ScalingStacks

1.2. Generalised Gibbons-Hawking ansatz [03YP]

Original official author HTML, exact retained edition. Historical TeX conversion verdicts remain unchanged. Cited-edition alignment and mathematical self-containment are not assessed.

Complete original source context Β· Original author HTML

1.2. Generalised Gibbons-Hawking ansatz

The materials in this Section draws heavily from the presentation of Zharkov [31]. Suppose MM is a complex NN-dimensional KΓ€hler manifold with a nonvanishing holomorphic volume form admitting a holomorphic isometric free T𝔫T^{\mathfrak{n}} action. The generalised Gibbons-Hawking ansatz expresses the KΓ€hler and Calabi-Yau conditions in terms of the 𝔫\mathfrak{n} moment map coordinates and the Nβˆ’π”«N-\mathfrak{n} holomorphic coordinates on the KΓ€hler quotient.

Let 𝔱\mathfrak{t} denote the Lie algebra of T𝔫T^{\mathfrak{n}}, and let 𝔱℀\mathfrak{t}_{\mathbb{Z}} be the natural integral lattice in 𝔱\mathfrak{t}. A choice of basis in 𝔱℀\mathfrak{t}_{\mathbb{Z}} defines linear coordinates ΞΌi\mu_{i} on the dual space π”±βˆ—β‰ƒβ„π”«\mathfrak{t}^{*}\simeq\mathbb{R}^{\mathfrak{n}}. Let YY be either β„‚Nβˆ’π”«\mathbb{C}^{N-\mathfrak{n}} or (β„‚βˆ—)Nβˆ’π”«(\mathbb{C}^{*})^{N-\mathfrak{n}}, and let Ξ·p\eta_{p} denote the standard complex coordinates on β„‚Nβˆ’π”«\mathbb{C}^{N-\mathfrak{n}} or the logarithmic coordinates on (β„‚βˆ—)Nβˆ’π”«(\mathbb{C}^{*})^{N-\mathfrak{n}} with period 1 (in which case e2​π​i​ηpe^{2\pi i\eta_{p}} are the standard coordinates on (β„‚βˆ—)Nβˆ’π”«(\mathbb{C}^{*})^{N-\mathfrak{n}}). Consider a principal T𝔫T^{\mathfrak{n}}-bundle Ο€:M→ℬ0\pi:M\to\mathcal{B}^{0} over an open set ℬ0\mathcal{B}^{0} in π”±βˆ—Γ—Y\mathfrak{t}^{*}\times Y, whose first Chern class is an element c1∈H2​(ℬ0,𝔱℀)c_{1}\in H^{2}(\mathcal{B}^{0},\mathfrak{t}_{\mathbb{Z}}). Later we will partially compactify MM into a singular T𝔫T^{\mathfrak{n}}-bundle. Summation convention will be used throughout.

Theorem 1.5.

(cf. Theorem 2.1 in [31]) Let Vi​jV^{ij}, respectively Wp​qΒ―W^{p\bar{q}}, be real symmetric positive definite/Hermitian matrices of smooth functions on ℬ0\mathcal{B}^{0}, locally given by some potential function Ξ¦\Phi:

(1.5) Vi​j=βˆ‚2Ξ¦βˆ‚ΞΌiβ€‹βˆ‚ΞΌj,Wp​qΒ―=βˆ’4β€‹βˆ‚2Ξ¦βˆ‚Ξ·pβ€‹βˆ‚Ξ·Β―q,1≀i,j≀𝔫,1≀p,q≀Nβˆ’π”«.V^{ij}=\frac{\partial^{2}\Phi}{\partial\mu_{i}\partial\mu_{j}},\quad W^{p\bar{q}}=-4\frac{\partial^{2}\Phi}{\partial\eta_{p}\partial\bar{\eta}_{q}},\quad 1\leq i,j\leq\mathfrak{n},\quad 1\leq p,q\leq N-\mathfrak{n}.

Then the following 𝔱\mathfrak{t}-valued real 2-form is closed:

(1.6) Fj=βˆ’1​(12β€‹βˆ‚Wp​qΒ―βˆ‚ΞΌj​d​ηp∧d​η¯q+βˆ‚Vi​jβˆ‚Ξ·p​d​μi∧d​ηpβˆ’βˆ‚Vi​jβˆ‚Ξ·Β―q​d​μi∧d​η¯q).F_{j}=\sqrt{-1}\left(\frac{1}{2}\frac{\partial W^{p\bar{q}}}{\partial\mu_{j}}d\eta_{p}\wedge d\bar{\eta}_{q}+\frac{\partial V^{ij}}{\partial\eta_{p}}d\mu_{i}\wedge d\eta_{p}-\frac{\partial V^{ij}}{\partial\bar{\eta}_{q}}d\mu_{i}\wedge d\bar{\eta}_{q}\right).

