) then there is a weak homotopy equivalence ℬ→Map P(X,BSU(2))\mathcal{B}\rightarrow Map_P(X,BSU(2)) in the homotopy class of the classifying map of PP. (Don. 86, Lem. (2.4), Don. & Kro. 97, Prop. 5.1.4) Currying it corresponds to a map ℬ×X→BSU(2)\mathcal{B}\times X\rightarrow BSU(2) and through pullback a principal SU(2)SU(2)-bundle P˜↠ℬ×X\widetilde{P}\twoheadrightarrow\mathcal{B}\times X. (Don. & Kro. 97, Eq. (5.1.5)) For every gauge class [A]∈ℬ[A]\in\mathcal{B}, one has P˜| {[A]}×X≅P↠X\widetilde{P}\vert_{\{[A]\}\times X}\cong P\twoheadrightarrow X. With the second Chern class c 2(P˜)∈H 4(ℬ×X,ℤ)c_2(\widetilde{P})\in H^4(\mathcal{B}\times X,\mathbb{Z}), the slant product yields a map: (Don. 86, Def. 2.5, Fre. & Uhl. 84, p. xi-xii, Don. & Kro. 97, p. 176-177)

μ:H 2(X,ℤ)→H 2(ℬ,ℤ),μ(α)≔c 2(P˜)/α. \mu\colon H_2(X,\mathbb{Z})\rightarrow H^2(\mathcal{B},\mathbb{Z}), \mu(\alpha)\coloneqq c_2(\widetilde{P})/\alpha.

With the subset ℳ⊂ℬ\mathcal{M}\subset\mathcal{B} of an (anti) self-dual Yang-Mills moduli space there is an induced map H 2(ℬ,ℤ)↪H 2(ℳ,ℤ)H^2(\mathcal{B},\mathbb{Z})\hookrightarrow H^2(\mathcal{M},\mathbb{Z}), which through postcomposition yields a map μ:H 2(X,ℤ)→H 2(ℳ,ℤ)\mu\colon H_2(X,\mathbb{Z})\rightarrow H^2(\mathcal{M},\mathbb{Z}). With the interpretation of the homology classes as embedded surfaces i:Σ↪Mi\colon\Sigma\hookrightarrow M with i *[Σ]∈H 2(M,ℤ)i_*[\Sigma]\in H_2(M,\mathbb{Z}) and the interpretation of the cohomology classes as differential forms of second order using de Rham cohomology H 2(ℳ,ℝ)≅H dR 2(ℳ)H^2(\mathcal{M},\mathbb{R})\cong H_{dR}^2(\mathcal{M}), one can see the map μ\mu as mapping embedded surfaces in MM to differential forms of second order in ℳ\mathcal{M}. Donaldson invariants are defined using the integration of these differential forms, which is disrupted by the missing compactness and the singularities of the moduli space. A compactification ℳ⊂ℳ¯\mathcal{M}\subset\overline\mathcal{M} can be constructed using Uhlenbeck's compactness theorem and leads to a ℤ\mathbb{Z}-linear map μ:H 2(X,ℤ)→H 2(ℳ¯,ℤ)\mu\colon H_2(X,\mathbb{Z})\rightarrow H^2(\overline\mathcal{M},\mathbb{Z}).

Let P k↠XP_k\twoheadrightarrow X be the unique principal SU(2)-bundle with ⟨c 2(P k),[X]⟩=k\langle c_2(P_k),[X]\rangle=k according to the above group isomorphism, then ⟨p 1Ad(P k),[X]⟩=−4k\langle p_1 Ad(P_k),[X]\rangle=-4k for the first Pontrjagin class of the adjoint bundle and the virtual dimension of its anti self-dual Yang-Mills (ASDYM) moduli space is given with the Atiyah-Singer index theorem by: (Sco. 05, p. 355)

dimℳ k=−8k−3(1−b 1(M)+b 2 +(M)). \dim\mathcal{M}_k =-8k-3(1-b_1(M)+b_2^+(M)).

