betaFeedback

WorksPhysics and Astronomy

Satisfying Freed–Witten quantization on smooth divisors in Calabi–Yau orientifolds

Abstract

Freed–Witten quantization conditions on several divisors wrapped by D-branes must be satisfied with one bulk BB-field. On a smooth Calabi–Yau threefold XX with a holomorphic O3/O7 involution σσ, we show that the common background B2,+⋆=12[R]B_{2,+}^{\star}=\tfrac12[R], where RR is the total O7 divisor, permits Freed–Witten period quantization on the full two-cycle lattice of every smooth connected divisor, modulo torsion. The worldvolume fluxes are chosen individually and obey the orientifold relation between each divisor and its image. The central step concerns setwise invariant divisors DD: we prove that the bulk Freed–Witten class ηD⋆=12([R]+[D])∣Dη_D^\star=\tfrac12([R]+[D])|_D has integer periods on every σσ-even integral two-cycle class, including inherited and autochthonous classes. Its periods on purely σσ-odd classes vanish, so any remaining non-integer periods occur only on classes that are neither σσ-even nor σσ-odd. We show how to choose σσ-odd worldvolume fluxes on these invariant divisors that make all periods integral. For exchanged divisor pairs, quantizing the flux on one member automatically quantizes its orientifold image. On invariant divisors with h−2,0(D)=0h^{2,0}_-(D)=0, the constructed flux is of type (1,1)(1,1) and admits a primitive representative; this includes every smooth rigid invariant divisor. We also study zero worldvolume flux on invariant divisors. For h−1,1(X)≤1h^{1,1}_-(X)\leq1, we construct common bulk backgrounds satisfying every inherited intersection period on all such divisors. A binary compatibility test determines whether a common shift exists for larger h−1,1(X)h^{1,1}_-(X). Non-inherited period tests then determine whether the resulting backgrounds allow zero flux on the full worldvolume lattices.

Introduction

Freed–Witten anomaly cancellation is a necessary condition for a consistent D-brane configuration [1]. In Type IIB compactifications on a Calabi–Yau threefold XX with a holomorphic O3/O7 involution σ\sigma, the associated worldvolume flux quantization conditions apply both to spacetime-filling D7-branes and to Euclidean D3-branes wrapping divisors.

Our main motivation comes from Euclidean D3-brane instantons, which can generate nonperturbative potentials for Kähler moduli and their axionic partners. Whether a given divisor contributes depends on several conditions: its worldvolume flux must be consistently quantized, the orientifold projection must leave the required universal fermion zero modes, and additional neutral or charged modes must not obstruct the amplitude [2, 3]. These conditions are logically distinct. In this paper we address the Freed–Witten period quantization condition. Choices of bulk BB-field and consistently quantized worldvolume fluxes have been studied in fluxed-instanton and D7-brane constructions [4, 5, 6, 7]. Worldvolume flux can also lift deformation zero modes, allowing non-rigid divisors to support nonperturbative superpotential contributions [8, 9].

The bulk BB-field enters the Freed–Witten quantization condition together with the canonical class of the wrapped divisor. A choice that compensates the half-integer shift on one divisor need not compensate it on another. All divisors share the same bulk background, while their worldvolume fluxes can differ. These fluxes

must also obey the orientifold relation between each divisor and its image. On a setwise invariant divisor, this requires the gauge-invariant flux FD\mathcal{F}_D to be σ\sigma-odd [5]. We therefore ask whether one bulk BB-field permits consistently quantized fluxes on every smooth connected divisor, together with its orientifold image.

The background B2,+∗=12[R]B^*_{2,+} = \frac{1}{2}[R], where RR is the total O7 divisor, has been used to permit vanishing gauge-invariant flux on D7 stacks supported on the O7 locus [10, 11]. Other common backgrounds can also work in particular geometries [12, 13]. Here we establish a general guarantee: under the assumptions stated in Sec. II, the single background B2,+∗B^*_{2,+} permits Freed–Witten period quantization on the full two-cycle lattice of every smooth connected divisor. The worldvolume fluxes are chosen individually and obey the orientifold image relation. No restriction on h−1,1(X)h^{1,1}_{-}(X) is required.

The essential step concerns smooth connected setwise invariant divisors D⊂XD \subset X, with σ(D)=D\sigma(D)=D. Such a divisor need not be supported on the O7 locus. We prove that the corresponding bulk Freed–Witten class ηD∗=12([R]+[D])∣D\eta^*_D = \frac{1}{2}([R] + [D])|_D (see Eq. (5)) has integer periods on every invariant (σ\sigma-even) integral worldvolume two-cycle class. This includes inherited and autochthonous classes. The relevant condition is invariance of the class; its representative need not be invariant or holomorphic. The proof combines the canonical-class form of Wu’s formula with a classical intersection identity for smooth involutions [14, 15, 16].

Since ηD∗\eta^*_D is σ\sigma-even, its periods on purely σ\sigma-odd classes vanish. Any non-integer periods can therefore occur only on integral classes of mixed σ\sigma-parity. We show how to repair these periods with a σ\sigma-odd worldvolume flux. The construction uses an integral class ℓD\ell_D whose σ\sigma-even pro jection is ηD∗\eta_D^\ast. Taking its σ\sigma-odd projection as FD\mathcal{F}_D makes ηD∗+FD=ℓD\eta_D^\ast+\mathcal{F}_D=\ell_D integral on the full worldvolume lattice. If h2,0(D)=0h^{2,0}(D)=0, the flux is of type (1,1)(1,1) and admits a primitive representative for an invariant Kähler metric. This includes every smooth rigid invariant divisor.

For an exchanged pair D′=σ(D)≠DD'=\sigma(D)\ne D, the construction is simpler. One chooses a quantized flux on DD and assigns FD′=−σ∗FD\mathcal{F}_{D'}=-\sigma^\ast\mathcal{F}_D. At B2,+⋆B_{2,+}^\star, the image flux then satisfies quantization automatically. Together with the invariant-divisor construction, this establishes the common-background result for any collection of smooth divisors and their images. Primitivity on exchanged divisors must, however, be checked separately.

We also study when setwise invariant divisors permit the stronger choice FD=0\mathcal{F}_D=0. This choice is particularly relevant in statistical studies of axions, where contributions from many instanton divisors are included at once [17, 18, 19, 20]. We give an exact test on arbitrary integral worldvolume classes. When h−1,1(X)=0h^{1,1}_-(X)=0, every inherited intersection class is invariant, so B2,+⋆B_{2,+}^\star already satisfies every inherited period on every smooth connected invariant divisor. This requires no assumptions on h2,0(D)h^{2,0}(D) or h−1,1(D)h^{1,1}_-(D). When bulk classes are exchanged, integer periods on their invariant sums need not imply integer periods on the individual classes. We construct common half-sum shifts of B2,+⋆B_{2,+}^\star that satisfy every inherited period when h−1,1(X)≤1h^{1,1}_-(X)\le1. For several exchanged pairs, a binary compatibility test determines which common shifts work. Retaining zero flux on the full worldvolume also requires the non-inherited periods to be integral.

Section III fixes the flux conventions and cycle notation. Section III proves the invariant-period theorem. Section IV constructs the worldvolume fluxes on invariant divisors and exchanged pairs, establishes the common-background result, and discusses the supersymmetry conditions. Sections V and VI study zero flux on invariant divisors, first on the full worldvolume lattice and then on inherited lattices. Appendix A describes the binary search over general invariant backgrounds. The remaining appendices give worked computations and explicit geometric examples.

Flux quantization and worldvolume classes

Let σ\sigma be a holomorphic involution of a smooth projective Calabi–Yau threefold XX, with σ∗ΩX=−ΩX\sigma^\ast\Omega_X=-\Omega_X. Its fixed locus consists of smooth divisor components and isolated points, corresponding to O7- and O3-planes. We write

R=⋃i=1nRRi,[R]=∑i=1nR[Ri].(1)R=\bigcup_{i=1}^{n_R}R_i,\qquad[R]=\sum_{i=1}^{n_R}[R_i]. \tag*{(1)}

where the RiR_i are the mutually disjoint O7 components, each counted once. Each RiR_i is fixed pointwise, so each individual class [Ri][R_i] is invariant.

For D7-branes or an ED3 instanton wrapping a smooth connected divisor DD, let ι:D↪X\iota:D\hookrightarrow X denote the inclusion. The pullback ι∗\iota^\ast restricts bulk forms and their cohomology classes to DD. In the usual flux normalization,

FD=FD−ι∗B2,\mathcal{F}_D=F_D-\iota^\ast B_2,

the Freed–Witten quantization condition is

FD+12c1(KD)∈H2(D,Z).(2)F_D+\frac{1}{2}c_1(K_D)\in H^2(D,\mathbb{Z}). \tag*{(2)}

Here FDF_D is the worldvolume gauge-field strength, B2B_2 is the bulk Neveu–Schwarz two-form, and FD\mathcal{F}_D is their gauge-invariant combination. The canonical line bundle KDK_D has holomorphic two-forms as its local sections, and c1(KD)c_1(K_D) is the canonical class of type (1,1)(1,1).

To express the canonical class in terms of bulk divisor classes, we use adjunction. Since KXK_X is trivial, the canonical class is just the restriction of [D][D]:

KD≃OX(D)∣D,c1(KD)=ι∗[D].(3)K_D\simeq\mathcal{O}_X(D)|_D,\qquad c_1(K_D)=\iota^\ast[D]. \tag*{(3)}

In particular, c1(KD)c_1(K_D) is integral. When DD is setwise invariant, it is also σ\sigma-even.

Let D′=σ(D)D'=\sigma(D) be the image divisor. We consider Abelian worldvolume flux classes satisfying the orientifold image relation

FD′=−σ∗FD.(4)\mathcal{F}_{D'}=-\sigma^\ast\mathcal{F}_D. \tag*{(4)}

Here σ∗\sigma^\ast transports classes from DD to D′D'. For a setwise invariant divisor, this becomes σ∗FD=−FD\sigma^\ast\mathcal{F}_D=-\mathcal{F}_D, the parity condition relevant for O(1)O(1) instantons [5]. We ask whether one common bulk BB-field permits quantized fluxes satisfying this relation on every smooth connected divisor.