Suppose further that 12​π​(F1,…,F𝔫)\frac{1}{2\pi}(F_{1},\ldots,F_{\mathfrak{n}}) is in the cohomology class c1∈H2​(ℬ0,𝔱℀)c_{1}\in H^{2}(\mathcal{B}^{0},\mathfrak{t}_{\mathbb{Z}}). Then there exists a connection Ο‘\vartheta on the principal bundle M→ℬ0M\to\mathcal{B}^{0} with curvature d​ϑi=Fid\vartheta_{i}=F_{i} for i=1,…,𝔫i=1,\ldots,\mathfrak{n}, such that MM is a KΓ€hler manifold with metric tensor

(1.7) h=(Vβˆ’1)i​j​΢iβŠ—ΞΆΒ―j+Wp​q¯​d​ηpβŠ—d​η¯q,Ο‰=d​μjβˆ§Ο‘j+βˆ’12​Wp​q¯​d​ηp∧d​η¯q,h=(V^{-1})^{ij}\zeta_{i}\otimes\bar{\zeta}_{j}+W^{p\bar{q}}d\eta_{p}\otimes d\bar{\eta}_{q},\quad\omega=d\mu_{j}\wedge\vartheta_{j}+\frac{\sqrt{-1}}{2}W^{p\bar{q}}d\eta_{p}\wedge d\bar{\eta}_{q},

where ΞΆj=Vi​j​d​μi+βˆ’1​ϑj\zeta_{j}=V^{ij}d\mu_{i}+\sqrt{-1}\vartheta_{j} and Ξ·p\eta_{p} form a basis of type (1,0) forms which defines an integrable complex structure. There is a nowhere vanishing holomorphic form on MM:

(1.8) Ξ©=∧j=1𝔫(βˆ’βˆ’1ΞΆj)β‹€βˆ§p=1Nβˆ’π”«dΞ·p.\Omega=\wedge_{j=1}^{\mathfrak{n}}(-\sqrt{-1}\zeta_{j})\bigwedge\wedge_{p=1}^{N-\mathfrak{n}}d\eta_{p}.

The Calabi-Yau condition Ο‰N=N!2Nβ€‹βˆ’1N2β€‹Ξ©βˆ§Ξ©Β―\omega^{N}=\frac{N!}{2^{N}}\sqrt{-1}^{N^{2}}\Omega\wedge\overline{\Omega} is equivalent to the equation

(1.9) det(Vi​j)=det(Wp​qΒ―).\det(V^{ij})=\det(W^{p\bar{q}}).
Remark 1.5.

The inverse matrix (Vβˆ’1)i​j(V^{-1})^{ij} describes the metric restricted to the torus fibres, and the matrix Wp​qΒ―W^{p\bar{q}} describes the metric induced on the KΓ€hler quotients. This viewpoint is taken by Pedersen and Poon [23], whose argument shows that Calabi-Yau manifolds with Hamiltonian torus symmetries necessarily arise from this construction locally. The local existence of the potential Ξ¦\Phi is equivalent to the linear integrability condition

(1.10) βˆ‚Vi​jβˆ‚ΞΌk=βˆ‚Vi​kβˆ‚ΞΌj,βˆ‚Wp​qΒ―βˆ‚Ξ·r=βˆ‚Wr​qβˆ‚Ξ·p,βˆ‚Wp​rβˆ‚Ξ·Β―q=βˆ‚Wp​qΒ―βˆ‚Ξ·Β―r,\frac{\partial V^{ij}}{\partial\mu_{k}}=\frac{\partial V^{ik}}{\partial\mu_{j}},\quad\frac{\partial W^{p\bar{q}}}{\partial\eta_{r}}=\frac{\partial W^{rq}}{\partial\eta_{p}},\quad\frac{\partial W^{pr}}{\partial\bar{\eta}_{q}}=\frac{\partial W^{p\bar{q}}}{\partial\bar{\eta}_{r}},

and

(1.11) βˆ‚2Wp​qΒ―βˆ‚ΞΌiβ€‹βˆ‚ΞΌj+4β€‹βˆ‚2Vi​jβˆ‚Ξ·pβ€‹βˆ‚Ξ·Β―q=0.\frac{\partial^{2}W^{p\bar{q}}}{\partial\mu_{i}\partial\mu_{j}}+4\frac{\partial^{2}V^{ij}}{\partial\eta_{p}\partial\bar{\eta}_{q}}=0.

In particular when N=2,𝔫=1N=2,\mathfrak{n}=1, the Calabi-Yau condition is V=WV=W and we recover the usual Gibbons-Hawking equation from (1.11).

Proof.