If −b 1(M)+b 2 +(M)-b_1(M)+b_2^+(M) is odd, then the virtual dimension is an even number dimℳ k=2d k\dim\mathcal{M}_k=2d_k, potentially negative given the choice of kk. With a fundamental class [ℳ k]∈H 2d k(ℳ k,ℤ)≅ℤ b 0(ℳ k)\left[\mathcal{M}_k\right]\in H_{2d_k}(\mathcal{M}_k,\mathbb{Z}) \cong\mathbb{Z}^{b_0\left(\mathcal{M}_k\right)} since the ASDYM moduli space is orientable, the Donaldson invariant is given with the cup product and the Kronecker pairing: (Sco. 05, p. 355)

𝒟 X,k:Sym d kH 2(X,ℤ)→ℤ,𝒟 X,k(α 1,…,α d k)≔⟨μ(α 1)⌣…⌣μ(α d k),[ℳ k]⟩. \mathcal{D}_{X,k}\colon Sym^{d_k}H_2(X,\mathbb{Z})\rightarrow\mathbb{Z}, \mathcal{D}_{X,k}(\alpha_1,\ldots,\alpha_{d_k})\coloneqq\left\langle\mu(\alpha_1)\smile\ldots\smile\mu(\alpha_{d_k}), \left[\mathcal{M}_k\right]\right\rangle.

Considering the second homology classes in a symmetric setting only up to permutation comes from the symmetry of the cup product, hence the irrelevance of any order. It can furthermore be shown, that the Donaldson invariants also don’t depend on the Riemannian metric and therefore only on the smooth structure.

Relation to Poincaré duality

Let e∈H 2(X,ℤ)e\in H^2(X,\mathbb{Z}) be a cohomology class class with self-intersection Q(e,e)=e 2[X]=−1Q(e,e)=e^2[X]=-1. If e˜:ℂP 2↪ℳ\widetilde{e}\colon\mathbb{C}P^2\hookrightarrow\mathcal{M} is the boundary inclusion of a neighborhood of its singularity and e˜ *:H 2(ℳ,ℤ)→H 2(ℂP 2,ℤ)≅ℤ\widetilde{e}^*\colon H^2(\mathcal{M},\mathbb{Z})\rightarrow H^2(\mathbb{C}P^2,\mathbb{Z})\cong\mathbb{Z} is the induced map on cohomology with the latter generated by the first Chern class c 1(H)c_1(H) of the tautological line bundle H↠ℂP 2H\twoheadrightarrow\mathbb{C}P^2, then its composition is exactly: (Don. 1986, Lemma 2.28)

e˜ *∘μ=−⟨e,−⟩c 1(H):H 2(X,ℤ)→H 2(ℂP 2,ℤ)≅ℤ. \widetilde{e}^*\circ\mu =-\langle e,-\rangle c_1(H)\colon H_2(X,\mathbb{Z})\rightarrow H^2(\mathbb{C}P^2,\mathbb{Z}) \cong\mathbb{Z}.

As mentioned in the definition, the boundary of the moduli space ℳ\mathcal{M} is XX (corresponding to the positions of the instantons when shrinking their scale down and considering the limit) and gives a compactification ℳ¯=ℳ∪X\overline\mathcal{M}=\mathcal{M}\cup X. If τ:X↪ℳ¯\tau\colon X\hookrightarrow\overline\mathcal{M} is the boundary inclusion and τ *:H 2(ℳ¯,ℤ)→H 2(X,ℤ)\tau^*\colon H^2(\overline\mathcal{M},\mathbb{Z})\rightarrow H^2(X,\mathbb{Z}) is the induced map on cohomology, then its composition with the Donaldson invariant is exactly Poincaré duality: (Don. 1986, Lemma 3.8)

PD=τ *∘μ:H 2(X,ℤ)→H 2(X,ℤ) PD =\tau^*\circ\mu\colon H_2(X,\mathbb{Z})\rightarrow H^2(X,\mathbb{Z})

Both these results can be combined in the intersection form. Let α,β∈H 2(X,ℤ)\alpha,\beta\in H_2(X,\mathbb{Z}), then using that pullbacks (like τ *\tau^* and e˜ *\widetilde{e}^*) commute with the cup product and under the Kronecker pairing become pushouts (like τ *\tau_* and e˜ *\widetilde{e}_*) yields: (Don. 1986, p. 291)