We set the continuous σ\sigma-odd bulk background B2B_2, to zero and work with closed worldvolume fluxes.2 The bulk BB-field and canonical contributions to the Freed–Witten condition combine into

ηD:=ι∗B2,++12c1(KD)=ι∗(B2,++12[D]).(5)\eta_D:=\iota^\ast B_{2,+}+\frac{1}{2}c_1(K_D) =\iota^\ast\left(B_{2,+}+\frac{1}{2}[D]\right). \tag*{(5)}

We refer to ηD\eta_D as the bulk Freed–Witten class on DD, since both contributions are restrictions of bulk classes. The full quantization condition is

ηD+FD∈H2(D,Z).(6)\eta_D+\mathcal{F}_D\in H^2(D,\mathbb{Z}). \tag*{(6)}

We first consider setwise invariant divisors. For the discrete backgrounds considered here, both terms in ηD\eta_D are σ\sigma-even and of type (1,1)(1,1). However, the σ\sigma-even and σ\sigma-odd projections of an integral class need not themselves be integral. Thus ηD+FD\eta_D+\mathcal{F}_D can be integral even when its σ\sigma-even part ηD\eta_D is not. The σ\sigma-odd part of the integral sum is then FD\mathcal{F}_D. The choice FD=0\mathcal{F}_D=0 instead requires

ηD∈H+1,1(D,Z).\eta_D\in H^{1,1}_{+}(D,\mathbb{Z}).

Being of type (1,1)(1,1) and σ\sigma-even does not by itself establish integrality. Nor does the Hodge type restrict the quantization test to holomorphic curve-cycle classes. The condition must hold on every integral two-cycle, as described below.

We test quantization through periods on integral worldvolume two-cycle classes γ∈H2(D,Z)\gamma\in H_2(D,\mathbb{Z}). For a closed two-form α\alpha, we use ∫γα\int_\gamma\alpha to denote its integral over any representative of γ\gamma. The result depends only on the class, and we use the same notation when α\alpha denotes its cohomology class. The required condition is an integer period of ηD+FD\eta_D+\mathcal{F}_D on every integral worldvolume two-cycle in the covering Calabi–Yau, including classes without holomorphic curve-cycle representatives.3

The action of σ\sigma on the worldvolume classes of DD determines which periods are guaranteed. An invariant (σ\sigma-even) class satisfies σ∗γ=γ\sigma_*\gamma=\gamma. This is a condition on the class and does not require an invariant or holomorphic representative. A σ\sigma-odd class satisfies σ∗γ=−γ\sigma_*\gamma=-\gamma and has zero period of the invariant class ηD\eta_D. We say that an integral class has mixed σ\sigma-parity when it is neither σ\sigma-even nor σ\sigma-odd.

A σ\sigma-odd flux has zero period on every σ\sigma-even class. It therefore cannot repair a half-integer period of ηD\eta_D on such a class. Integer periods on all σ\sigma-even integral classes are necessary for a σ\sigma-odd flux to satisfy Eq. (6). We will show that this necessary condition is also sufficient.

The common background

The discrete choice

B2,+⋆=12[R](7)B^\star_{2,+}=\frac{1}{2}[R] \tag*{(7)}

is invariant under the geometric involution, since σ∗[R]=[R]\sigma^*[R]=[R]. The full orientifold also includes worldsheet orientation reversal, which sends B2↦−σ∗B2B_2\mapsto-\sigma^*B_2. Backgrounds differing by integral shifts are gauge equivalent, so a geometrically invariant background is allowed when 2B2,+2B_{2,+} is integral. This holds here because 2B2,+⋆=[R]2B^\star_{2,+}=[R]. In what follows, ηD⋆\eta_D^\star denotes the bulk Freed–Witten class evaluated at this background.

Theorem 1. Let D⊂XD\subset X be a smooth connected divisor satisfying σ(D)=D\sigma(D)=D. For every invariant integral worldvolume two-cycle class γ∈H2(D,Z)\gamma\in H_2(D,\mathbb{Z}),

∫γηD⋆∈Z.(8)\int_\gamma\eta_D^\star\in\mathbb{Z}. \tag*{(8)}

Thus the common background B2,+⋆=12[R]B^\star_{2,+}=\frac{1}{2}[R] gives integer Freed–Witten periods on every such class. No invariant or holomorphic representative is required. Purely σ\sigma-odd integral classes have zero period. If DD is an O7 component, the condition holds on every integral worldvolume two-cycle.

O7 components

If DD is an O7 component, write R=D+D′R=D+D'. The other fixed components do not meet DD, so ι∗[R]=0\iota^*[R]=0. Adjunction gives

ηD⋆=12ι∗[R]+12c1(KD)=ι∗[D].(9)\eta_D^\star=\frac{1}{2}\iota^*[R]+\frac{1}{2}c_1(K_D)=\iota^*[D]. \tag*{(9)}

This is an integral class and has integer periods on every integral two-cycle. It explains the familiar cancellation on D7 stacks supported on O7 components [10, 11].

Invariant divisors that are not O7 components

Suppose DD is not an O7 component. It is preserved setwise, but is not fixed pointwise, so its induced involution is nontrivial. We first relate its fixed curves to the O7 locus, and then use two intersection identities to prove the required integrality.

Fixed curves on the divisor

Let DrD_r be the union of the fixed curves on DD, each counted once. Near an O7 component, choose local coordinates with

σ(z1,z2,z3)=(−z1,z2,z3),R={z1=0}.(10)\sigma(z_1,z_2,z_3)=(-z_1,z_2,z_3),\qquad R=\{z_1=0\}. \tag*{(10)}

Setwise invariance allows a local defining equation for DD to be chosen either even or odd in z1z_1. A defining equation odd in z1z_1 would be divisible by z1z_1, and smoothness would then make DD coincide locally with the O7 component. Since DD is not an O7 component, we may therefore use a defining equation even in z1z_1. Moreover, smoothness requires a nonzero derivative in a direction tangent to RR. After relabeling z2z_2 if necessary, the local defining equation for DD can therefore be written as z2=f(z12,z3)z_2=f(z_1^2,z_3). Here (z1,z3)(z_1,z_3) are local coordinates on DD, so DD meets RR transversely and the fixed curve is cut out once by z1=0z_1=0. Consequently,

[Dr]=ι∗[R]∈H2(D,Z).(11)[D_r]=\iota^*[R]\in H^2(D,\mathbb{Z}). \tag*{(11)}

At an O3 point the induced action is locally (u,v)↦(−u,−v)(u,v) \mapsto(-u,-v). This gives an isolated fixed point, with no curve contribution.4

Two intersection identities and the proof

Write γ2\gamma^2 for the self-intersection number of γ\gamma and γ⋅σ∗γ\gamma\cdot\sigma_*\gamma for its intersection number with its image. These numbers are computed using smooth oriented surface representatives in DD, perturbed without changing their classes so that they meet transversely at isolated points. Each intersection is counted with its orientation sign. For γ2\gamma^2, we use two representatives of the same class. These perturbations need not preserve holomorphicity, even when a chosen representative is a rigid holomorphic curve.

The canonical-class form of Wu’s formula is [14]

∫γc1(KD)≡γ2(mod2).(12)\int_\gamma c_1(K_D) \equiv\gamma^2 \pmod{2}. \tag*{(12)}

For a smooth holomorphic curve CC, the same congruence follows from adjunction:

∫Cc1(KD)=2g(C)−2−C2≡C2(mod2).\int_C c_1(K_D) = 2g(C) - 2 - C^2 \equiv C^2 \pmod{2}.

Wu’s formula establishes this congruence for every integral two-cycle class, without requiring a holomorphic representative.

The second identity is [15, 16]

∫γ[Dr]≡γ⋅σ∗γ(mod2).(13)\int_\gamma[D_r] \equiv\gamma\cdot\sigma_*\gamma\pmod{2}. \tag*{(13)}

This is the four-dimensional case of Arnold’s fixed-locus intersection identity. See Proposition 3.2.3 of Ref. [16]. Intersections of a surface with its image away from the fixed locus occur in pairs. The unpaired contribution is its intersection with the fixed curves. Surface representatives can avoid the isolated fixed points.

Adding Eqs. (12) and (13), and using Eq. (11), gives

2∫γηD∗≡γ2+γ⋅σ∗γ(mod2).(14)2\int_\gamma\eta^*_{D} \equiv\gamma^2 + \gamma\cdot\sigma_*\gamma\pmod{2}. \tag*{(14)}

For an invariant class, the right-hand side is 2γ22\gamma^2, which is divisible by two. This proves Theorem 1. Invariance is used only at this last step. No invariant representative was assumed. Equation (14) also holds on O7 components. There σ∗γ=γ\sigma_*\gamma=\gamma, so its right-hand side is even, while Eq. (9) makes its left-hand side even as well.

For involutions with only O3 planes, there are no O7 components and thus no fixed curves on DD. Equation (13) then gives γ⋅σ∗γ≡0(mod2)\gamma\cdot\sigma_*\gamma\equiv0\pmod{2}. An invariant class therefore has self-intersection divisible by two. Equation (12) then makes its canonical period divisible by two as well. The theorem holds with B2,⋆+=0B^+_{2,\star}=0.

An immediate consequence gives a sufficient condition for zero flux on the full worldvolume lattice.

Corollary 1 (zero odd second cohomology). If a smooth connected invariant divisor DD satisfies

h−1,1(D)=h−2,0(D)=0,(15)h^{1,1}_-(D)=h^{2,0}_-(D)=0, \tag*{(15)}

then B2,⋆+B^+_{2,\star} satisfies every integral worldvolume period condition with FD=0\mathcal{F}_D=0.

Indeed, h−2,0(D)=0h^{2,0}_-(D)=0 as well, so the σ\sigma-odd part of H2(D,C)H^2(D,\mathbb{C}) vanishes. Every class in the free integral two-cycle lattice is therefore invariant, and Theorem 1 applies to the entire lattice. No vanishing of h+2,0(D)h^{2,0}_+(D) is needed. For a rigid divisor, the condition reduces to h−1,1(D)=0h^{1,1}_-(D)=0.