(Theorem 1.5, sketch) By formula (1.6) and the integrability condition (1.10),

(1.12) d​Fj=βˆ’12​(βˆ‚2Wp​qΒ―βˆ‚ΞΌiβ€‹βˆ‚ΞΌj+4β€‹βˆ‚2Vi​jβˆ‚Ξ·pβ€‹βˆ‚Ξ·Β―q)​d​μi∧d​ηp∧d​η¯q,dF_{j}=\frac{\sqrt{-1}}{2}\left(\frac{\partial^{2}W^{p\bar{q}}}{\partial\mu_{i}\partial\mu_{j}}+4\frac{\partial^{2}V^{ij}}{\partial\eta_{p}\partial\bar{\eta}_{q}}\right)d\mu_{i}\wedge d\eta_{p}\wedge d\bar{\eta}_{q},

so the closedness of FjF_{j} is equivalent to (1.11). Since 12​π​F\frac{1}{2\pi}F represents the appropriate first Chern class, FF must be the curvature of a T𝔫T^{\mathfrak{n}}-connection Ο‘\vartheta. Modulo gauge Ο‘\vartheta admits the local formula

(1.13) Ο‘j=βˆ’1​{βˆ‚2Ξ¦βˆ‚Ξ·pβ€‹βˆ‚ΞΌβ€‹d​ηpβˆ’βˆ‚2Ξ¦βˆ‚Ξ·Β―pβ€‹βˆ‚ΞΌβ€‹d​η¯p}.\vartheta_{j}=\sqrt{-1}\{\frac{\partial^{2}\Phi}{\partial\eta_{p}\partial\mu}d\eta_{p}-\frac{\partial^{2}\Phi}{\partial\bar{\eta}_{p}\partial\mu}d\bar{\eta}_{p}\}.

Gauge equivalent choices of the connection define the structures on MM up to holomorphic isometry.

The integrability of the complex structure follows from the fact that the differential ideal generated by (1,0)(1,0) forms is closed:

(1.14) d​΢j=(12β€‹βˆ‚Wp​qΒ―βˆ‚ΞΌj​d​η¯qβˆ’2β€‹βˆ‚Vi​jβˆ‚Ξ·p​d​μi)∧d​ηp,d\zeta_{j}=\left(\frac{1}{2}\frac{\partial W^{p\bar{q}}}{\partial\mu_{j}}d\bar{\eta}_{q}-2\frac{\partial V^{ij}}{\partial\eta_{p}}d\mu_{i}\right)\wedge d\eta_{p},

using (1.10)(1.6) and the definition of ΞΆj\zeta_{j}. The KΓ€hler condition d​ω=0d\omega=0 follows from (1.10)(1.6). The Calabi-Yau condition follows from the more general formula

Ο‰N=det(Wp​qΒ―)​det(Vi​j)βˆ’1​N!2Nβ€‹βˆ’1N2β€‹Ξ©βˆ§Ξ©Β―.\omega^{N}=\det(W^{p\bar{q}})\det(V^{ij})^{-1}\frac{N!}{2^{N}}\sqrt{-1}^{N^{2}}\Omega\wedge\overline{\Omega}.

∎

Remark 1.6.

If βˆ‚βˆ‚ΞΈj\frac{\partial}{\partial\theta_{j}} are the Hamiltonian vector fields dual to Ο‘i\vartheta_{i}, namely Ο‘i​(βˆ‚βˆ‚ΞΈj)=Ξ΄i​j\vartheta_{i}(\frac{\partial}{\partial\theta_{j}})=\delta_{ij}, then d​μi=βˆ’ΞΉβˆ‚βˆ‚ΞΈi​ωd\mu_{i}=-\iota_{\frac{\partial}{\partial\theta_{i}}}\omega, namely ΞΌi\mu_{i} are the symplectic moment coordinates up to sign. When Nβˆ’π”«=1N-\mathfrak{n}=1, namely there is only one Ξ·\eta coordinate, then dΞ·=Ξ©(βˆ‚βˆ‚ΞΈ1,…,βˆ‚βˆ‚ΞΈπ”«,β‹…)d\eta=\Omega(\frac{\partial}{\partial\theta_{1}},\ldots,\frac{\partial}{\partial\theta_{\mathfrak{n}}},\cdot), and accordingly we refer to Ξ·\eta as the holomorphic moment coordinate. In this situation MM admits a fibration

Mβ†’(ΞΌ1,…,μ𝔫,Im​(Ξ·))π”±βˆ—Γ—β„,M\xrightarrow{(\mu_{1},\ldots,\mu_{\mathfrak{n}},\text{Im}(\eta))}\mathfrak{t}^{*}\times\mathbb{R},

where fibres are special Lagrangians with phase zero: the Lagrangian condition follows from d​μi=βˆ’ΞΉβˆ‚βˆ‚ΞΈi​ω=0d\mu_{i}=-\iota_{\frac{\partial}{\partial\theta_{i}}}\omega=0 on fibres, while the special condition is equivalent to ImΞ©(βˆ‚βˆ‚ΞΈ1,β€¦βˆ‚βˆ‚ΞΈπ”«,β‹…)=dImΞ·=0\text{Im}\Omega(\frac{\partial}{\partial\theta_{1}},\ldots\frac{\partial}{\partial\theta_{\mathfrak{n}}},\cdot)=d\text{Im}\eta=0.

Remark 1.7.

The information contained in the generalised Gibbons-Hawking ansatz can be encoded by a Riemannian metric on the base, written in distinguished coordinates as

(1.15) gℬ0=Vi​j​d​μiβŠ—d​μj+Re​(Wp​q¯​d​ηpβŠ—d​η¯q),g_{\mathcal{B}^{0}}=V^{ij}d\mu_{i}\otimes d\mu_{j}+\text{Re}(W^{p\bar{q}}d\eta_{p}\otimes d\bar{\eta}_{q}),

such that the map M→ℬ0M\to\mathcal{B}^{0} is a Riemannian submersion.