Q(α,β) =⟨PD(α)⌣PD(β),[X]⟩=⟨τ *μ(α)⌣τ *μ(β),[X]⟩ =⟨μ(α)⌣μ(β),τ *[X]⟩=12∑ e 2[X]=−1± e⟨μ(α)⌣μ(β),e˜ *[ℂP 2]⟩ =12∑ e 2[X]=−1± e⟨e˜ *μ(α)⌣e˜ *μ(β),[ℂP 2]⟩=12∑ e 2[X]=−1± e⟨α,e⟩⟨β,e⟩⟨c 1 2(H),[ℂP 2]⟩⏟ =1\begin{aligned} Q(\alpha,\beta) &=\langle PD(\alpha)\smile PD(\beta),[X]\rangle =\langle\tau^*\mu(\alpha)\smile\tau^*\mu(\beta),[X]\rangle \\ &=\langle\mu(\alpha)\smile\mu(\beta),\tau_*[X]\rangle =\frac{1}{2}\sum_{e^2[X]=-1}\pm_e\langle\mu(\alpha)\smile\mu(\beta),\widetilde{e}_*[\mathbb{C}P^2]\rangle \\ &=\frac{1}{2}\sum_{e^2[X]=-1}\pm_e\langle\widetilde{e}^*\mu(\alpha)\smile\widetilde{e}^*\mu(\beta),[\mathbb{C}P^2]\rangle =\frac{1}{2}\sum_{e^2[X]=-1}\pm_e\langle\alpha,e\rangle\langle\beta,e\rangle\underbrace{\langle c_1^2(H),[\mathbb{C}P^2]\rangle}_{=1} \end{aligned}

In the middle, the ± e\pm_e account for potential sign differences due to change of orientation and the bordism of the moduli space between XX and a disjoint union of ℂP 2\mathbb{C}P^2 is used to shift τ *[X]\tau_*[X] to the various e˜ *[ℂP 2]\widetilde{e}_*[\mathbb{C}P^2].

Change of instanton number

According to the above formula, dimℳ k+1=dimℳ k+8\dim\mathcal{M}_{k+1}=\dim\mathcal{M}_k+8 holds under a change of the instanton number, expanding the input of the Donaldson invariant by four new second homology classes. More generally, the slant product yields a ℤ\mathbb{Z}-linear map μ:H n(X,ℤ)→H 4−n(ℳ¯,ℤ)\mu\colon H_n(X,\mathbb{Z})\rightarrow H^{4-n}(\overline\mathcal{M},\mathbb{Z}), (Don. & Kro. 97, p. 176-177, Sco. 05, p. 355) so for XX connected with H 0(X,ℤ)≅ℤH_0(X,\mathbb{Z})\cong\mathbb{Z} a map μ:ℤ→H 4(ℳ¯,ℤ)\mu\colon\mathbb{Z}\rightarrow H^4(\overline\mathcal{M},\mathbb{Z}). Hence one can also expand the Donaldson invariant by two images of a generator.

SO(3) Donaldson invariants

Let P↠XP\twoheadrightarrow X be a principal SO(3)-bundle, then there exists a unique homotopy class represented by a classifying map τ:X→BSO(3)\tau\colon X\rightarrow BSO(3) into the classifying space for SO(3) giving a pullback P=τ *ESO(3)P=\tau^* ESO(3). Now the second Stiefel-Whitney class and the first Pontrjagin class give a group isomorphism:

(w 2,⟨p 1(−),[X]⟩):Prin SO(3)(X)→≅{(v,n)∈H 2(X,ℤ 2)×ℤ|⟨v 2,[X]⟩≡nmod4}. (w_2,\langle p_1(-),[X]\rangle)\colon Prin_{SO(3)}(X)\xrightarrow\cong\{(v,n)\in H^2(X,\mathbb{Z}_2)\times\mathbb{Z}|\langle v^2,[X]\rangle\equiv n mod 4\}.

In this case, the first Pontrjagin class replaces the second Chern class in the slant product.

Donaldson invariants of compactification

Compactification to two dimensions

According to the classification of surfaces, every oriented closed 2-manifold (surface) is homeomorphic to a Sigma surface Σ g=#gT 2\Sigma_g=\# gT^2, given as a connected sum of 2-tori T 2=S 1×S 1T^2=S^1\times S^1 with the convention Σ 0=S 2\Sigma_0=S^2. Now the Donaldson invariants can be considered on the 4-manifold Σ g×S 2\Sigma_g\times S^2.

Compactification to three dimensions

For an oriented closed 3-manifold YY, one can consider the Donaldson invariants on the 4-manifold Y×S 1Y\times S^1. For general structure groups, they relate to the Rozansky-Witten invariants of YY and for the structure group SU(2)SU(2), they relate to the Casson-Walker-Lescop invariant? of YY.

References

Donaldson's theorem was introduced and improved in:

Donaldson invariants were introduced and improved in the original articles:

Reviews include:

See also:

Last revised on October 13, 2026 at 08:49:53. See the history of this page for a list of all contributions to it.