Worldvolume fluxes and full quantization

For a setwise invariant divisor, we first show that integer periods on invariant classes are necessary and sufficient for completing quantization with a σ\sigma-odd flux. Theorem 1 guarantees these periods at B2,⋆+B^+_{2,\star}. We then treat exchanged pairs and combine the two constructions.

Constructing the worldvolume fluxes

Proposition 1 (odd-flux completion). Let DD be setwise invariant and let η∈H2(D,Q)\eta\in H^2(D,\mathbb{Q}) be invariant. There exists a σ\sigma-odd rational class F\mathcal{F} for which η+F\eta+\mathcal{F} has integer periods on every integral two-cycle if and only if η\eta has integer periods on every invariant integral two-cycle class.

Necessity follows because a σ\sigma-odd class has zero period on an invariant cycle. To prove sufficiency, write

L=H2(D,Z)/Tor⁡H2(D,Z),L=H_2(D,\mathbb{Z})/\operatorname{Tor} H_2(D,\mathbb{Z}),
L+={γ∈L:σ∗γ=γ}.(16)L_+=\{\gamma\in L:\sigma_*\gamma=\gamma\}. \tag*{(16)}

The quotient removes torsion, so LL has a finite integral basis. If a nonzero integer multiple nγn\gamma lies in L+L_+, then n(σ∗γ−γ)=0n(\sigma_*\gamma-\gamma)=0. Since LL is torsion-free, γ\gamma itself lies in L+L_+. This property ensures that an integral basis of L+L_+ can be extended to a full integral basis of LL. Choose such a basis,

γ1,…,γk,μ1,…,μs,(17)\gamma_1,\ldots,\gamma_k,\mu_1,\ldots,\mu_s, \tag*{(17)}

where the γa\gamma_a form a basis of L+L_+. The added classes μb\mu_b need not have definite σ\sigma-parity. An integral cohomology class can be specified by assigning integer periods on a full integral two-cycle basis. Choose ℓ\ell to agree with η\eta on the invariant basis cycles and to have zero periods on the added cycles:

∫γaℓ=∫γaη∈Z,∫μbℓ=0.(18)\int_{\gamma_a}\ell=\int_{\gamma_a}\eta\in\mathbb{Z},\qquad\int_{\mu_b}\ell=0. \tag*{(18)}

For any γ∈L\gamma\in L, the sum γ+σ∗γ\gamma+\sigma^*\gamma is invariant. Consequently,

∫γ(ℓ+σ∗ℓ)=∫γ+σ∗γℓ=∫γ+σ∗γη=2∫γη.(19)\begin{aligned} \int_\gamma(\ell+\sigma^*\ell)=\int_{\gamma+\sigma^*\gamma}\ell\\ &= \int_{\gamma+\sigma^*\gamma}\eta=2\int_\gamma\eta. \tag*{(19)} \end{aligned}

Thus 12(ℓ+σ∗ℓ)=η\frac{1}{2}(\ell+\sigma^*\ell)=\eta in rational cohomology: η\eta is the σ\sigma-even projection of ℓ\ell. Define the flux to be its σ\sigma-odd projection:

F=12(ℓ−σ∗ℓ),η+F=ℓ.(20)\mathcal{F}=\frac{1}{2}(\ell-\sigma^*\ell),\qquad\eta+\mathcal{F}=\ell. \tag*{(20)}

The flux is σ\sigma-odd and the sum is integral, proving the proposition. The zero periods assigned to the μb\mu_b were arbitrary; any integer choices give a solution.

Theorem 2 (a common background for all smooth divisors). Under the assumptions of Sec. II, consider any collection of smooth connected divisors together with their orientifold images. The single bulk background B2,+⋆=12[R]B_{2,+}^\star=\frac{1}{2}[R] admits worldvolume flux classes satisfying

Fσ(D)=−σ∗FD,ηD⋆+FD∈H2(D,Z)free(21)\mathcal{F}_{\sigma(D)}=-\sigma^*\mathcal{F}_D,\qquad\eta_D^\star+\mathcal{F}_D\in H^2(D,\mathbb{Z})_{\mathrm{free}} \tag*{(21)}

on every divisor in the collection. The fluxes can be chosen with 2FD2\mathcal{F}_D integral. Quantization holds on the full worldvolume two-cycle lattices, including inherited and autochthonous classes, with no restriction on h−1,1(X)h^{1,1}_-(X).

If DD is setwise invariant, Theorem 1 supplies the integer invariant-cycle periods required by Proposition 1. The proposition produces an integral class ℓD\ell_D whose σ\sigma-even projection is ηD⋆\eta_D^\star. Its σ\sigma-odd projection FD=ℓD−ηD⋆\mathcal{F}_D=\ell_D-\eta_D^\star satisfies both conditions.

Now suppose D′=σ(D)≠DD'=\sigma(D)\ne D. Choose any integral class ℓD\ell_D on one member of the pair and set

FD=ℓD−ηD⋆,FD′=−σ∗FD.(22)\mathcal{F}_D=\ell_D-\eta_D^\star,\qquad\mathcal{F}_{D'}=-\sigma^*\mathcal{F}_D. \tag*{(22)}

On DD, quantization holds because ηD⋆+FD=ℓD\eta_D^\star+\mathcal{F}_D=\ell_D is integral by construction. It remains to check the image divisor. Since ηD′⋆=σ∗ηD⋆\eta_{D'}^\star=\sigma^*\eta_D^\star, its image satisfies

ηD′⋆+FD′=σ∗(ηD⋆−FD)=σ∗(2ηD⋆−ℓD).(23)\begin{aligned} \eta_{D'}^\star+\mathcal{F}_{D'}=\sigma^*(\eta_D^\star-\mathcal{F}_D) \\ &=\sigma^*(2\eta_D^\star-\ell_D). \tag*{(23)} \end{aligned}

This is integral because

2ηD⋆=ι∗([R]+[D])2\eta_D^\star=\iota^*([R]+[D])

is integral.

The exchanged-pair construction works for any discrete σ\sigma-even background with integral 2B2,+2B_{2,+}. Theorem 1 ensures that B2,+⋆B_{2,+}^\star also permits the required odd fluxes on invariant divisors. The two constructions therefore provide one common background for the entire collection. In both cases, 2FD2\mathcal{F}_D is integral. On O7 components and on the invariant divisors of Corollary 1, FD=0\mathcal{F}_D=0 already satisfies quantization.

On a setwise invariant divisor, starting from any one solution FD(0)\mathcal{F}_D^{(0)}, we obtain the full family of quantized σ\sigma-odd fluxes by adding an arbitrary integral σ\sigma-odd class:

FD=FD(0)+ξD,ξD∈H−2(D,Z)free.(24)\mathcal{F}_D=\mathcal{F}_D^{(0)}+\xi_D,\qquad\xi_D\in H^2_-(D,\mathbb{Z})_{\mathrm{free}}. \tag*{(24)}

The construction does not require these fluxes to be inherited from the bulk. Implementing it can require the full integral worldvolume lattice and its involution action, beyond the bulk toric intersection data. Appendix D gives the corresponding integer system and a simple example.

When the flux satisfies the supersymmetry conditions

We first consider the σ\sigma-odd flux on a setwise invariant divisor. For a supersymmetric instanton, this flux must also be of type (1,1)(1,1) and primitive [5]. A sufficient condition is the absence of σ\sigma-odd holomorphic two-forms.

Corollary 2 (odd (1,1)(1,1) flux). On a smooth connected setwise invariant divisor DD with h−2,0(D)=0h^{2,0}_-(D)=0, the flux in Theorem 2 is of type (1,1)(1,1) and admits a primitive representative for an invariant Kähler form. In particular, this holds on every smooth rigid invariant divisor, without requiring h−1,1(D)=0h^{1,1}_-(D)=0.

The flux class FD\mathcal{F}_D is real and σ\sigma-odd. For a σ\sigma-even Kähler form JJ, parity gives

∫Dι∗J∧FD=0.(25)\int_D \iota^*J\wedge\mathcal{F}_D=0. \tag*{(25)}

On a connected Kähler surface, the component of a harmonic two-form proportional to the Kähler form has a constant coefficient. Equation (25) forces this coefficient to vanish, so the harmonic representative of FD\mathcal{F}_D is primitive. If, in addition, h−2,0(D)=0h^{2,0}_-(D)=0, its (2,0)(2,0) and (0,2)(0,2) components vanish, since their harmonic representative is also of type (1,1)(1,1). When h−2,0(D)≠0h^{2,0}_-(D)\ne0, the (1,1)(1,1) condition must be tested separately.

For an exchanged pair, choosing ℓD=0\ell_D=0 in Eq. (22) gives FD=−ηD⋆\mathcal{F}_D=-\eta_D^\star and FD′=ηD′⋆\mathcal{F}_{D'}=\eta_{D'}^\star. These classes are of type (1,1)(1,1). The image relation does not, however, force ∫Dι∗J∧FD=0\int_D\iota^*J\wedge\mathcal{F}_D=0 on either divisor individually. Primitivity must therefore be checked separately.

Zero flux on invariant divisors

We now return to setwise invariant divisors and ask when ηD∗\eta_D^* itself has integer periods on every integral two-cycle, allowing FD=0\mathcal{F}_D=0. Corollary 1 covers divisors with no σ\sigma-odd second cohomology. On any other invariant divisor, Theorem 1 still guarantees the σ\sigma-even periods, including invariant autochthonous classes, and the σ\sigma-odd periods vanish. The remaining tests are on integral classes of mixed σ\sigma-parity.

For an arbitrary integral worldvolume class γ\gamma, define δγ=γ−σ∗γ\delta_\gamma=\gamma-\sigma_*\gamma. Preservation of the intersection pairing gives

δγ2=2(γ2−γ⋅σ∗γ).(26)\delta_\gamma^2=2(\gamma^2-\gamma\cdot\sigma_*\gamma). \tag*{(26)}

Equation (14) therefore becomes

2∫γηD∗=12δγ2(mod2).(27)2\int_\gamma\eta_D^*=\frac{1}{2}\delta_\gamma^2\pmod{2}. \tag*{(27)}

Therefore, the exact period test is

∫γηD∗∈Z⟺δγ2∈4Z.(28)\int_\gamma\eta_D^*\in\mathbb{Z}\quad\Longleftrightarrow\quad\delta_\gamma^2\in4\mathbb{Z}. \tag*{(28)}

An integral class of mixed σ\sigma-parity need not violate quantization: Eq. (28) determines whether its period is integral.