Remark 1.8.

There is an additional freedom to twist the connection Ο‘\vartheta by a flat connection. Up to gauge equivalence, the choice of Ο‘\vartheta is parametrised by H1​(ℬ0,Tn)H^{1}(\mathcal{B}^{0},T^{n}).

1.2.1. Elementary examples

Example 1.6.

(Constant solution) The simplest solution is where Vi​jV^{ij} and Wp​qΒ―W^{p\bar{q}} are independent of the base variables and satisfy (1.9). We shall see that many interesting solutions can be thought heuristically as perturbation of the constant solution after introducing some topology. Some important special cases for us are:

  • β€’

    N=2,𝔫=1N=2,\mathfrak{n}=1, V=W=A>0V=W=A>0. The subcase where Ξ·\eta takes value in β„‚\mathbb{C} is relevant for the Taub-NUT metric (cf. Example 1.8), and the subcase where Ξ·\eta is a periodic variable is relevant for the Ooguri-Vafa metric (cf. Section 1.3). In the periodic case the choice of the connection Ο‘\vartheta is parametrised by H1​(S1×ℝ2,S1)≃S1H^{1}(S^{1}\times\mathbb{R}^{2},S^{1})\simeq S^{1}.

  • β€’

    N=3,𝔫=2N=3,\mathfrak{n}=2, Vi​j=ai​jV^{ij}=a_{ij} is symmetric positive definite, and W=A=det(ai​j)W=A=\det(a_{ij}). We write ga=ai​j​d​μi​d​μj+A​|d​η|2g_{a}=a_{ij}d\mu_{i}d\mu_{j}+A|d\eta|^{2}. The subcase where Ξ·\eta takes value in β„‚\mathbb{C} will be relevant for constructing new Taub-NUT type Calabi-Yau metrics on β„‚3\mathbb{C}^{3}, and the subcase where Ξ·\eta is a periodic variable will be relevant for the positive vertex. In the periodic case the choice of the connection Ο‘\vartheta is parametrised by H1​(S1×ℝ3,T2)≃T2H^{1}(S^{1}\times\mathbb{R}^{3},T^{2})\simeq T^{2}.

  • β€’

    N=3,𝔫=1N=3,\mathfrak{n}=1, Wp​qΒ―=ap​qΒ―W^{p\bar{q}}=a_{p\bar{q}} is Hermitian, and V=A=det(ap​qΒ―)V=A=\det(a_{p\bar{q}}). We write ga=Re​(ap​q¯​d​ηp​d​η¯q)+A​dβ€‹ΞΌβŠ—d​μg_{a}=\text{Re}(a_{p\bar{q}}d\eta_{p}d\bar{\eta}_{q})+Ad\mu\otimes d\mu. Here Ξ·1,Ξ·2\eta_{1},\eta_{2} are periodic coordinates with period 1. The choice of the connection Ο‘\vartheta is parametrised by H1​(T2×ℝ3,S1)≃T2H^{1}(T^{2}\times\mathbb{R}^{3},S^{1})\simeq T^{2}. This case will be relevant for the negative vertex. Notice that if we demand that the fibration on MM induced by ΞΌ,Im​(Ξ·1),Im​(Ξ·2)\mu,\text{Im}(\eta_{1}),\text{Im}(\eta_{2}) is a special Lagrangian fibration with phase zero, then we would need a1​2Β―=a2​1Β―a_{1\bar{2}}=a_{2\bar{1}}, namely ap​q=ap​qΒ―a_{pq}=a_{p\bar{q}} is symmetric.

Example 1.7.

(β„‚N\mathbb{C}^{N} Harvey-Lawson example) The affine space β„‚N\mathbb{C}^{N} with the standard Euclidean metric Ο‰=βˆ’12β€‹βˆ‘i=0Nβˆ’1d​zi∧d​zΒ―i\omega=\frac{\sqrt{-1}}{2}\sum_{i=0}^{N-1}dz_{i}\wedge d\bar{z}_{i} and holomorphic volume form Ξ©=βˆ’1Nβˆ’1​d​z0βˆ§β€¦β€‹d​zNβˆ’1\Omega=\sqrt{-1}^{N-1}dz_{0}\wedge\ldots dz_{N-1} admits a diagonal TNβˆ’1T^{N-1}-action, where the kk-th circle factor acts by

ei​θkβ‹…(z0,z1,…,zNβˆ’1)=(eβˆ’i​θk​z0,z1,…,ei​θk​zk,zk+1,…,zNβˆ’1).e^{i\theta_{k}}\cdot(z_{0},z_{1},\ldots,z_{N-1})=(e^{-i\theta_{k}}z_{0},z_{1},\ldots,e^{i\theta_{k}}z_{k},z_{k+1},\ldots,z_{N-1}).