If curves CC and σ(C)\sigma(C) are exchanged, invariance of ηD∗\eta_D^* gives

∫CηD∗=∫σ(C)ηD∗=12∫C+σ(C)ηD∗.(29)\int_C\eta_D^*=\int_{\sigma(C)}\eta_D^*=\frac{1}{2}\int_{C+\sigma(C)}\eta_D^*. \tag*{(29)}

An integer period on the sum can coexist with half-integer periods on the components. Both component periods must satisfy quantization.

Appendix F gives an explicit example in which every inherited period is integral, while some autochthonous periods are half-integer.

Zero flux on inherited lattices

We continue to impose FD=0\mathcal{F}_D=0 and first test the intersection classes inherited from the bulk. Theorem 1 guarantees integer periods on invariant inherited classes. We identify the individual tests for exchanged classes and show when and how a common bulk shift can make every inherited period integral on all setwise invariant target divisors. Zero flux on the full worldvolume additionally requires the autochthonous periods to be integral.

The inherited corollary

For an arbitrary integral bulk divisor class [E]∈H2(X,Z)[E]\in H^2(X,\mathbb{Z}), invariant or otherwise, let γE\gamma_E denote its inherited intersection class on DD. It is characterized, for closed two-forms α\alpha, by

∫γEα=∫Dι∗[E]∧α.(30)\int_{\gamma_E}\alpha=\int_D\iota^*[E]\wedge\alpha. \tag*{(30)}

If the geometric intersection D∩ED\cap E consists of curves CjC_j with multiplicities njn_j, then γE=∑jnj[Cj]\gamma_E=\sum_j n_j[C_j] is their homology class. Integer combinations of the classes γE\gamma_E, as [E][E] ranges over the integral bulk divisor lattice, form the inherited intersection curve lattice of DD. An individual component CjC_j need not be inherited. Classes outside this lattice are non-inherited, or autochthonous.

This definition also includes E=DE=D: by adjunction, γD\gamma_D corresponds to ι∗[D]=c1(KD)\iota^*[D]=c_1(K_D). It is defined by restriction even if [D][D] has no other holomorphic representative with which to intersect DD.

At B2,+∗B^*_{2,+}, define the inherited period

I(D,E):=∫γEηD∗=12∫X[D][E]([D]+[R]).(31)I(D,E):=\int_{\gamma_E}\eta_D^* =\frac{1}{2}\int_X[D][E]\bigl([D]+[R]\bigr). \tag*{(31)}

This period may be half-integer, whereas Freed–Witten quantization requires it to be an integer.

Corollary 3. If the inherited class γE\gamma_E is invariant, then

I(D,E)∈Z.(32)I(D,E)\in\mathbb{Z}. \tag*{(32)}

In particular, this holds for every invariant integral bulk class [E][E], including E=DE=D. When h−1,1(X)=0h^{1,1}_-(X)=0, it holds on the full inherited intersection curve lattice of $D.

Indeed, restriction to invariant DD commutes with the involution. An invariant bulk class therefore restricts to an invariant worldvolume class. A non-invariant bulk class can also restrict to an invariant class if its σ\sigma-odd projection restricts to zero. No condition on h2,0(D)h^{2,0}(D) or h−1,1(D)h^{1,1}_-(D) is needed for this corollary. A purely σ\sigma-odd bulk class has zero period trivially.

A bulk shift for an exchanged pair

The target divisor DD remains setwise invariant. Let [e]↔[e′][e]\leftrightarrow[e'] be an exchanged pair of integral bulk divisor classes. Their inherited classes γe\gamma_e and γe′=σ∗γe\gamma_{e'}=\sigma^*\gamma_e have equal periods of ηD∗\eta_D^*, but these periods may be half-integer.

The two intersection numbers entering Eq. (14) are

γe2=∫X[D][e]2,γe⋅σγe=∫X[D][e][e′].\gamma_e^2=\int_X[D][e]^2,\qquad \gamma_e\cdot\sigma\gamma_e=\int_X[D][e][e'].

That equation therefore reads

2∫γeηD∗=2I(D,e)≡∫X[D][e]([e]+[e′])(mod2).(33)2\int_{\gamma_e}\eta_D^*=2I(D,e)\equiv\int_X[D][e]\bigl([e]+[e']\bigr)\pmod{2}. \tag*{(33)}

In other words, the periods of ηD⋆\eta_D^\star and 12ι⋆([e]+[e′])\frac{1}{2}\iota^\star([e]+[e']) on γe\gamma_e differ by an integer. Subtracting this half-sum from the bulk background therefore makes the inherited period integral:

B~2,+⋆=12[R]−12([e]+[e′]).(34)\widetilde{B}_{2,+}^\star=\frac{1}{2}[R]-\frac{1}{2}([e]+[e']). \tag*{(34)}

Since this background is σ\sigma-even, the image class σ∗γe\sigma_\ast\gamma_e has the same period and passes the same test.

The periods on invariant integral worldvolume classes remain integers, since

12∫γι⋆([e]+[e′])=∫γι⋆[e]∈Z(σ∗γ=γ).(35)\frac{1}{2}\int_\gamma\iota^\star([e]+[e'])=\int_\gamma\iota^\star[e]\in\mathbb{Z}\qquad(\sigma_\ast\gamma=\gamma). \tag*{(35)}

On an O7 component every worldvolume class is invariant, so integrality is preserved on its full lattice. Purely σ\sigma-odd periods remain zero. The background B~2,+⋆\widetilde{B}_{2,+}^\star also obeys the orientifold condition: it is σ\sigma-even and twice its class is integral.

Theorem 3. Suppose σ\sigma exchanges one pair [e]↔[e′][e]\leftrightarrow[e'] in an integral basis of divisor classes on XX and leaves all other basis classes invariant. Then the common background (34) satisfies Freed–Witten period quantization on the full inherited intersection curve lattice of every smooth connected setwise invariant divisor DD.

The inherited periods associated with [e][e] and [e′][e'] are integral by the preceding calculation. All remaining basis classes are invariant, and their periods remain integral by Eq. (35). Since every inherited class is an integer combination of these basis intersections, this proves the theorem.

If several pairs of basis classes are exchanged, shifting the background for one pair can change the periods of another. We must therefore choose a common combination of shifts that makes every inherited period integral on all target divisors.

Choosing a common combination

Choose an integral divisor basis in which σ\sigma exchanges mm pairs and every remaining class is σ\sigma-even or σ\sigma-odd.5 Let [Aα][A_\alpha], α=1,…,m\alpha=1,\ldots,m, denote one class from each exchanged pair. Fixed basis classes contribute only integral shifts, while σ\sigma-odd basis classes contribute no shift. Only the half-sums of exchanged pairs therefore need to be tested.

For each pair, we may include or omit the shift 12([Aα]+σ⋆[Aα])\frac{1}{2}([A_\alpha]+\sigma^\star[A_\alpha]). There are therefore 2m2^m possible combinations. Write

[A]=∑α=1mzα[Aα],zα∈Z2,(36)[A]=\sum_{\alpha=1}^{m}z^\alpha[A_\alpha],\qquad z^\alpha\in\mathbb{Z}_2, \tag*{(36)}
B~2,+⋆=12[R]−12([A]+σ⋆[A]).(37)\widetilde{B}_{2,+}^\star=\frac{1}{2}[R]-\frac{1}{2}([A]+\sigma^\star[A]). \tag*{(37)}

The corresponding bulk Freed–Witten class is

η~D⋆=ηD⋆−12ι⋆([A]+σ⋆[A]).(38)\widetilde{\eta}_D^\star=\eta_D^\star-\frac{1}{2}\iota^\star([A]+\sigma^\star[A]). \tag*{(38)}

For every candidate background, the invariant periods remain integral by Eq. (35). It suffices to test one class from each exchanged pair: its image has the same period because the background B~2,+⋆\widetilde{B}_{2,+}^\star is σ\sigma-even. All remaining basis classes have definite σ\sigma-parity, so their inherited periods are already integral or zero.

For m=0m=0, take [A]=0[A]=0. For m=1m=1, take [A]=[A1][A]=[A_1], using the one-pair argument of Theorem 3. Each exchanged pair contributes one σ\sigma-odd direction, so m≤h−1,1(X)m\le h^{1,1}_{-}(X). Thus one common shift always works when h−1,1(X)≤1h^{1,1}_{-}(X)\le1, and also at higher odd rank when at most one pair is exchanged in this basis.

For several independent shifts, their effects on all selected periods must be included. For each target divisor DiD_i, define

Mαβ(i):=∫X[Di][Aα]([Aβ]+σ⋆[Aβ]) mod 2,dα(i):=2I(Di,Aα) mod 2=Mαα(i).(39)M_{\alpha\beta}^{(i)}:=\int_X[D_i][A_\alpha]([A_\beta]+\sigma^\star[A_\beta])\bmod2, \qquad d_\alpha^{(i)}:=2\mathcal{I}(D_i,A_\alpha)\bmod2=M_{\alpha\alpha}^{(i)}. \tag*{(39)}

These matrices are symmetric: the term containing [Aα][Aβ][A_\alpha][A_\beta] is manifestly symmetric, and applying σ\sigma to the other term exchanges α\alpha and β\beta while leaving [Di][D_i] invariant. The equality in the second line of Eq. (39) follows from Eq. (33). An entry dα(i)=1d_\alpha^{(i)}=1 marks a half-integer period at B~2,+⋆\widetilde{B}_{2,+}^\star. An entry Mαβ(i)=1M_{\alpha\beta}^{(i)}=1 means that including shift β\beta changes period α\alpha by a half-integer. All inherited periods on DiD_i, with this background, are therefore integral precisely when

M(i)z=d(i)over Z2.(40)M^{(i)}z=d^{(i)}\quad\text{over }\mathbb{Z}_2. \tag*{(40)}

One common background requires the same zz for every target. Stack their equations into the single binary system

(M(1)M(2)⋮)⏟Mz=(d(1)d(2)⋮)⏟d.(41)\underbrace{\begin{pmatrix} M^{(1)}\\ M^{(2)}\\ \vdots \end{pmatrix}}_{\mathcal{M}}z = \underbrace{\begin{pmatrix} d^{(1)}\\ d^{(2)}\\ \vdots \end{pmatrix}}_{\mathbf{d}}. \tag*{(41)}

Gaussian elimination using XOR row additions finds every compatible combination, if any. Appendix C illustrates why including all half-sums need not work, even when a smaller combination does. For m>1m>1, consistency is not guaranteed by our argument. An inconsistent system excludes this family of shifts for the chosen targets.

For a sufficient test on all smooth invariant divisors, one may instead form MM and its diagonal on an integral basis of the invariant bulk divisor lattice. Both depend linearly on the target class [D][D].

Appendix A extends the search to other invariant discrete backgrounds, including choices outside the half-sum family.

Full worldvolume tests for a shifted background

After the inherited tests, the remaining autochthonous periods must also be checked. For any such integral class γ\gamma, Eqs. (38) and (27) show that ∫γη~D⋆\int_\gamma\widetilde{\eta}^{\star}_{D} is an integer precisely when

12δγ2−∫γ+σ∗γι∗[A]≡0(mod2).(42)\frac{1}{2}\delta_{\gamma}^{2}-\int_{\gamma+\sigma_{*}\gamma}\iota^{*}[A]\equiv0\pmod{2}. \tag*{(42)}

The same calculation also identifies an inherited odd flux at the original background:

ηD⋆+12ι∗([A]−σ∗[A])=η~D⋆+ι∗[A].(43)\eta_{D}^{\star}+\frac{1}{2}\iota^{*}\left([A]-\sigma^{*}[A]\right)=\widetilde{\eta}_{D}^{\star}+\iota^{*}[A]. \tag*{(43)}

Since ι∗[A]\iota^{*}[A] is integral, the flux FD=12ι∗([A]−σ∗[A])F_{D}=\frac{1}{2}\iota^{*}\left([A]-\sigma^{*}[A]\right) at B2,+⋆B_{2,+}^{\star} satisfies the same period tests as zero flux at B~2,+⋆\widetilde{B}_{2,+}^{\star}. The general odd-flux construction of Theorem 2 also allows fluxes that are not inherited from the bulk.

Conclusions

We have shown that the single bulk background B2,+⋆=12[R]B_{2,+}^{\star}=\frac{1}{2}[R] permits Freed–Witten period quantization on any collection of smooth connected divisors and their orientifold images, under the assumptions of Sec. II. The worldvolume fluxes are chosen individually, obey the orientifold image relation, and satisfy quantization on the full two-cycle lattices, including inherited and autochthonous classes. No restriction on h−1,1(X)h^{1,1}_{-}(X) is required.

The essential ingredient is the invariant-period theorem for setwise invariant divisors. On every such divisor, ηD⋆\eta_{D}^{\star} has integer periods on every invariant integral two-cycle class, without requiring an invariant or holomorphic representative. Purely σ\sigma-odd classes have zero period, so any non-integer periods occur only on classes of mixed σ\sigma-parity. This invariant-period statement is precisely the condition needed to complete quantization with an odd flux. For exchanged pairs, a quantized choice on one member also quantizes its orientifold image. The worldvolume fluxes need not be inherited from the bulk. On O7 components, zero worldvolume flux already satisfies quantization on every integral two-cycle.

On a setwise invariant divisor, the harmonic representative of the odd flux is primitive for an invariant Kähler metric. When h−2,0(D)=0h^{2,0}_{-}(D)=0, it is also of type (1,1)(1,1) and therefore satisfies the flux conditions for a supersymmetric instanton. This includes every smooth rigid invariant divisor, without requiring h−2,0(D)=0h^{2,0}_{-}(D)=0. If h−2,0(D)≠0h^{2,0}_{-}(D)\ne0, the (1,1)(1,1) condition must be tested separately. For exchanged divisors, the image relation alone does not ensure primitivity on either member. The Chan–Paton projection and the neutral and charged zero modes remain additional conditions for a contributing instanton.

We have also studied when invariant divisors permit FD=0F_{D}=0. This requires the bulk Freed–Witten class itself to have integer periods on the full worldvolume lattice.

At B2,+⋆B_{2,+}^{\star}, only classes of mixed σ\sigma-parity require further tests, and the criterion δγ2∈4Z\delta_{\gamma}^{2}\in4\mathbb{Z} determines whether each period is integral. The full condition is automatic when the σ\sigma-odd second cohomology of DD vanishes.

On inherited lattices of invariant divisors, B2,+⋆B_{2,+}^{\star} already works whenever h−1,1(X)=0h^{1,1}_{-}(X)=0. Common half-sum shifts extend this inherited guarantee to h−1,1(X)≤1h^{1,1}_{-}(X)\le1, and also to higher odd rank when at most one pair is exchanged in an integral bulk basis. These inherited guarantees require no conditions on h−2,0(D)h^{2,0}_{-}(D) or h−1,1(D)h^{1,1}_{-}(D). For several exchanged pairs, the binary equations account for the effects of each shift on the other periods and determine which common combinations work. The remaining autochthonous periods must also be integral to retain zero flux on the full worldvolume. The double-octic example makes this distinction explicit: every inherited period is integral at B2,+⋆B_{2,+}^{\star}, while some autochthonous periods are half-integer. The alternative zero-flux background in that example lies outside the half-sum family.

The common background B2,+⋆B_{2,+}^{\star} can therefore be used for instanton quantization while exploring D7 configurations beyond componentwise tadpole cancellation. The exchanged-pair construction also ensures period quantization for smooth D7 divisors that are not individually invariant. This freedom matters for chiral model building, since the componentwise prescription can obstruct a net chiral Standard Model quark-doublet spectrum [23]. Numerical tests and applications to axion spectra generated by rigid ED3 instantons will be presented separately [24].

Acknowledgments

This work is supported by the U.S. Department of Energy (DOE) under Grant No. DE-SC0010103. I am thankful to Federico Carta for useful discussions, and to Andreas Schachner for helpful comments on the manuscript.

Computing other possible zero-flux backgrounds

For setwise invariant target divisors, we now search over all invariant discrete backgrounds with FD=0F_{D}=0, allowing choices outside the family of shifts (37).6 We first impose the inherited period conditions on them and then add any further worldvolume tests. The calculation applies to any smooth involution and can be used for orientifolds of Kreuzer–Skarke hypersurfaces [25, 26], including systematic collections of divisor exchanges [27, 28].

The full inherited system

Set B2,−=0B_{2,-}=0. Let ωa\omega_a, a=1,…,n+a=1,\ldots,n_+, be an integral basis of the invariant bulk divisor lattice, where n+=h+1,1(X)n_+=h^{1,1}_+(X). Modulo integral shifts, write

B2,+(x)=12xaωa,xa∈Z2,(44)B_{2,+}(x)=\frac{1}{2}x^a\omega_a,\qquad x^a\in\mathbb{Z}_2, \tag*{(44)}

and [Di]=Ciaωa[D_i]=C_i^a\omega_a with integer CiaC_i^a. Then

ηDi(x)=12(xa+Cia)ℓi∗ωa.(45)\eta_{D_i}(x)=\frac{1}{2}\left(x^a+C_i^a\right)\ell_i^*\omega_a. \tag*{(45)}

Use a full integral bulk basis [eA][e_A], without requiring its elements to have definite σ\sigma-parity. Every inherited period is tested by

Ax=qover Z2,(46)Ax=q\quad\text{over }\mathbb{Z}_2, \tag*{(46)}

with rows

AAa(i)=∫X[Di][eA]ωa mod⁡2,A_{Aa}^{(i)}=\int_X[D_i][e_A]\omega_a\ \operatorname{mod}2,
qA(i)=∫X[Di]2[eA] mod⁡2.(47)q_A^{(i)}=\int_X[D_i]^2[e_A]\ \operatorname{mod}2. \tag*{(47)}

When m=0m=0, the known solution is xR=[R]mod⁡2x_R=[R]\operatorname{mod}2 in the invariant basis. When m=1m=1, the one-pair argument supplies the known solution from [R]−[A1]−σ∗[A1][R]-[A_1]-\sigma^*[A_1]. More generally, Eq. (41) can provide a common solution for the selected collection. The unrestricted system (A3) may also admit solutions outside that family of shifts.

If x0x_0 is one solution, all other solutions are

x=x0+v,Av=0over Z2.(48)x=x_0+v,\qquad Av=0\quad\text{over }\mathbb{Z}_2. \tag*{(48)}

Gaussian elimination both tests consistency and finds these solutions. Row addition is exclusive-OR (XOR): 1+1=01+1=0. If the system is consistent and AA has binary rank ss, there are 2n+−s2^{n_+-s} backgrounds satisfying all tested periods. Appendix B gives a worked example.

For h−1,1(X)=0h^{1,1}_-(X)=0, choosing [eA]=ωA[e_A]=\omega_A recovers

Aba(i)=Cikκkba mod⁡2,A_{ba}^{(i)}=C_i^k\kappa_{kba}\ \operatorname{mod}2,
qb(i)=CikCidκkcd mod⁡2,κabc=∫Xωaωbωc.(49)q_b^{(i)}=C_i^kC_i^d\kappa_{kcd}\ \operatorname{mod}2,\qquad\kappa_{abc}=\int_X\omega_a\omega_b\omega_c. \tag*{(49)}

When h−1,1(X)>0h^{1,1}_-(X)>0, the full basis in Eq. (47) is essential.