The corresponding moment coordinates are

ΞΌi=12(|zi|2βˆ’|z0|2),i=1,2,…Nβˆ’1,Β andΒ Ξ·=z0z1…zNβˆ’1.\mu_{i}=\frac{1}{2}(|z_{i}|^{2}-|z_{0}|^{2}),\quad i=1,2,\ldots N-1,\text{ and }\eta=z_{0}z_{1}\ldots z_{N-1}.

This defines a TNβˆ’1T^{N-1}-bundle away from the singular locus ⋃{zi=zj=0}\bigcup\{z_{i}=z_{j}=0\}. Special cases include (1.1)(1.3). The inverse matrices are

(Vβˆ’1)i​j=|z0|2+Ξ΄i​j​|zi|2,Wβˆ’1=|z0​z1​…​zNβˆ’1|2​(1|z0|2+…+1|zNβˆ’1|2),(V^{-1})^{ij}=|z_{0}|^{2}+\delta_{ij}|z_{i}|^{2},\quad W^{-1}=|z_{0}z_{1}\ldots z_{N-1}|^{2}\left(\frac{1}{|z_{0}|^{2}}+\ldots+\frac{1}{|z_{N-1}|^{2}}\right),

viewed as functions of ΞΌi\mu_{i} and Ξ·\eta. The special Lagrangian fibration described by Remark 1.6 is the well known Harvey-Lawson example.

We notice in particular when N=3N=3 that the discriminant locus of the singular T2T^{2}-bundle is given by π”‡βŠ‚β„ΞΌ1,ΞΌ22Γ—{0}βŠ‚β„ΞΌ1,ΞΌ22Γ—β„‚Ξ·\mathfrak{D}\subset\mathbb{R}^{2}_{\mu_{1},\mu_{2}}\times\{0\}\subset\mathbb{R}^{2}_{\mu_{1},\mu_{2}}\times\mathbb{C}_{\eta} as in (1.2). This is not an accidental feature of the Euclidean metric:

Lemma 1.6.

Let β„‚3\mathbb{C}^{3} be equipped with the holomorphic volume form Ξ©\Omega above, and let Ο‰\omega be any T2T^{2}-invariant KΓ€hler form with infinite volume on the singular loci {zi=zj=0}\{z_{i}=z_{j}=0\} for any i,ji,j. Then the discriminant locus of the singular T2T^{2}-bundle is π”‡βŠ‚β„ΞΌ1,ΞΌ22Γ—β„‚Ξ·\mathfrak{D}\subset\mathbb{R}^{2}_{\mu_{1},\mu_{2}}\times\mathbb{C}_{\eta} in moment coordinates ΞΌ1,ΞΌ2\mu_{1},\mu_{2} and Ξ·\eta.

Proof.

The discriminant locus is the image of the singular locus π’ži​j={zi=zj=0}\mathcal{C}_{ij}=\{z_{i}=z_{j}=0\} under the moment map. We shall focus on π’ž01\mathcal{C}_{01}. The holomorphic moment coordinate Ξ·\eta depends only on Ξ©\Omega and the T2T^{2} action, so Ξ·=z0​z1​z2\eta=z_{0}z_{1}z_{2} as before and vanishes on π’ž01\mathcal{C}_{01}. The symplectic moment coordinates are defined by d​μi=βˆ’ΞΉβˆ‚βˆ‚ΞΈi​ωd\mu_{i}=-\iota_{\frac{\partial}{\partial\theta_{i}}}\omega, and are normalised to be zero at (z1,z2,z3)=0(z_{1},z_{2},z_{3})=0. In particular since the Hamiltonian vector field βˆ‚βˆ‚ΞΈ1\frac{\partial}{\partial\theta_{1}} vanishes on π’ž01\mathcal{C}_{01}, the moment ΞΌ1\mu_{1} must be the constant zero on π’ž01\mathcal{C}_{01}. Furthermore ΞΌ2>0\mu_{2}>0 on π’ž01\mathcal{C}_{01} by considering the weight of the remaining S1S^{1} action at the fixed point, so the image of π’ž01\mathcal{C}_{01} is contained in 𝔇1βˆͺ{0}\mathfrak{D}_{1}\cup\{0\}. The infinite volume condition and the formula

0β‰€βˆ«π’ž01∩{ΞΌ2<m}Ο‰=βˆ’2Ο€βˆ«ΞΌ2=m0dΞΌ2=2Ο€m,βˆ€mβ‰₯0,0\leq\int_{\mathcal{C}_{01}\cap\{\mu_{2}<m\}}\omega=-2\pi\int_{\mu_{2}=m}^{0}d\mu_{2}=2\pi m,\quad\forall m\geq 0,

ensure that ΞΌ2\mu_{2} stretches to infinity, so 𝔇1βˆͺ{0}\mathfrak{D}_{1}\cup\{0\} is the image of π’ž01\mathcal{C}_{01}. Likewise the image of π’ž02\mathcal{C}_{02} is 𝔇2βˆͺ{0}\mathfrak{D}_{2}\cup\{0\} and the image of π’ž12\mathcal{C}_{12} is 𝔇3βˆͺ{0}\mathfrak{D}_{3}\cup\{0\}. ∎

Remark 1.9.

The same method shows the complex geometry of the positive vertex in Section 1.1.6 is compatible with the discriminant locus described in Section 1.1.3.

Example 1.8.

(Taub-NUT) We take N=2,𝔫=1N=2,\mathfrak{n}=1, and V=W=12​μ2+|Ξ·|2+AV=W=\frac{1}{2\sqrt{\mu^{2}+|\eta|^{2}}}+A, where AA is a positive constant. This defines a Calabi-Yau metric whose asymptotic geometry at infinity approaches the constant solution (cf. Example 1.6), with asymptotic circles of length 2​πA\frac{2\pi}{\sqrt{A}} fibred over a flat 3-dimensional base. Different choices of AA define the same metric up to scaling. The first Chern class c1c_{1} of the S1S^{1}-bundle over (ℝμ×ℂη)βˆ–{0}(\mathbb{R}_{\mu}\times\mathbb{C}_{\eta})\setminus\{0\} evaluates to βˆ’1-1 on any sphere around the origin in ℝμ×ℂη\mathbb{R}_{\mu}\times\mathbb{C}_{\eta}; equivalently, the 3-current βˆ’12​π​d​F-\frac{1}{2\pi}dF is represented by the origin 0βˆˆβ„ΞΌΓ—β„‚Ξ·0\in\mathbb{R}_{\mu}\times\mathbb{C}_{\eta} viewed as a codimension 3 cycle. Written in terms of the delta function,

12​π​d​F=12​π​(βˆ‚2βˆ‚ΞΌ2+4β€‹βˆ‚2βˆ‚Ξ·β€‹βˆ‚Ξ·Β―)​V​dβ€‹ΞΌβˆ§d​Reβ€‹Ξ·βˆ§d​Im​η=βˆ’Ξ΄β‘(ΞΌ,Ξ·,Ξ·Β―)​dβ€‹ΞΌβˆ§d​Reβ€‹Ξ·βˆ§d​Im​η.\frac{1}{2\pi}dF=\frac{1}{2\pi}(\frac{\partial^{2}}{\partial\mu^{2}}+4\frac{\partial^{2}}{\partial\eta\partial\bar{\eta}})Vd\mu\wedge d\text{Re}\eta\wedge d\text{Im}{\eta}=-\delta(\mu,\eta,\bar{\eta})d\mu\wedge d\text{Re}\eta\wedge d\text{Im}{\eta}.

From the holomorphic perspective, LeBrun [17] observes that the Taub-NUT space is biholomorphic to β„‚2\mathbb{C}^{2}. To see this, recall ΞΆ=V​d​μ+βˆ’1​ϑ\zeta=Vd\mu+\sqrt{-1}\vartheta and notice the (1,0) form ΞΆβˆ’ΞΌ2​η​μ2+|Ξ·|2​d​η\zeta-\frac{\mu}{2\eta\sqrt{\mu^{2}+|\eta|^{2}}}d\eta is closed, so locally is the differential of a holomorphic function. The line integrals

logz1=∫΢+(12β€‹Ξ·βˆ’ΞΌ2​η​μ2+|Ξ·|2)dΞ·,logz0=βˆ«βˆ’ΞΆ+(12​η+ΞΌ2​η​μ2+|Ξ·|2)dΞ·\log z_{1}=\int\zeta+(\frac{1}{2\eta}-\frac{\mu}{2\eta\sqrt{\mu^{2}+|\eta|^{2}}})d\eta,\quad\log z_{0}=\int-\zeta+(\frac{1}{2\eta}+\frac{\mu}{2\eta\sqrt{\mu^{2}+|\eta|^{2}}})d\eta

define holomorphic functions up to 2β€‹Ο€β€‹βˆ’1​℀2\pi\sqrt{-1}\mathbb{Z} over the regions β„Γ—β„‚βˆ–{Ξ·=0,μ≀0}\mathbb{R}\times\mathbb{C}\setminus\{\eta=0,\mu\leq 0\} and β„Γ—β„‚βˆ–{Ξ·=0,ΞΌβ‰₯0}\mathbb{R}\times\mathbb{C}\setminus\{\eta=0,\mu\geq 0\} respectively, so z1z_{1} and z0z_{0} are well defined over the respective regions. Since d​log⁑z1+d​log⁑z0=d​log⁑ηd\log z_{1}+d\log z_{0}=d\log\eta we can normalise z0,z1z_{0},z_{1} to satisfy the functional equation z1​z0=Ξ·z_{1}z_{0}=\eta, whence z1z_{1} and z0z_{0} extend as global holomorphic functions. These coordinates exhibit the biholomorphism to β„‚2\mathbb{C}^{2}. By considering the Hamiltonian vector field acting on log⁑z1,log⁑z0\log z_{1},\log z_{0}, we idenitfy the S1S^{1} action as

ei​θ⋅(z1,z0)=(ei​θ​z1,eβˆ’i​θ​z0).e^{i\theta}\cdot(z_{1},z_{0})=(e^{i\theta}z_{1},e^{-i\theta}z_{0}).