Further periods and physical conditions

For any integral two-cycle class γ\gamma on DiD_i, define its integer bulk charges

na(γ)=∫γℓi∗ωa.(50)n_a(\gamma)=\int_\gamma\ell_i^*\omega_a. \tag*{(50)}

Its period is

∫γηDi(x)=12(xa+Cia)na(γ).(51)\int_\gamma\eta_{D_i}(x)=\frac{1}{2}\left(x^a+C_i^a\right)n_a(\gamma). \tag*{(51)}

Thus integrality adds the row

na(γ)xa=Ciana(γ)(mod2).(52)n_a(\gamma)x^a=C_i^a n_a(\gamma)\pmod2. \tag*{(52)}

An integral basis of worldvolume classes tests every period. For exchanged curves, invariance of ωa\omega_a gives na(σ(C))=na(C)n_a(\sigma(C))=n_a(C): either component supplies the same row, while their sum doubles every charge and gives 0=00=0 modulo two. A class trivial in XX has all na=0n_a=0 and imposes no additional condition. An autochthonous class need not be trivial in XX.

Starting from a known solution, shifts must preserve the rows already satisfied. Substituting x=x0+vx=x_0+v therefore requires Av=0Av=0 together with the additional equations. If AA has rank n+n_+, there is no nontrivial shift within Eq. (44). The enlarged system may have fewer solutions, or none.

An inconsistent enlarged system excludes a common zero-flux background among the candidates being tested. It does not exclude the odd-flux solutions at B2,+∗B^*_{2,+}, guaranteed by Theorem 2. Their supersymmetry condition is discussed in Sec. IV.

An illustrative h+1,1=4h^{1,1}_+=4 XOR computation

We illustrate the calculation with four invariant bulk classes, h−1,1=h+1,1=4h^{1,1}_-=h^{1,1}_+=4. The following intersection data modulo two are illustrative input, rather than a claim about a specified Calabi–Yau geometry.

For one smooth connected invariant divisor, take

[D]=ω1,[R]=ω2+ω3+ω4.(53)[D]=\omega_1,\qquad[R]=\omega_2+\omega_3+\omega_4. \tag*{(53)}

The configuration B2,+∗B^*_{2,+} corresponds to xR=(0,1,1,1)Tx_R=(0,1,1,1)^T. Suppose

Aba=∫X[D]ωbωa mod⁡2=(1111110010101001).(54)A_{ba}=\int_X[D]\omega_b\omega_a\ \operatorname{mod}2= \begin{pmatrix} 1&1&1&1\\ 1&1&0&0\\ 1&0&1&0\\ 1&0&0&1 \end{pmatrix}. \tag*{(54)}

Since [D]=ω1[D]=\omega_1, the right-hand side is the first column:

qb=Ab1,q=(1,1,1,1)T.(55)q_b=A_{b1},\qquad q=(1,1,1,1)^T. \tag*{(55)}

Direct substitution verifies AxR=qAx_R=q. Gaussian elimination over Z2\mathbb{Z}_2 gives

(11111110011010110011)⟶(10011010100011000000).(56)\left( \begin{array}{cccc|c} 1&1&1&1&1\\ 1&1&0&0&1\\ 1&0&1&0&1\\ 1&0&0&1&1 \end{array} \right) \longrightarrow \left( \begin{array}{cccc|c} 1&0&0&1&1\\ 0&1&0&1&0\\ 0&0&1&1&0\\ 0&0&0&0&0 \end{array} \right). \tag*{(56)}

Every row addition is an XOR. For example, adding the original second and third equations cancels x1x^{1} and gives x2+x3=0x^{2}+x^{3}=0, since 1+1=01+1=0.

There are three independent equations and one free variable. Setting x4=t∈Z2x^{4}=t\in\mathbb{Z}_{2}, the solutions are

x=(1+t,t,t,t)T.(57)x=(1+t,t,t,t)^{T}. \tag*{(57)}

Thus there are 24−3=22^{4-3}=2 backgrounds:

t=0:B2,+(0)=12ω1,t=0:\quad B_{2,+}^{(0)}=\frac{1}{2}\omega_{1},

t=1:B2,+(1)=12(ω2+ω3+ω4)=B2,+⋆.t=1:\quad B_{2,+}^{(1)}=\frac{1}{2}(\omega_{2}+\omega_{3}+\omega_{4})=B_{2,+}^{\star}. {#eq:B6}

Changing tt adds v=(1,1,1,1)Tv=(1,1,1,1)^{T}, with Av=0Av=0. Both choices satisfy the four inherited tests.

To illustrate an additional test, suppose an integral class γ\gamma has charges na(γ)=(0,0,0,1)n_{a}(\gamma)=(0,0,0,1). Its period is x4/2x^{4}/2, so integrality requires x4=0x^{4}=0 and leaves only B2,+(0)B_{2,+}^{(0)}. Such a class would have mixed σ\sigma-parity: its half-integer period at B2,+⋆B_{2,+}^{\star} excludes invariance, and its nonzero period excludes σ\sigma-odd parity.

Choosing a common shift: a two-pair example

We illustrate how two exchanged-pair shifts can affect one another. The binary matrices below are illustrative data. Appendix E gives a compact Calabi–Yau example of the one-pair construction.

Take two exchanged pairs in a full integral basis, [e1]↔[e1′][e_{1}]\leftrightarrow[e'_{1}] and [e2]↔[e2′][e_{2}]\leftrightarrow[e'_{2}], with all remaining basis classes invariant. Choose [A1]=[e1][A_{1}]=[e_{1}] and [A2]=[e2][A_{2}]=[e_{2}]. The four candidate backgrounds are

B~2,+⋆=12[R]−z12([e1]+[e1′])−z22([e2]+[e2′]),z1,z2∈Z2.(58)\widetilde{B}_{2,+}^{\star}=\frac{1}{2}[R]-\frac{z^{1}}{2}([e_{1}]+[e'_{1}])-\frac{z^{2}}{2}([e_{2}]+[e'_{2}]),\qquad z^{1},z^{2}\in\mathbb{Z}_{2}. \tag*{(58)}

Suppose the pairings on two target divisors reduce to

M(1)=(1000),M(2)=(1111).(59)M^{(1)}=\begin{pmatrix}1&0\\0&0\end{pmatrix},\qquad M^{(2)}=\begin{pmatrix}1&1\\1&1\end{pmatrix}. \tag*{(59)}

Their diagonal vectors are d(1)=(1,0)Td^{(1)}=(1,0)^{T} and d(2)=(1,1)Td^{(2)}=(1,1)^{T}. The first target requires z1=1z^{1}=1. The second requires z1+z2=1z^{1}+z^{2}=1. Stacking and reducing gives

\begin{pmatrix} 1&0&1\\ 0&0&0\\ 1&1&1\\ 1&1&1 \end{pmatrix} \longrightarrow \begin{pmatrix} 1&0&1\\ 0&1&0\\ 0&0&0\\ 0&0&0 \end{pmatrix}. \tag*{(60)}\end{pmatrix}

The unique common solution is z=(1,0)Tz=(1,0)^{T}, so [A]=[A1]=[e1][A]=[A_{1}]=[e_{1}]. It selects

B~2,+⋆=12[R]−12([e1]+[e1′]).(61)\widetilde{B}_{2,+}^{\star}=\frac{1}{2}[R]-\frac{1}{2}([e_{1}]+[e'_{1}]). \tag*{(61)}

Including both shifts, z=(1,1)Tz=(1,1)^{T}, would fail the second target.

For comparison, a further block

M(3)=(0110),d(3)=(0,0)T,(62)M^{(3)}=\begin{pmatrix}0&1\\1&0\end{pmatrix},\qquad d^{(3)}=(0,0)^{T}, \tag*{(62)}

would require z1=z2=0z^{1}=z^{2}=0 and contradict the first target. For any symmetric binary matrix MM, the system Mz=diag⁡(M)Mz=\operatorname{diag}(M) has at least one solution [29]. Thus each target admits a solution individually, but the three targets have no common solution. Establishing whether such a pattern occurs in a Calabi–Yau geometry requires geometric intersection data.

Constructing an odd worldvolume flux

For a setwise invariant divisor, the completion in Proposition 1 can be found by an integer linear system once the full worldvolume lattice is known. Choose an integral basis ωi\omega_{i} of H2(D,Z)freeH^{2}(D,\mathbb{Z})_{\mathrm{free}} and write

σ∗ωj=(TD)ijωi,2ηD=kiωi.(63)\sigma^{*}\omega_{j}=(T_{D})^{i}{}_{j}\omega_{i},\qquad2\eta_{D}=k^{i}\omega_{i}. \tag*{(63)}

Both TDT_{D} and kk have integer entries. Writing ℓD=niωi\ell_{D}=n^{i}\omega_{i}, the condition ℓD+σ∗ℓD=2ηD\ell_{D}+\sigma^{*}\ell_{D}=2\eta_{D} becomes

(I+TD)n=k,n∈Zb2(D).(64)(I+T_{D})n=k,\qquad n\in\mathbb{Z}^{b_{2}(D)}. \tag*{(64)}

Here b2(D)=dim⁡RH2(D,R)b_{2}(D)=\dim_{\mathbb{R}}H^{2}(D,\mathbb{R}) is the second Betti number, and II is the b2(D)×b2(D)b_{2}(D)\times b_{2}(D) identity matrix. Theorem 2 guarantees an integer solution. For any such solution,

FD=∑i(ni−12ki)ωi=12∑i((I−TD)n)iωi.(65)\mathcal{F}_{D}=\sum_{i}\left(n^{i}-\frac{1}{2}k^{i}\right)\omega_{i}=\frac{1}{2}\sum_{i}\left((I-T_{D})n\right)^{i}\omega_{i}. \tag*{(65)}

This requires the worldvolume matrix TDT_{D} and the coefficients of ηDi\eta_{D}^{i} in a full integral worldvolume basis. The bulk toric divisor lattice alone need not supply these data.

For a simple period example, take two integral basis cycles a,ba,b exchanged by σ\sigma, with

∫aη=∫bη=12.(66)\int_{a}\eta=\int_{b}\eta=\frac{1}{2}. \tag*{(66)}

The invariant subgroup is generated by a+ba+b, whose period is one. Its basis can be extended to the full integral cycle basis {a+b,a}\{a+b,a\}, since b=(a+b)−ab=(a+b)-a. Choose an integral class with periods

∫aℓ=1,∫bℓ=0.(67)\int_{a}\ell=1,\qquad\int_{b}\ell=0. \tag*{(67)}

Its σ\sigma-even projection is η\eta, while its σ\sigma-odd projection has periods 12\frac{1}{2} and −12-\frac{1}{2} on a,ba,b. The completed periods are therefore 11 and 00. The basis {a+b,a−b}\{a+b,a-b\} would not be a full integral basis, since recovering aa or bb would require halves.