The holomorphic volume form is

Ξ©=βˆ’βˆ’1β€‹ΞΆβˆ§d​η=βˆ’βˆ’1​d​log⁑z1∧d⁑(z1​z0)=βˆ’1​d​z0∧d​z1.\Omega=-\sqrt{-1}\zeta\wedge d\eta=-\sqrt{-1}d\log z_{1}\wedge d(z_{1}z_{0})=\sqrt{-1}dz_{0}\wedge dz_{1}.

Thus Ξ·=z0​z1\eta=z_{0}z_{1} defines holomorphic fibration of β„‚2\mathbb{C}^{2} by affine quadrics. The generic quadric fibre is topologically a cylinder, and metrically is also approaching the flat cylindrical metric near spatial infinity. When Ξ·=0\eta=0, the quadric fibre degenerates into a union of two complex lines with simple normal crossing, where each line looks metrically like a cylinder with one capped end.

1.2.2. Compactification and distributional equation

When we partially compactify the principal TnT^{n}-bundle over ℬ0\mathcal{B}^{0} to a singular T𝔫T^{\mathfrak{n}}-bundle over β„¬βŠƒβ„¬0\mathcal{B}\supset\mathcal{B}^{0} by allowing torus fibres to degenerate, we need to encode the topology into the generalised Gibbons-Hawking ansatz, by changing the RHS of (1.11) into a distributional term reflecting the nontriviality of the first Chern class (cf. the Taub-NUT example 1.8). This has been worked out by Zharkov [31] in general dimensions; here we will focus on the vertices in N=3N=3.

Example 1.9.

(Positive vertex and Taub-NUT type β„‚3\mathbb{C}^{3}) Recall from Section 1.1.3 that e1,e2e_{1},e_{2} are the homology classes of the two circle factors in the T2T^{2}-fibre, or equivalently an integral basis in 𝔱\mathfrak{t}. The 𝔱\mathfrak{t}-valued curvature 2-form F=F1​e1+F2​e2F=F_{1}e_{1}+F_{2}e_{2} satisfies (cf. (1.12))

12​π​d​FjβŠ—ej=βˆ’14​π​(βˆ‚2Wβˆ‚ΞΌiβ€‹βˆ‚ΞΌj+4β€‹βˆ‚2Vi​jβˆ‚Ξ·β€‹βˆ‚Ξ·Β―)​d​μi∧dβ€‹Ξ·βˆ§dβ€‹Ξ·Β―βŠ—ej,\frac{1}{2\pi}dF_{j}\otimes e_{j}=\frac{\sqrt{-1}}{4\pi}\left(\frac{\partial^{2}W}{\partial\mu_{i}\partial\mu_{j}}+4\frac{\partial^{2}V^{ij}}{\partial\eta\partial\bar{\eta}}\right)d\mu_{i}\wedge d\eta\wedge d\bar{\eta}\otimes e_{j},

which is a 𝔱\mathfrak{t}-valued 3-current supported on the codimension 3 discriminant locus π”‡βŠ‚β„ΞΌ1,ΞΌ22Γ—(S1×ℝ)Ξ·\mathfrak{D}\subset\mathbb{R}^{2}_{\mu_{1},\mu_{2}}\times(S^{1}\times\mathbb{R})_{\eta}. Now take small 3-balls transverse to 𝔇1,𝔇2,𝔇3\mathfrak{D}_{1},\mathfrak{D}_{2},\mathfrak{D}_{3} respectively. The integrals of 12​π​d​FjβŠ—ej\frac{1}{2\pi}dF_{j}\otimes e_{j} over the balls are equal to the integrals of the Chern class representative 12​π​F\frac{1}{2\pi}F over the S2S^{2} linking 𝔇1,𝔇2,𝔇3\mathfrak{D}_{1},\mathfrak{D}_{2},\mathfrak{D}_{3}, which by Section 1.1.3 are e1,βˆ’e2,βˆ’e1+e2e_{1},-e_{2},-e_{1}+e_{2} up to orientation issues. Thus

(1.16) βˆ’14​π​(βˆ‚2Wβˆ‚ΞΌiβ€‹βˆ‚ΞΌj+4β€‹βˆ‚2Vi​jβˆ‚Ξ·β€‹βˆ‚Ξ·Β―)​d​μi∧dβ€‹Ξ·βˆ§dβ€‹Ξ·Β―βŠ—ej=𝔇1βŠ—e1βˆ’π”‡2βŠ—e2+𝔇3βŠ—(e2βˆ’e1),\frac{\sqrt{-1}}{4\pi}\left(\frac{\partial^{2}W}{\partial\mu_{i}\partial\mu_{j}}+4\frac{\partial^{2}V^{ij}}{\partial\eta\partial\bar{\eta}}\right)d\mu_{i}\wedge d\eta\wedge d\bar{\eta}\otimes e_{j}=\mathfrak{D}_{1}\otimes e_{1}-\mathfrak{D}_{2}\otimes e_{2}+\mathfrak{D}_{3}\otimes(e_{2}-e_{1}),