Zero flux and divisor exchange

Consider the compact hypersurface Q4(dP9)Q_4^{(dP_9)} of Ref. [30]. Its involution has h+1,1=4h^{1,1}_{+}=4, h−1,1=1h^{1,1}_{-}=1, exchanges D7D_7 and D8D_8, and preserves D5D_5, D6D_6, D9D_9. There is one O7 divisor, of class

[R]=[D5]+[D7]+[D8].(68)[R]=[D_5]+[D_7]+[D_8]. \tag*{(68)}

The divisor D6D_6 is a smooth rigid dP9dP_9. An O(1)O(1) instanton configuration on it is explicitly discussed in Ref. [31]. This geometry illustrates how the one-pair shift of Theorem 3 removes a half-integer inherited period at B2,+⋆B^{\star}_{2,+}. Take the invariant target D=D6D=D_6 and the inherited curve C=D6∩D7C=D_6\cap D_7. Equations (28)–(29) of Ref. [31] give

∫X[D6]2[D7]=−1,∫X[D5][D6][D7]=1,\int_X [D_6]^2[D_7]=-1,\qquad\int_X [D_5][D_6][D_7]=1,
∫X[D6][D7]2=−1,∫X[D6][D7][D8]=0.(69)\int_X [D_6][D_7]^2=-1,\qquad\int_X [D_6][D_7][D_8]=0. \tag*{(69)}

Consequently,

I(D6,D7)=12∫X[D6][D7]([D6]+[R])=12(−1+1−1+0)=−12.(70)\begin{aligned} I(D_6,D_7)=\frac{1}{2}\int_X[D_6][D_7]([D_6]+[R]) \\ &=\frac{1}{2}(-1+1-1+0)=-\frac{1}{2}. \tag*{(70)} \end{aligned}

At B2,+⋆=12[R]B^{\star}_{2,+}=\frac{1}{2}[R] with F=0\mathcal{F}=0, this gives

∫CηD6⋆=−12.(71)\int_C\eta^\star_{D_6}=-\frac{1}{2}. \tag*{(71)}

The image C′=D6∩D8C'=D_6\cap D_8 has the same period, while the invariant sum has period −1-1. This is an inherited failure on a rigid invariant divisor, rather than an additional autochthonous period.

For comparison, the alternative background B2,+=12[D5]B_{2,+}=\frac{1}{2}[D_5] used in Ref. [31] gives

∫X[D6][E]([D6]+[D5])=0(72)\int_X[D_6][E]([D_6]+[D_5])=0 \tag*{(72)}

for each basis class [E]=[D5],…,[D9][E]=[D_5],\ldots,[D_9]. All inherited periods on D6D_6 then vanish. This alternative is precisely the one-pair shift:

B2,+⋆−12([D7]+[D8])=12[D5].B^{\star}_{2,+}-\frac{1}{2}([D_7]+[D_8])=\frac{1}{2}[D_5].

Theorem 3 therefore guarantees every inherited period on every smooth connected invariant divisor in this geometry. The shift system reduces to one bit. Choosing [A1]=[D7][A_1]=[D_7] gives

M(D6)=(1),d(D6)=(1),(73)M^{(D_6)}=(1),\qquad d^{(D_6)}=(1), \tag*{(73)}

since

∫X[D6][D7]([D7]+[D8])=−1.\int_X[D_6][D_7]([D_7]+[D_8])=-1.

The equation is z1=1z^1=1, which gives [A]=[D7][A]=[D_7] directly.

The same curve C=D6∩D7C=D_6\cap D_7, now regarded as a curve on the exchanged divisor D7D_7, gives

∫CηD7⋆=12∫X[D7][D6]([R]+[D7])=12(1−2+0)=−12.(74)\begin{aligned} \int_C\eta^\star_{D_7}=\frac{1}{2}\int_X[D_7][D_6]([R]+[D_7]) \\ &=\frac{1}{2}(1-2+0)=-\frac{1}{2}. \tag*{(74)} \end{aligned}

Thus zero worldvolume flux fails on the exchanged pair D7D_7, D8D_8 at B2,+⋆B^{\star}_{2,+}. Nevertheless, the choice FD7=−ηD7⋆\mathcal{F}_{D_7}=-\eta^\star_{D_7} and FD8=ηD8⋆\mathcal{F}_{D_8}=\eta^\star_{D_8} satisfies the image relation and quantizes all periods, as in Theorem 2.

A double-octic example

We use the double-octic construction of Ref. [32] to illustrate a case in which inherited periods permit zero flux but autochthonous periods do not. Consider

X:w2=x18−x28+x38+x48⊂P1,1,1,1,44.(75)X:\quad w^2=x_1^8-x_2^8+x_3^8+x_4^8\subset\mathbb{P}^4_{1,1,1,1,4}. \tag*{(75)}

This is smooth. The coordinate ww has weight four. The projection π:X→P3\pi:X\to\mathbb{P}^3 onto the coordinates [x1:x2:x3:x4][x_1:x_2:x_3:x_4] is a double cover, with involution σ:w↦−w\sigma:w\mapsto-w. If HH is the pullback of the hyperplane class, then7

∫XH3=2,[R]=4H.(76)\int_X H^3=2,\qquad[R]=4H. \tag*{(76)}

The bulk divisor lattice is generated by the invariant class HH, so h−1,1(X)=0h^{1,1}_{-}(X)=0.

Take the smooth connected invariant divisor

D=X∩{x3=x1−x2},h=ι∗H.(77)D=X\cap\{x_3=x_1-x_2\},\qquad h=\iota^*H. \tag*{(77)}

The image of DD is the plane P={x3=x1−x2}≃P2P=\{x_3=x_1-x_2\}\simeq\mathbb{P}^2 in P3\mathbb{P}^3. The restricted double cover D→PD\to P has equation

w2=G(x1,x2,x4),w^2=G(x_1,x_2,x_4),
G=x18−x28+(x1−x2)8+x48.(78)G=x_1^8-x_2^8+(x_1-x_2)^8+x_4^8. \tag*{(78)}

Its branch curve is G=0G=0 in PP, above which the two values of ww coincide at w=0w=0. The preimage of this curve is D∩RD\cap R. The surface DD itself is not contained in RR. The branch curve is smooth because the binary form x18−x28+(x1−x2)8x_1^8-x_2^8+(x_1-x_2)^8 has no repeated projective root, so the double-cover surface DD is smooth as well. Since [D]=H[D]=H,

c1(KD)=h,∫Dh2=2.(79)c_1(K_D)=h,\qquad\int_D h^2=2. \tag*{(79)}

At the common background,

B2,+⋆=2H,ηD⋆=52h.(80)B^{\star}_{2,+}=2H,\qquad\eta^\star_D=\frac{5}{2}h. \tag*{(80)}

The two curves

C±:x1=x2,x3=0,w=±x4(81)C_{\pm}: \quad x_1=x_2,\quad x_3=0,\quad w=\pm x_4 \tag*{(81)}

lie on DD and are exchanged by σ\sigma. Each maps isomorphically to the line {x1=x2, x3=0}\{x_1=x_2,\ x_3=0\} in P3\mathbb{P}^3. Thus C±≃P1C_{\pm}\simeq\mathbb{P}^1. A generic hyperplane intersects this line once. Since hh is the restricted pullback of the hyperplane class, ∫C±h=1\int_{C_{\pm}}h=1. Their sum represents the inherited hyperplane intersection class γH\gamma_H, for which ∫γHh=2\int_{\gamma_H}h=2. Therefore

∫C±ηD⋆=52,∫C++C−ηD⋆=5.(82)\int_{C_{\pm}}\eta_D^\star=\frac{5}{2},\qquad\int_{C_++C_-}\eta_D^\star=5. \tag*{(82)}

Every inherited class is an integer multiple of γH\gamma_H and passes, while the individual component periods fail.

Write γ±=[C±]∈H2(D,Z)\gamma_{\pm}=[C_{\pm}]\in H_2(D,\mathbb{Z}). Each C±C_{\pm} is rational and satisfies ∫C±c1(KD)=∫C±h=1\int_{C_{\pm}}c_1(K_D)=\int_{C_{\pm}}h=1. Adjunction on DD therefore gives

γ±2=2g(C±)−2−∫C±c1(KD)=−2−1=−3.\gamma_{\pm}^2=2g(C_{\pm})-2-\int_{C_{\pm}}c_1(K_D)=-2-1=-3.

Since (γ++γ−)2=2(\gamma_++\gamma_-)^2=2, it follows that γ+⋅γ−=4\gamma_+\cdot\gamma_-=4. In particular, the two classes are distinct. Thus their classes, as well as their representatives, are exchanged. Neither component class is an integer multiple of γH\gamma_H. They are autochthonous. For γ=γ+\gamma=\gamma_+, the σ\sigma-odd difference has square −14-14, agreeing with Eq. (28).

Theorem 2 can be realized explicitly on this surface. Let c±∈H2(D,Z)c_{\pm}\in H^2(D,\mathbb{Z}) be the integral divisor classes of the curves C±C_{\pm} on DD. Their sum is hh, and the involution exchanges them. The flux

FD=12(c+−c−)(83)F_D=\frac{1}{2}(c_+-c_-) \tag*{(83)}

is σ\sigma-odd and of type (1,1)(1,1). At the original bulk background,

ηD⋆+FD=3c++2c−∈H2(D,Z).(84)\eta_D^\star+F_D=3c_++2c_-\in H^2(D,\mathbb{Z}). \tag*{(84)}

The two component periods become −1-1 and 66, respectively. Since c±c_{\pm} are integral classes, this identity proves quantization on every integral two-cycle of DD.