where the RHS is a 𝔱\mathfrak{t}-valued codimension 3 cycle. The orientation here is decided by comparing with the Taub-NUT example. If d​μ1∧d​μ2∧d​Reβ€‹Ξ·βˆ§d​Im​ηd\mu_{1}\wedge d\mu_{2}\wedge d\text{Re}\eta\wedge d\text{Im}\eta is an orientation form on ℝμ1,ΞΌ22Γ—(S1×ℝ)Ξ·\mathbb{R}^{2}_{\mu_{1},\mu_{2}}\times(S^{1}\times\mathbb{R})_{\eta}, then the orientation forms on 𝔇1,𝔇2,𝔇3\mathfrak{D}_{1},\mathfrak{D}_{2},\mathfrak{D}_{3} are d​μ2,d​μ1,βˆ’d​μ1d\mu_{2},d\mu_{1},-d\mu_{1}, compatible with the directions pointing to infinity.

In the variant situation where Ξ·βˆˆβ„‚\eta\in\mathbb{C} instead of being periodic, to which previous discussions still apply, the generalised Gibbons-Hawking ansatz has a scaling symmetry compatible with the distributional equation (1.16): a new solution ΦΛ\Phi_{\Lambda} may be constructed from an old solution Ξ¦\Phi by

(1.17) {ΦΛ​(ΞΌ1,ΞΌ2,Ξ·)=Ξ›βˆ’1​Φ​(Λ​μ1,Λ​μ2,Ξ›1.5​η),VΞ›i​j​(ΞΌ1,ΞΌ2,Ξ·)=Λ​Vi​j​(Λ​μ1,Λ​μ2,Ξ›1.5​η),WΛ​(ΞΌ1,ΞΌ2,Ξ·)=Ξ›2​W​(Λ​μ1,Λ​μ2,Ξ›1.5​η)\begin{cases}\Phi_{\Lambda}(\mu_{1},\mu_{2},\eta)=\Lambda^{-1}\Phi(\Lambda\mu_{1},\Lambda\mu_{2},\Lambda^{1.5}\eta),\\ V^{ij}_{\Lambda}(\mu_{1},\mu_{2},\eta)=\Lambda V^{ij}(\Lambda\mu_{1},\Lambda\mu_{2},\Lambda^{1.5}\eta),\\ W_{\Lambda}(\mu_{1},\mu_{2},\eta)=\Lambda^{2}W(\Lambda\mu_{1},\Lambda\mu_{2},\Lambda^{1.5}\eta)\end{cases}

These solutions are isometric up to a scaling factor, analogous to Taub-NUT metrics with different asymptotic circle lengths. The presence of the periodicty condition (or more abstractly an integral lattice structure) breaks down scaling symmetry by singling out a special scale.

Example 1.10.

(Negative vertex) By a similar argument, in the negative vertex setting (cf. Section 1.1.5) the curvature 2-form FF satisfies

(1.18) βˆ’12​π​d​F=βˆ’βˆ’14​π​(βˆ‚2Wp​qΒ―βˆ‚ΞΌβ€‹βˆ‚ΞΌ+4β€‹βˆ‚2Vβˆ‚Ξ·pβ€‹βˆ‚Ξ·Β―q)​dβ€‹ΞΌβˆ§d​ηp∧d​η¯q=S,-\frac{1}{2\pi}dF=-\frac{\sqrt{-1}}{4\pi}\left(\frac{\partial^{2}W^{p\bar{q}}}{\partial\mu\partial\mu}+4\frac{\partial^{2}V}{\partial\eta_{p}\partial\bar{\eta}_{q}}\right)d\mu\wedge d\eta_{p}\wedge d\bar{\eta}_{q}=S,

where S={z1+z2=1}={e2​π​i​η1+e2​π​i​η2=1}βŠ‚β„‚z1βˆ—Γ—β„‚z2βˆ—Γ—{0}βŠ‚β„‚z1βˆ—Γ—β„‚z2βˆ—Γ—β„ΞΌS=\{z_{1}+z_{2}=1\}=\{e^{2\pi i\eta_{1}}+e^{2\pi i\eta_{2}}=1\}\subset\mathbb{C}^{*}_{z_{1}}\times\mathbb{C}^{*}_{z_{2}}\times\{0\}\subset\mathbb{C}^{*}_{z_{1}}\times\mathbb{C}^{*}_{z_{2}}\times\mathbb{R}_{\mu} defines a codimension 3 cycle. Here SS is endowed with the complex orientation, and the orientation on β„‚z2βˆ—Γ—β„ΞΌ\mathbb{C}^{*}_{z_{2}}\times\mathbb{R}_{\mu} is defined by the form dβ€‹ΞΌβˆ§d​Re​η1∧d​Im​η1∧d​Re​η2∧d​Im​η2d\mu\wedge d\text{Re}\eta_{1}\wedge d\text{Im}\eta_{1}\wedge d\text{Re}\eta_{2}\wedge d\text{Im}\eta_{2}.

Original mathematics by the credited authors. Source-backed reader collection; mathematical self-containment is not assessed.