The alternative background B2,+⋆=12HB_{2,+}^\star=\frac{1}{2}H gives ηD=h\eta_D=h, with integer periods on every integral worldvolume two-cycle. This background lies outside the family (37) in this example. Since h−1,1(X)=0h_-^{1,1}(X)=0, every integral [A][A] satisfies σ⋆[A]=[A]\sigma^\star[A]=[A], and the shift changes B2,+⋆B_{2,+}^\star only by an integral shift. It cannot repair the half-integer component periods. Thus no [A][A] in the proposed family satisfies all worldvolume periods with FD=0F_D=0 on this DD, even though its full inherited lattice passes.

Invariant autochthonous classes

Invariant worldvolume classes need not be inherited from bulk divisors. A direct example is provided by the Fermat quintic

X:x05+x15+x25+x35+x45=0⊂P4,(85)X:\quad x_0^5+x_1^5+x_2^5+x_3^5+x_4^5=0\subset\mathbb{P}^4, \tag*{(85)}

with involution

σ:[x0:x1:x2:x3:x4]⟼[x1:x0:x2:x3:x4].(86)\sigma:[x_0:x_1:x_2:x_3:x_4]\longmapsto[x_1:x_0:x_2:x_3:x_4]. \tag*{(86)}

This preserves the quintic and sends ΩX\Omega_X to −ΩX-\Omega_X. Its O7 divisor is R=X∩{x0=x1}R=X\cap\{x_0=x_1\}, and it has the isolated fixed point [1:−1:0:0:0][1:-1:0:0:0]. Both XX and the invariant divisor D=X∩{x4=0}D=X\cap\{x_4=0\} are smooth. The divisor DD is not fixed pointwise.

Let HH denote the hyperplane class of XX and h=ι⋆Hh=\iota^\star H. The bulk divisor lattice is ZH\mathbb{Z}H, so the inherited lattice of DD is generated by γH\gamma_H, with

∫γHh=∫Dh2=5.(87)\int_{\gamma_H}h=\int_D h^2=5. \tag*{(87)}

Now consider the line

C:x0+x1=x2+x3=x4=0.(88)C:\quad x_0+x_1=x_2+x_3=x_4=0. \tag*{(88)}

It lies on DD and is preserved setwise by σ\sigma. Its class is therefore invariant. Since CC is a line, ∫Ch=1\int_C h=1, so [C][C] cannot be an integer multiple of γH\gamma_H. It is autochthonous.

Here [R]=[D]=H[R]=[D]=H, so

B2,+⋆=12H,ηD⋆=h,∫CηD⋆=1.(89)B_{2,+}^\star=\frac{1}{2}H,\qquad\eta_D^\star=h,\qquad\int_C\eta_D^\star=1. \tag*{(89)}

Thus CC supplies an explicit invariant autochthonous class covered by Theorem 1 on a divisor outside the O7 locus. A rigid example appears in Sec. 4.3 of Ref. [30]. The threefold has h1,1(X)=4h^{1,1}(X)=4, while the setwise invariant rational elliptic surface D5≃dP9D_5\simeq dP_9 has h1,1(D5)=6h^{1,1}(D_5)=6. The inherited lattice has rank at most four, so at least two independent invariant classes lie outside its rational span.

References

  1. [1]D. S. Freed and E. Witten, Anomalies in string theory with D-branes, Asian J. Math. 3, 819 (1999), arXiv:hep-th/9907189.arxiv.org/abs/hep-th/9907189
  2. [2]E. Witten, Nonperturbative superpotentials in string theory, Nucl. Phys. B 474, 343 (1996), arXiv:hep-th/9604030.
  3. [3]R. Blumenhagen, M. Cvetic, S. Kachru, and T. Weigand, D-Brane Instantons in Type II Orientifolds, Ann. Rev. Nucl. Part. Sci. 59, 269 (2009), arXiv:0902.3251 [hep-th].DOI
  4. [4]A. Collinucci and R. Savelli, On Flux Quantization in F-Theory, JHEP 02, 015, arXiv:1011.6388 [hep-th].DOI
  5. [5]T. W. Grimm, M. Kerstan, E. Palti, and T. Weigand, On Fluxed Instantons and Moduli Stabilisation in IIB Orientifolds and F-theory, Phys. Rev. D 84, 066001 (2011), arXiv:1105.3193 [hep-th].
  6. [6]M. Cicoli, C. Mayrhofer, and R. Valandro, Moduli Stabilisation for Chiral Global Models, JHEP 02, 062, arXiv:1110.3333 [hep-th].DOI
  7. [7]M. Kerstan and T. Weigand, Fluxed M5-instantons in F-theory, Nucl. Phys. B 864, 597 (2012), arXiv:1205.4720 [hep-th].DOI
  8. [8]M. Bianchi, A. Collinucci, and L. Martucci, Magnetized E3-brane instantons in F-theory, JHEP 12, 045, arXiv:1107.3732 [hep-th].DOI
  9. [9]M. Bianchi, A. Collinucci, and L. Martucci, Freezing E3-brane instantons with fluxes, Fortsch. Phys. 60, 914 (2012), arXiv:1202.5045 [hep-th].DOI
  10. [10]M. Demirtas, M. Kim, L. McAllister, J. Moritz, and A. Rios-Tascon, Small cosmological constants in string theory, JHEP 12, 136, arXiv:2107.09064 [hep-th].
  11. [11]L. McAllister, J. Moritz, R. Nally, and A. Schachner, Candidate de Sitter vacua, Phys. Rev. D 111, 086015 (2025), arXiv:2406.13751 [hep-th].arxiv.org/abs/2406.13751
  12. [12]J. Louis, M. Rummel, R. Valandro, and A. Westphal, Building an explicit de Sitter, JHEP 10, 163, arXiv:1208.098 [hep-th].
  13. [13]C. Crinò, F. Quevedo, A. Schachner, and R. Valandro, A database of Calabi-Yau orientifolds and the size of D3-tadpoles, JHEP 08, 050, arXiv:2024.13115 [hep-th].
  14. [14]R. E. Gompf and A. I. Stipsicz, 4-Manifolds and Kirby Calculus, Graduate Studies in Mathematics, Vol. 20 (American Mathematical Society, Providence, 1999).DOI
  15. [15]V. I. Arnold, Distribution of ovals of the real plane of algebraic curves, of involutions of four-dimensional smooth manifolds, and the arithmetic of integer-valued quadratic forms, Functional Analysis and Its Applications 5, 169 (1971).DOI
  16. [16]A. Degtyarev, On the Pontrjagin–Viro form (1997), arXiv:alg-geom/9705011 [math.AG].arxiv.org/abs/alg-geom/9705011
  17. [17]M. Demirtas, C. Long, L. McAllister, and M. Stillman, The Kreuzer–Skarke Axiverse, JHEP 04, 138, arXiv:1808.01282 [hep-th].arxiv.org/abs/1808.01282
  18. [18]V. M. Mehta, M. Demirtas, C. Long, D. J. E. Marsh, L. McAllister, and M. J. Scott, Superradiance in string theory, JCAP 07, 033, arXiv:2103.06812 [hep-th].DOI
  19. [19]E. Sheridan, F. Carta, N. Gendler, M. Jain, D. J. E. Marsh, L. McAllister, N. Righi, K. K. Rogers, and A. Schachner, Fuzzy axions and associated relics, JHEP 09, 016, arXiv:2412.12012 [hep-th].
  20. [20]M. Jain, E. Sheridan, D. J. E. Marsh, E. Heyes, K. K. Rogers, and A. Schachner, Bayesian inference on Calabi-Yau moduli spaces and the axiverse: Experimental data meets string theory, Phys. Rev. D 114, 026026 (2026), arXiv:2512.00144 [hep-th].arxiv.org/abs/2512.00144
  21. [21]A. Calabri, C. Ciliberto, and M. M. Lopes, Numerical godeaux surfaces with an involution (2005), arXiv:math/0501551 [math.AG].DOI
  22. [22]M. Borovoi and D. A. Timashev, Galois cohomology and component group of a real reductive group, Israel J. Math. 259, 937 (2024), arXiv:2110.13062 [math.GR].DOI
  23. [23]M. Jain, Can componentwise D7-tadpole cancellation yield a chiral Standard Model? A no-go for smooth O7-orientifolds (2026), arXiv:2610.28730 [hep-th].
  24. [24]M. Jain, F. Carta, and D. J. E. Marsh, The C4 axiverse in Kreuzer–Skarke orientifolds: rigid O(1) instantons, and D7 tadpole constraints (forthcoming).
  25. [25]M. Kreuzer and H. Skarke, Complete classification of reflexive polyhedra in four-dimensions, Adv. Theor. Math. Phys. 4, 1209 (2000), arXiv:hep-th/0002240.DOI
  26. [26]J. Moritz, Orientifolds Kreuzer-Skarke (2023), arXiv:2305.06363 [hep-th].
  27. [27]R. Altman, J. Carifio, X. Gao, and B. D. Nelson, Orientifold Calabi-Yau threefolds with divisor involutions and string landscape, JHEP 03, 087, arXiv:2111.03078 [hep-th].DOI
  28. [28]X. Cao, H. Gao, and X. Gao, Orientifold Calabi-Yau threefolds: divisor exchanges and multi-reflections, JHEP 10, 188, arXiv:2407.02565 [hep-th].arxiv.org/abs/2407.02565
  29. [29]I. Minevich, Symmetric Matrices over F2 and the Lights Out Problem (2012), arXiv:1206.2973 [math.RA].arxiv.org/abs/1206.2973
  30. [30]R. Blumenhagen, V. Braun, T. W. Grimm, and T. Weigand, GUTs in Type IIB Orientifold Compactifications, Nucl. Phys. B 815, 1 (2009), arXiv:0811.2936 [hep-th].DOI
  31. [31]D. Green and T. Weigand, Retrofitting and the μ problem (2009), arXiv:0906.0595 [hep-th].arxiv.org/abs/0906.0595
  32. [32]A. Collinucci and R. Savelli, On Flux Quantization in F-Theory II: Unitary and Symplectic Gauge Groups, JHEP 08, 094, arXiv:1203.4542 [hep-th].DOI

Paper details

vlnc:W_m2hf27df
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents