WorksMathematics
Spherical space forms in K3: an instanton parity argument
Abstract
Let be a connected subdiagram of the Dynkin diagram , and let be the spherical three-manifold which bounds the corresponding plumbing of -spheres. We show that if a negative-definite four-manifold with and oriented boundary is embedded in the K3 surface, then the intersection form of is the negative-definite root lattice . As a consequence, the traces of and surgery on the left-handed trefoil do not embed in K3. Heegaard Floer correction terms and the theory of lattices reduce the theorem to the exclusion of one smaller lattice for each of , , and . In these cases, stretching the K3 surface along exhibits a zero-dimensional moduli space of instantons as the zero set of a section of a real two-plane bundle over a compact surface. The quaternionic nature of the linearized anti-self-duality equations on K3 shows that the Euler number of this bundle is even, while the corresponding Donaldson invariant is odd.
Foreword
With the exception of this foreword, the present paper is entirely AI-generated, in response to prompts provided by the human authors. The mathematics was produced by GPT-6 Astra, and the paper was redrafted by Claude Fable 5.2. The main result is Theorem 1.1. A particular case states that the 4-manifold obtained by adding a 2-handle to the 4-ball along a left-handed trefoil with framing cannot be smoothly embedded in a K3 surface. (See Corollary 1.2.)
These results arose from prompts motivated by a question which the authors learned of from Anubhav Mukherjee. Let be the standard 3-sphere in the neck of the connected sum . Does the connected sum contain any sphere different from ? Here one can interpret different as up to isotopy; or one can ask whether is the only non-zero class in that is represented by an embedding. In the latter version, the question is closely related to the question of whether a diffeomorphism of can “mix” the homology of the two summands. The clearest strategy to construct such a mixing diffeomorphism starts with a hypothetical embedded 2-sphere in with such that the class has non-zero projection onto both sides. Then reflection in arises from a diffeomorphism of which does not preserve the direct sum decomposition. (Indeed, if is the sum of two such primitive classes, then it turns out that this would imply that the diffeomorphism group of acts by the full isometry group of the homology lattice.) To try and construct such a 2-sphere, one might start with a cuspidal elliptic fiber on one side, and seek on the other side an embedded disk with relative self-intersection number and boundary a trefoil in the 3-sphere . Tubing the trefoil in to the cusp gives the sought-after 2-sphere. But finding the disk in punctured is equivalent to embedding the 4-dimensional handlebody mentioned above, which Corollary 1.2 says cannot be done. Of course, only the simplest construction of such a sphere is ruled out by this argument, because the hypothetical sphere might meet in an arbitrary knot or link instead of the trefoil.
ChatGPT proved the corollary first, for the case of the framed left-handed trefoil. It began by giving an argument conditional on the superconformal simple type conjecture [12, 7]. If embeds in , then the manifold obtained by replacing this piece with an plumbing is a manifold whose Seiberg-Witten invariants can be computed sufficiently to obtain a contradiction to that conjecture. The AI then found an unconditional proof using Donaldson invariants, which is the one presented here.
In outline, here is the proof by contradiction as it applies to a hypothetical embedding of in . With a suitable bundle on , one obtains a zero-dimensional ASD moduli space, and the Donaldson invariant is known to be . On the other hand, stretching the metric along leads to a gluing description which presents the moduli space as the zero set of a section of an obstruction bundle , where has real rank 2 and is a compact, oriented 2-manifold. This describes the Donaldson invariant as the Euler class . However, has an interpretation as the element of arising from the index of the ASD deformation complex. The quaternionic structure of the tangent bundle of implies that this element is even, and it follows that the Donaldson invariant of , computed as , is even, which contradicts the known result.
The justification of the required gluing result in section 5 of the paper is verbose. The AI initially found no suitable models for the required analytic steps in the literature. The human authors pointed to the cited papers of Taubes, which provided both the overall strategy for the gluing theorem and the analytic details. With this background, GPT-6 wrote a gluing argument which was later condensed by Fable.
After completing the proof of the corollary, the generalization given in Theorem 1.1 was suggested by the authors, as something that might follow from similar arguments. As noted in the paper, the instanton parity argument is needed only in the cases , , and , to rule out the lattices , , and respectively.
The above corollary and the parity argument in its proof are both new to the authors. Distributing the preprint in this form, the authors would be glad to hear from others if these results are known.
Acknowledgements.
The work of the first author was supported by NSF grant DMS-2304877. Both authors were supported by Simons Foundation Awards #994330 and #14760-09, Simons Collaboration on New Structures in Low-Dimensional Topology.
Introduction
Let be a Dynkin diagram of type , or . We use the same letter for the corresponding root lattice, which we take to be positive definite. Let be the four-manifold obtained by plumbing disk bundles of Euler number over the two-sphere according to the diagram , so that the intersection form of is , and let be its oriented boundary. Thus is the quotient of by a finite subgroup of . This paper is concerned with embeddings of the manifolds in the K3 surface. Our main result is the following.
Theorem 1.1. Let be a connected subdiagram of the Dynkin diagram , and let be a compact, connected, codimension-zero submanifold of the K3 surface whose oriented boundary is . Suppose that is negative definite and that . Then the intersection form of is isomorphic to .
The diagrams to which the theorem applies are for , for , and , and . With the hypotheses of the theorem, the group is free and is zero (Proposition 7.1), so has the same integral homology as the plumbing .
In three cases, the conditions which the theorem imposes on do not need to be assumed. If is , or , and is embedded in the K3 surface in any manner, then it divides the K3 surface into two pieces, and one of the two is negative definite, has oriented boundary , and has vanishing first homology. Theorem 1.1 applies to that piece, whose intersection form is therefore . See Theorems 2.1 and 7.2.
For a knot and an integer , let denote the trace of -surgery on : the four-manifold obtained by attaching a two-handle to the four-ball along with framing .
Corollary 1.2. Let be the left-handed trefoil. Then the trace does not admit an orientation-preserving smooth embedding in the K3 surface. The same is true of .
Proof. As oriented manifolds, the boundary of is [1 Lemma 1.2], and the boundary of is [15]. The traces are simply connected, and their intersection forms are and respectively, rather than and . □
Outline of the proof
The first part of the proof of Theorem 1.1 uses Heegaard Floer correction terms and the theory of lattices. The manifold is spin, so the inequality of Ozsváth and Szabó [18] bounds in terms of the correction terms of , and also determines modulo 8. The classification of even definite lattices of small rank and determinant then shows that the intersection form of is when is , or . The same holds for , by an argument which also uses Donaldson’s diagonalization theorem. In the four remaining cases, there is one possibility for the intersection form other than :
To exclude these four possibilities we use instantons. Consider the case , and suppose that a K3 surface is decomposed as , where and has intersection form . Let be an bundle with whose second Stiefel–Whitney class is non-zero on the generator of . The moduli space of anti-self-dual connections in has formal dimension zero, and the corresponding Donaldson invariant, which counts the points of this moduli space with signs, is . We examine the moduli space as the metric on is stretched along . On the manifold with a cylindrical end attached, the bundle carries a reducible anti-self-dual connection of action , asymptotic to a reducible flat connection on . A dimension count shows that, when the neck is long, every anti-self-dual connection in is close to on , and on is close to a point of a moduli space of instantons asymptotic to , where is a compact two-manifold. The connection , however, belongs to a moduli space of formal dimension . Its stabilizer is a circle and, however the equations are perturbed, the cohomology group of its deformation complex is not zero: it is a copy of on which the stabilizer acts with weight one. In this situation, gluing theory describes the moduli space on not as a copy of but as the zero set of a section of a real two-plane bundle , the bundle of obstruction spaces. The Donaldson invariant is therefore equal, modulo 2, to the Euler number of .
The proof is completed by showing that this Euler number is even. The gluing construction provides a family of connections on parametrized by , and the index of the family of linearized operators is the element of . For a hyperkähler metric on , the operator can be identified with a Dirac operator coupled to the real vector bundle , and so commutes with a quaternionic structure. The index of any family of such operators therefore lies in the image of , and in particular its second Stiefel–Whitney class is zero. It follows that , so the Euler number of has the same parity as the Euler number of the surface , which is even.
The cases , and are dealt with in the same way, and differ only in the numerical data.
The method of gluing in the presence of an obstruction goes back to Taubes [22] and Donaldson [3], and is a standard part of the theory [4, 5]. Obstructed gluing at a reducible connection, for cobordisms between rational homology spheres, is studied in [2 Section 5]. We have collected the analysis in § 6, and the reader who is willing to take Proposition 4.1 on trust can omit that section.
Organization
Organization. Sections 2–4 deal with the case . § 3 contains the lattice-theoretic part of the argument, § 4 describes the moduli spaces on the two pieces, and § 5 contains the parity argument. The transversality and gluing results on which these depend are proved in § 6. The cases , and are treated in § 7, and the remaining cases in § 8, where the proof of Theorem 1.1 is completed.
The case : constraints on the lattice
In this section and the next two, denotes the manifold . It is the quotient of by the binary octahedral group , a group of order 48 whose abelianization is . For this manifold we prove a statement which is a little stronger than Theorem 1.1, in that the hypotheses on become conclusions.
Theorem 2.1. Suppose that is smoothly embedded in a surface . Then separates into two pieces,
where has oriented boundary and is negative definite. The groups and are zero, and the intersection form of is isomorphic to . In particular, and .
The proof occupies this section and the next three. Here we prove the following.
Proposition 2.2. In the situation of Theorem 2.1, one of the two pieces, , is negative definite and has oriented boundary . The groups and are zero, and the intersection form is isomorphic either to or to .
Proof. Since , the submanifold separates into two pieces. Let denote either of them. In the Mayer–Vietoris sequence, the groups and are zero, so injects into and is therefore free, while is a quotient of and is therefore finite. The exact sequence of the pair and Poincaré–Lefschetz duality then give
The left-hand side is , so and . The form is even, because is spin.
The manifold is an -space and is symplectic, so one of the two pieces has [16 Theorem 1.2]. (If were positive on both sides, then the Ozsváth–Szabó invariants of would factor through , and they would vanish, which they do not for a symplectic manifold.) We call this piece . It is negative definite, because is non-degenerate. To see which way the boundary of is oriented, we use the correction terms of . There are two structures on , both of them spin, and their correction terms are
See [1 Lemma 1.2], where the values are given for . If is a negative-definite spin manifold whose boundary is a rational homology sphere, and is the spin structure which the boundary inherits, then
by [18 Theorem 9.6]. Since both correction terms of are negative, the oriented boundary of is and not , and the rank is at most . It is at least , because .
Let be the lattice , which is even and positive definite, of determinant . Its discriminant group has one non-zero element, on which the discriminant quadratic form takes a value which is either or modulo . The Milgram formula [14 Appendix 4] gives
so if , and if . So either , in which case , or and . In the second case, the lattice has an even unimodular overlattice of index . This overlattice is positive definite of rank , and is therefore isomorphic to [20 Chapter V]. In it, is the orthogonal complement of a root, so is isomorphic to . □
To prove Theorem 2.1, it remains to show that cannot be .
The moduli spaces when has rank one
We suppose from now on that the second alternative in Proposition 2.2 occurs, and we write for . So is a K3 surface, decomposed as
@ and our aim is to derive a contradiction. By Proposition 2.2, the first homology groups of and are zero. The group is generated by a class with , and we have
The class is primitive in , because the intersection form of is even.
The bundle
Since is primitive and has square , there is an automorphism of the lattice which carries to , where and are the standard basis of a hyperbolic summand [8 Proposition 3.3(i)]. Let and be the standard basis of a second hyperbolic summand, orthogonal to the first, and let be the class dual to . Then
Let be the bundle with and , and let be the associated bundle, so that
We adopt the usual conventions for such bundles, as in [11]: we consider connections in , and the gauge group consists of the automorphisms of of determinant one, so that an irreducible connection has stabilizer in . We write for the action of a connection, normalized so that on a closed manifold. For the bundle we have , and the moduli space of anti-self-dual connections has formal dimension
The corresponding Donaldson invariant, which counts the points of a regular moduli space with signs, is
This is the standard calculation for the K3 surface [11 Proposition 7.13]. We shall use only the fact that the number of points is odd.
Flat connections on
The restriction of to is the generator of with , and its restriction to is the non-zero element of . The flat connections in , modulo gauge transformations of determinant one, therefore correspond to the conjugacy classes of homomorphisms whose determinant is the non-trivial character .
Lemma 3.1. (a) There are two such conjugacy classes. They are represented by and by , the pull-back of the irreducible two-dimensional representation of the symmetric group by the quotient maps .
(b) The rho-invariants of the corresponding flat connections in are
(c) Their Chern–Simons invariants satisfy
Here is the Atiyah–Patodi–Singer rho-invariant of the flat bundle , and cs is the Chern–Simons invariant with values in . Our sign conventions for both are fixed by the way they enter the formulae (7) and (8) below. The proof of the lemma is a calculation with the character table of , and is given in § 4.4.
The flat connection is reducible: because the characters and are distinct, its stabilizer in is a circle. The flat connection is irreducible. Both are non-degenerate, in that
because is finite. We write for the dimension of the stabilizer of , so and .
Remark 3.2. The holonomy group of the connection has order two, so its centralizer in is a copy of . The elements in the non-identity component of are automorphisms of which do not lift to automorphisms of of determinant one, and they play no role in what follows.
Moduli spaces on and
Let and be the manifolds obtained from and by attaching cylindrical ends and , with metrics which are products on the ends. For , let be the Riemannian manifold
which is a copy of containing a neck of length .
Let be either or , and let be either or . We write for the moduli space of anti-self-dual connections in on which are asymptotic to on the end and have action . By the Atiyah–Patodi–Singer index theorem, the formal dimension of this moduli space is
where and .
On , the action is determined modulo by . To find these residues, we consider reducible connections on . A reduction of to the circle has the form , where is a complex line bundle with , so that with odd. Because and , each such line bundle carries an anti-self-dual connection on with curvature in , unique up to gauge. Since , the corresponding connection in has action
The reductions for and give the same connection in . So the reducible anti-self-dual connections on have action , , and so on, and there is one of each. We write for the reducible connection with action . The limit of on the end is , because is the only reducible flat connection. So on , and from Lemma 3.1(c) we obtain
The residues on are obtained by subtracting these from . In the following table we list the pairs of non-negative values and allowed by (8) with , together with the formal dimensions and of and given by (7), (4) and Lemma 3.1(b).
In each row, is zero, as it must be because has formal dimension zero.
The connection belongs to a moduli space of formal dimension . Its deformation complex decomposes according to the decomposition . The summand with coefficients in has , the Lie algebra of the stabilizer of , and its other cohomology groups are zero because and are zero. The summand with coefficients in is a complex of complex vector spaces, with , and its index is . So has complex dimension at least one, with equality if and only if .
In the next proposition, the moduli spaces are defined using perturbations of the anti-self-duality equations on and , which are of the type described in § 6.1 and are supported in compact subsets of the interiors of and . The same perturbations are used on to define . Perturbing the equations does not affect the formal dimensions or the residues of . If the perturbations are small, then there is still exactly one reducible solution on with action , and we continue to call it .
Proposition 3.3. The perturbations can be chosen so that the following hold.
(a) The moduli space consists of the single point . The cohomology group is zero, and is isomorphic to , with the stabilizer of acting with weight one.
(b) The moduli spaces and are empty. The moduli space is a compact smooth two-manifold. It consists of irreducible connections with .
(c) Let be a sequence tending to infinity, and let belong to . Then, after passing to a subsequence, there is a point in such that the connections converge to on compact subsets of and to on compact subsets of . No action is lost in the limit: .
Proof. In § 6.1 we show that the perturbations can be chosen so that the following hold. First, the connection has and , and the irreducible parts of the moduli spaces on are regular (Lemma 5.1). Second, the moduli spaces on with contain no reducibles and are regular (Lemma 5.2). Third, the action of every solution on or is greater than ; since the residues of are and on , and and on , the action is then non-negative.
A regular moduli space of negative formal dimension is empty. So the irreducible part of is empty, which proves (a), and the two moduli spaces on in the second and third rows of (9) are empty also. The moduli space is a smooth two-manifold. To see that it is compact, recall that a sequence in has a subsequence which converges weakly to a point of some with . If is less than , then the difference is accounted for by bubbles, each of which carries action at least , and by anti-self-dual connections on the cylinder . The only possibilities allowed by the residues are and , and these are empty.
The proof of (c) is similar. A sequence as in (c) has a subsequence which converges weakly to a pair of solutions on and , belonging to moduli spaces and , where and are non-negative and . Equality holds only if there are no bubbles and no action is lost in the neck, in which case . Among the moduli spaces on with , only can be non-empty. So and . Since is non-negative and congruent to or modulo , it is equal to , the limiting flat connection is , and no action is lost. By (a), the limit on is . □
Proof of Lemma 3.1
We refer to [9 Appendix A.4] for the character table of . The group has three irreducible two-dimensional representations: the defining representation of , the tensor product , and . The first two have determinant , and has determinant . A reducible representation with determinant is the sum of two characters whose product is , and so is . This proves (a).
For (b), let be the arguments of the eigenvalues of in the representation . With oriented as the boundary of the negative-definite plumbing, the rho-invariant is given by
The representations and factor through the quotient of , and each element of is the image of two elements of . In the following table, the third column is the sum of over the two elements of lying over an element of the given conjugacy class in .
| class in | size | |||
| (12) | 6 | 2 | ||
| (123) | 8 | 0 | ||
| (1234) | 6 | 6 |
Table 2.
The remaining conjugacy classes in make no contribution, because and are trivial there. So
For (c), recall that is cyclic of order 48, generated by . We have . To find we restrict to subgroups. On a Sylow 2-subgroup of , which is a generalized quaternion group of order 16, the representation is the sum of the trivial character and a non-trivial one, so is zero modulo 16. On a cyclic subgroup of order 3, the representation coincides with , so is equal to modulo 3. Hence . With the conventions of [9 Appendix B], the Chern–Simons invariant of is . Since and have the same determinant, it follows that
In (8), we have used the fact that if two connections on are asymptotic to and , then the difference of their actions is congruent to modulo . For this, the Chern–Simons invariant is defined using the form with , which is the normalization for which an anti-self-dual connection on a closed manifold has action .
The parity argument
We continue with the notation of § 4. The gluing theorem of § 6 leads to the following description of the moduli space on when is large. We write for the space of irreducible connections in modulo the gauge group , and for the perturbation of the equations on , so that is the set of solutions of in .
Proposition 4.1. If is sufficiently large then, after the addition of a further small perturbation to , the following hold.
(a) There are an open subset of , an oriented real two-plane bundle , and a smooth bundle map from to the bundle over whose fibre at is , such that the space
is cut out transversely and is a compact surface, diffeomorphic to .
(b) The moduli space is regular and is contained in . So is the zero set of the section of given by , and vanishes transversely. The number of its zeros is odd.
The surface is a thickened moduli space, of the kind which arises whenever a moduli space is described by a finite-dimensional obstruction bundle. To say that it is cut out transversely means the following. Write . At a point of , the linearization of the equation , together with the Coulomb gauge condition, is the operator
and we require this operator to be surjective. Its kernel is then the tangent space to at .
The last statement in part (b) holds because the number of points in the moduli space, counted with signs, is the Donaldson invariant (6). On the other hand, the number of zeros of a transverse section of is given, modulo 2, by the second Stiefel–Whitney class. So Proposition 4.1 has the following consequence, in which is the fundamental class with coefficients :
Because acts freely on the irreducible connections, there is a universal bundle over . Its restriction to carries a family of connections parametrized by . So the operators
for in , form a family of elliptic operators over , which has an index in .
Lemma 4.2. In , we have
Proof. The operators in (10) form a family of surjective Fredholm operators over , and the kernels form the bundle . So the index of this family is . On the other hand, differs from the operator by terms which are compact: the derivative of the perturbation is compact, and the terms involving have finite rank. The index of the family is . □
Proposition 4.3. Let be a K3 surface with any Riemannian metric, let be a compact space, and let be an bundle carrying a family of connections on parametrized by . Then the index of the family of operators lies in the image of the map . In particular, its first and second Stiefel–Whitney classes are zero.
Proof. The index of the family does not depend on the Riemannian metric, so we may suppose that the metric is hyperkähler. The spin bundles and of are quaternionic line bundles: we regard their fibres as right -modules, and Clifford multiplication by one-forms and the Dirac operator are then -linear. For the hyperkähler metric, the bundle is flat and trivial. Let be a parallel section of of unit length. Clifford multiplication on gives isomorphisms of real vector bundles
and because is parallel, these isomorphisms carry the operator to the Dirac operator , up to constant factors. In the same way, the operator becomes the Dirac operator coupled to the real vector bundle with the connection . This coupled Dirac operator is again -linear. The quaternionic structure is the same for every member of the family, so the index of the family is the class of a virtual quaternionic vector bundle on . Finally, the real vector bundle which underlies a quaternionic vector bundle has and . □
Proof of Theorem 2.1. By Proposition 2.2, we have only to exclude the case considered in § 4, in which is . In that case, let and be as in Proposition 4.1. By Proposition 4.3, the class in (12) has and , and so the Stiefel–Whitney classes of and are equal. Because is oriented, it follows that is orientable and that
The right-hand side is the Euler number of modulo 2, which is zero because is a closed orientable surface. This contradicts (11). (If is empty, then is empty and (11) fails for that reason.) So is isomorphic to , and the values of and follow. □
Regularity and gluing
In this section we prove the transversality statements which were used in the proof of Proposition 3.3, and we prove Proposition 4.1. The methods are not new. The technique for gluing in the presence of an obstruction, in which the equations are first solved modulo a finite-dimensional space associated with the small eigenvalues of the linearized operator, goes back to Taubes [22]; see also [3] and [5 Section 7.2]. For manifolds with cylindrical ends, we follow [4 Chapter 4]. Some constants in our argument grow polynomially with the length of the neck, while the error terms decay exponentially, and to deal with this one needs a slice theorem in which the constants are controlled by the lowest eigenvalue of the Laplacian . Results of that kind are also due to Taubes: see [23 Section 6] and [21 Lemma A.1], and the account in [6 Theorems 1.1 and 1.2]. The small eigenvalues of Dirac-type operators on manifolds with long necks, and the obstruction bundles to which they give rise in families, are discussed in [17]. The perturbations of the equations which we use are the holonomy perturbations of [10].
What is particular to the present situation is that the ends are rational homology spheres, on which the flat connections may be reducible. Moduli spaces of this sort are studied in detail in [13] and [2], and we refer to those papers at several points.
Perturbations and regularity
Let be an oriented Riemannian four-manifold with a cylindrical end , where is a rational homology sphere, and let be an bundle, with the conventions of § 4. For a non-degenerate flat connection on the end, the moduli spaces are defined in the standard way [4 Chapter 4], using Sobolev spaces with and a small exponential weight on the end. For a solution , we write , and for the cohomology groups of its deformation complex, so that is the Lie algebra of the stabilizer of and is the cokernel of the linearized equations. The solution is regular if . The elements of are represented by self-dual forms which decay exponentially on the end.
The equations are perturbed using holonomy perturbations, as in [10 Section 3]. In place of we consider
where is a sum of terms of the following kind. One takes a ball in the interior of , a map with , and a self-dual form supported in . The corresponding term is obtained from the holonomy of around the loops , for in , by multiplying by and projecting to . The perturbations form a Banach space, they are supported in a compact subset of the interior of , and they are gauge-equivariant: if is compatible with a reduction , then takes values in the summand . Their use on manifolds with cylindrical ends, in the presence of reducible connections, is described in [13 Section 4] and [2 Section 3.2].
In [10], the map is required to be a submersion. We require more, namely that each map is an embedding of . The derivative of is then built from pull-backs by diffeomorphisms, so that it and its formal adjoint are bounded on Sobolev spaces of every order, negative orders included, which we need in the gluing argument. The usual transversality arguments apply without change to perturbations of this restricted kind.
A solution of the perturbed equations with action satisfies
where is a positive constant which depends only on the normalization of . The norm of is bounded independently of , by a constant which is small when is small. So for any , all solutions have when is sufficiently small.
If a solution is compatible with a reduction , then the linearization of the equations at in Coulomb gauge is the sum of a tangential operator, which acts on ordinary one-forms, and a normal operator , which acts on one-forms with values in and is complex-linear. The circle in the stabilizer of acts on with weight one.
Lemma 5.1. Suppose that and . Let be a flat connection on the end with whose stabilizer is a circle. Let be a reduction of which is compatible with on the end, for which is non-zero in and for which the normal operators have complex index . Then for all sufficiently small perturbations ,
there is exactly one solution compatible with this reduction, up to gauge. For a residual set of small perturbations, the normal operator is injective, so that
and the irreducible solutions in the moduli spaces are regular, for all and for all below any given bound.
Proof. When , the solutions compatible with the reduction are the anti-self-dual connections in with curvature in , and there is exactly one of these up to gauge, because and . The tangential operator at this solution has kernel , and its cokernel consists only of the constants. So the solution persists when is small, by the implicit function theorem. No others appear, because solutions in a line bundle cannot bubble and cannot lose action on the end; compare [13], Section 4.4.
The normal operators at these solutions are complex Fredholm operators of index . Among such operators, those whose kernel has complex dimension form a submanifold of real codimension , whose normal space at an operator with kernel and cokernel is . Let and be the kernel and cokernel of . It is enough to show that the derivative of at maps onto . For surjectivity onto a finite-dimensional space is an open condition, so the map is then transverse to all these submanifolds for near zero, and by the Sard–Smale theorem is injective for a residual set of small . The surjectivity is proved by the argument which is used in [13], Section 4.6 and [2], Section 3.5 to obtain transversality in the directions normal to the reducible connections. A non-zero element of is detected by the derivative of the holonomy around some loop in a compact subset of , because the flat line bundle has no parallel sections. A non-zero element of cannot vanish on an open set. And the only constraint on the derivative of a holonomy perturbation in the directions normal to the reducible connections is equivariance under the circle, which for a real-linear map from to is the same as complex-linearity. The perturbations can be taken to vanish to first order at along the reducible locus, so that does not move.
The regularity of the irreducible solutions for generic is proved as in [10], § 4 and § 6; see also [13], Section 4.6. □
Lemma 5.2. Suppose that and , and that is non-zero on the end. Then for a residual set of small perturbations , the moduli spaces contain no reducible solutions and are regular, for all and for all below any given bound.
Proof. Since , every reducible connection is compatible with a reduction . The class reduces modulo 2 to , so it is non-zero on the end; and it is non-zero in , because has no torsion. The solutions compatible with such a reduction are the zeros of a Fredholm section of index , which is negative, so for generic there are none; compare [13], Section 4.4. The regularity of the irreducible solutions is proved as before. □
These lemmas supply what was needed in the proof of Proposition 3.3. Lemma 5.1 applies to the manifold of § 4, with the reduction for which : the formal dimension of is the sum of , from the stabilizer, and twice the complex index of the normal operator. Lemma 5.2 applies to , where . The lower bound for the action comes from (13), with .
The gluing theorem
Let be a closed, oriented four-manifold, decomposed as along a rational homology sphere , and let be an bundle, with the same conventions as before. We form , and as in § 4.3. We fix holonomy perturbations on and , and we use their sum on . Let be a flat connection on with whose stabilizer is a circle. We make the following hypotheses.
is a reducible solution on , asymptotic to , with . The space is isomorphic to , and the stabilizer , which is a circle, acts on it with weight one.
is a compact two-manifold which is a union of components of the moduli space of solutions on asymptotic to . It consists of irreducible, regular solutions.
If is a sequence tending to infinity, and belongs to , then after passing to a subsequence there is a point in such that the connections converge to on compact subsets of and to on compact subsets of , without loss of action.
Restriction to the end identifies with . Let be the moduli space of solutions framed at infinity, which is a principal -bundle over , and let
be the associated complex line bundle.
Theorem 5.3. Suppose that hypotheses (i)–(iii) hold. Then there are constants and such that the following is true for all . Let be a holonomy perturbation on , supported in the complement of the neck, whose norm is at most . Then assertion (a) of Proposition 4.1 holds for the equations on defined by , with a diffeomorphism which is covered by an isomorphism of oriented bundles ; and the moduli space is contained in . For a residual set of such , the moduli space is regular.
Remark 5.4. The gluing parameter for the pair is an element of . Because maps isomorphically to , the gluing parameter is absorbed by the action of , which is why is a copy of rather than of . In the situation of Remark 3.2, there is a second way to identify the two copies of over the neck, using the non-identity component of . The bundle which results is not : its second Stiefel–Whitney class differs from by the image of the non-zero element of under the coboundary map to , and this image is not zero because and are zero.
Proof of Theorem 5.3
We first take .
Approximate solutions. Because is non-degenerate, the connection approaches exponentially on the end of , and so do the solutions representing the points of on the end of , uniformly in . We use cut-off functions to make and equal to at distance greater than along the ends, and we join the results across the neck of . This requires an identification of the two copies of . The possible identifications form a copy of , and because maps onto , they all give gauge-equivalent connections. So we obtain connections , whose gauge-equivalence classes depend only on , with
and with similar bounds for the derivatives with respect to . Here and below, , and denote positive constants which are independent of and . The manifolds have uniformly bounded geometry, so the constants in the Sobolev and elliptic estimates do not depend on .
In the same way, the self-dual forms on which represent the elements of can be cut off and transferred to . The result depends on the identification, through the action of on , and we obtain linear maps
The linearized equations. Let be the derivative of at .
Lemma 5.5. There is a constant such that, for all and all sufficiently large , the operator has exactly two eigenvalues in the interval , counted with multiplicity, and no other spectrum in . If is the projection onto the span of the two eigenvectors, then is an isomorphism from to the image of .
Proof. The forms are annihilated by up to an error of size , which accounts for two small eigenvalues and for the last statement. If the lemma were false, there would be self-dual forms on , with , which have norm , are orthogonal to the image of , and have in . On the neck, where is the pull-back of , we identify self-dual forms with time-dependent one-forms on by , and becomes the operator
where is the Hodge Laplacian on . This Laplacian is bounded below by a positive constant, because , so the norm of cannot escape into the neck. A subsequence therefore converges on compact subsets of and to solutions of the adjoint equations, not both zero. But the limit on is zero, because the points of are regular, and the limit on is an element of which is orthogonal to . □
The space , which is not zero, does not enter in (15). It does affect the Laplacian on zero-forms, and so the Coulomb gauge condition. For , and for connections whose distance from is at most , we have
This is obtained from a Poincaré inequality on a compact subset of , where the connections representing the points of are irreducible, by integrating along the neck. So has an inverse , whose norms as an operator between Sobolev spaces are bounded by powers of , and is the projection onto the kernel of . The powers of do no harm, because they multiply terms which are exponentially small. For example, if and denotes the operator , then the gauge-invariance of gives , and so
where the last term has norm at most , by (14). With Lemma 5.5, this shows that has a right inverse whose image lies in the kernel of and whose norm, as an operator from to , is at most .
The thickened moduli space. For in , consider the equation
for a one-form in the image of and an element of . Its linearization at is the map , which is invertible, with an inverse of norm at most : this follows from the properties of and from Lemma 5.5. The left-hand side of (18) is of size at , and the non-linear terms are quadratic, with constants bounded by powers of . So by the contraction mapping theorem, there is a such that, for large , the equation has a unique solution with
The solution depends smoothly on .
The map embeds in , and up to an exponentially small error, the image of is a complement in the Coulomb slice at to the tangent space of the embedded surface. So the connections , with in the image of and , represent the points of an open neighborhood of the surface, each point once. This is an application of the slice theorem, in a form in which the size of the neighborhood is controlled by the lowest eigenvalue of , for which we have the bound (16); compare [23 Section 6] and [6 Theorem 1.2]. We define and the map of Proposition 4.1 by pulling back and the maps by the projection . Then is the set of points . It is a copy of , it is cut out transversely, and the section is .
*The moduli space is contained in We must show that, when is large, every element of is represented by a connection , with in the image of and satisfying (19). For then is a solution of (18), and it coincides with by uniqueness. Hypothesis (iii) does not give this directly, because the radius in (19) shrinks with .
Let be elements of , with , which converge to and as in hypothesis (iii). Since no action is lost, is close to on the neck; and since is non-degenerate, the difference , in a suitable gauge, decays exponentially from both ends of the neck [4 Chapter 4]:
Cutting at and extending the two halves by , we obtain connections on and at which is of size in the weighted norms. The linearized operator at is injective with closed range, because , and the points of the compact space are regular. So these two connections lie within distance of and of a solution representing a point of . It follows that, after a gauge transformation, is within distance of , and this is much smaller than when is large.
The additional perturbation. If is replaced by , where the norm of is at most and is large, then (14) still holds and the argument is unchanged. For a residual set of such , the moduli space is regular: perturbations supported in the complement of the neck are sufficient for this, because an element of the cokernel which vanishes there is zero, by unique continuation. This completes the proof of Theorem 5.3.
Proof of Proposition 4.1
In the situation of § 4, the stabilizer of is a circle, and hypotheses (i), (ii) and (iii) are the three parts of Proposition 3.3. So Theorem 5.3 applies. It gives part (a) of Proposition 4.1 and the inclusion of in . For a generic choice of the additional perturbation, the moduli space is regular, and its points are then the transverse zeros of the section . For a generic choice it is also the case that there are no reducible solutions in , and that the moduli spaces for the bundles with smaller action, whose formal dimensions are or less, are empty. The moduli space is then compact, and the number of its points, counted with signs, is the Donaldson invariant (6), which is odd.
The cases , and
Let be 5, 6 or 7, and write for . This manifold is the quotient of by the binary dihedral group
which has order . The group is when is odd and when is even. In this section we prove the following.
Proposition 6.1. Let be 5, 6 or 7, and let be a negative-definite submanifold of a surface , with oriented boundary and with . Then the intersection form of is not isomorphic to , where
The proof follows § 4–§ 6 closely, and we shall concentrate on the points at which it differs. Suppose then that , where is isomorphic to and .
The two pieces
The relation (2) holds for and for , with on the left. If were non-zero, it would have order 2 and would be unimodular. But is even, and its signature is , which is not divisible by 8. So . As in § 4, we then have
Because the first homology groups of and are zero, the lattices and are primitive sublattices of , each the orthogonal complement of the other, and the restriction map is surjective.
The bundle
We identify with the dual lattice , so that the square of a class in is , where denotes the norm in . We choose a class in as follows. When , it is a generator of . When , it is , where and are the basis of dual to the basis of roots. When , we use the standard model of as the lattice of vectors in with coordinate sum zero, and we take , where
In each case , and
This is clear when is 5 or 6. When , the vectors of norm in are the eight vectors , and for the vectors and have norms and , so they do not belong to .
Lemma 6.2. There is a class in with and .
Proof. Let . It is an even, indefinite lattice of rank , and its discriminant form is the negative of that of . When is odd, the discriminant group is cyclic of order 4, and the discriminant form takes the value or on a generator. When , it takes the values , and 1 on the three non-zero elements. In neither case is there a non-zero isotropic element. By Meyer’s theorem, contains a primitive vector with . Let be the positive generator of the ideal in . Then represents an isotropic element of the discriminant group, which is non-zero if . So , and there is a vector in with . After adding a multiple of to , we may suppose that , so that and span a hyperbolic plane in .
Now let be any class in which restricts to . If we identify with , then the classes which restrict to zero on are the elements of . So we may subtract a combination of and from to make it orthogonal to both, without changing its restriction to . The square of is even, and the class
has the required properties. □
Let be the adjoint bundle of the bundle with and . As in § 4, we have , the moduli space has formal dimension zero, and the Donaldson invariant is .
Flat connections
The group is the group of characters of , which has four elements. Let be the character with
and let be the character which corresponds to the restriction of to . When is odd, the class generates , so has order four and . When , the character is not trivial and is not , because its self-linking number is non-zero, whereas corresponds to the non-zero element of which has self-linking number zero. In all three cases, therefore,
and the four characters of are , , and .
Lemma 6.3. (a) Up to conjugacy, there are two homomorphisms with determinant . They are
(b) The rho-invariants of the corresponding flat connections are
Proof. The irreducible two-dimensional representations of are the representations , for , given by
See [9 Appendix A.2]. The determinant of is 1 or , according as is odd or even, and is never . A reducible representation with determinant is the sum of two characters whose product is , and there are two such pairs. This proves (a).
For (b), we use the formula for the rho-invariant from § 4.4, with in place of . For both and , the character of the adjoint representation is . So the contribution of is zero when is even, and when is odd it is . Each of the elements has , and the sum of over these elements is . Since
we obtain .
The flat connections and have the same adjoint representation, but they are not equivalent under the gauge group of determinant-one automorphisms, and we shall see that they are distinguished by the action. Both are reducible and non-degenerate. Their stabilizers in are circles, because the characters and are distinct. (When , the holonomy group of has order two, and Remarks 3.2 and 5.4 apply. When is odd, the holonomy group contains a rotation of order four, and its centralizer in is already a circle.)
The moduli spaces
Since and are both reducible, the residues of the action can be found from reducible connections on . A decomposition of as is determined by the class , which lies in . For each such there is a reducible anti-self-dual connection on , unique up to gauge; the classes and give the same connection. Its action is , and its limit on the end is the sum of the two characters which are the restrictions to of and . Let
For , the action is and the limit is . By (22), this is the only reducible connection with action , and we call it . Reducible connections with limit are obtained by taking as in the following table, in which .
| 5 | 6 | 7 | |
Table 3.
In each case, the classes restrict to and on , and the action is . So
From (7), (21) and Lemma 6.3, we obtain the following table, which takes the place of (9).
Table 24.
Lemma 5.1 applies to , with the reduction for which , and Lemma 5.2 applies to . We choose the perturbations so small that all solutions have , which by (23) implies that . The three parts of Proposition 3.3 then hold, with the following changes. In (a), the moduli space is . In (b), the moduli spaces on in the second and third rows of (24) are empty, and the compact two-manifold is
The proofs are the same. For the compactness of and for part (c), we note that an anti-self-dual connection on which runs between and has action at least , while a bubble, or a non-trivial anti-self-dual connection on the cylinder with the same limit at both ends, has action at least 1. So the only moduli spaces on which could arise from weak limits are the ones in the table.
The hypotheses of Theorem 5.3 are therefore satisfied, and Proposition 4.1 holds in the present situation. The argument of § 5 then gives a contradiction, as before, and Proposition 6.1 is proved.
The remaining cases
In this section we complete the proof of Theorem 1.1. Throughout, is a connected subdiagram of the Dynkin diagram , and is a submanifold of a K3 surface which satisfies the hypotheses of the theorem: it is negative definite, its oriented boundary is , and . We write for the closure of the complement of in .
Proposition 7.1. The group is free, and is zero. The intersection form either is isomorphic to , or is one of the four forms listed in (1). The group is zero, except possibly when , in which case it is either zero or .
Proof. The complement is connected, and both and are spin. As in the proof of Proposition 2.2, for or the group is free, the group is finite, and
The boundary map in the Mayer–Vietoris sequence is an isomorphism, so and are zero. From (24) we see that is an even, positive-definite lattice whose determinant is . Let be its rank.
Correction terms. The structures on are indexed by the elements of , in such a way that the spin structures correspond to the elements with , and the formula of [19 Corollary 1.5] for the boundary of a negative-definite plumbing gives
From the minimal norms of the cosets of the root lattices in their duals, we obtain the correction terms of the spin structures:
| , spin | |
| , even | |
| , odd | , |
| , odd | , |
| , even | , , , |
| , | |
Table 27.
(For the four cosets have minimal norms , , and , and the last two have order two only when is even. For the coset corresponding to has minimal norm , and it has order two when .)
Let be the spin structure which inherits from . Then
The inequality is [18 Theorem 9.6], and the congruence follows from the grading shift of the map which the punctured manifold induces on . If , then is a rational homology ball and .
The diagrams , and . From (27) and (26), the rank is for , it is or for , and it is or for . In the last case, is excluded because the correction term is not zero, and is an even unimodular lattice of rank , so it is isomorphic to . For , the discriminant form of is the negative of the discriminant form of . So has an even unimodular overlattice, which is isomorphic to , and is the orthogonal complement of a sublattice in , which is . The case of was treated in Proposition 2.2.
The diagrams . The determinant of is , by (24) and the hypothesis that is zero, so is not zero. (For and , the correction terms alone do not exclude .) From (27) and (26), we then have for , and
for , and respectively. An even positive-definite lattice of determinant is if its rank is , and is if its rank is . When , the discriminant form of is determined by the linking form of and the value of modulo , through Milgram’s formula, and it is the negative of the discriminant form of the lattice in the following table.
Table 1.
So has an even unimodular overlattice, isomorphic to , in which is the orthogonal complement of . The Weyl group of acts transitively on the sublattices of of each of the types , and , on the pairs of orthogonal roots, and on the vectors of norm . So the complement can be computed in the standard model, in which consists of the vectors in with even. A sublattice in the first coordinates, where , has complement in the remaining coordinates. The orthogonal roots and have complement , and the vector has complement . This gives the last row of the table.
The diagrams . When is odd, the spin structure with correction term cannot be the one inherited from . So , and from (26) and the fact that we obtain . To identify the lattice, let be the disk bundle over the two-sphere with Euler number . Its boundary is , so we can form the closed manifold . It is negative definite and has , so by Donaldson’s theorem its intersection form is diagonalizable. Changing the sign, we identify with the standard lattice . Let be the class of the zero section of , so that . The vector is primitive, because the quotient of by is a subgroup of , which is free. The lattice is contained in the orthogonal complement of , and it is equal to it because the two have the same rank and determinant.
Because is primitive, the reduction of modulo 2 is the kernel of the linear form on . Because is even, the linear form , which is not zero, vanishes on this kernel. So the two linear forms are equal, which means that every coordinate of is odd. Since , every coordinate is , and is isomorphic to .
The complement. By (25), the group is zero when is square-free. When , which is the case for and for , the group could only have order 2, and would then be even and unimodular. But the signature of is , with , and this is not divisible by 8. In the remaining case, , the relation (25) allows to have order 1 or 2.
Proof of Theorem 1.1. By Proposition 7.1, either is isomorphic to or we are in one of the four cases of [1]. The first of these is excluded by Theorem 2.1, in which is the piece called , and the other three are excluded by Proposition 6.1.
For and , as for , the hypotheses of Theorem 1.1 can be removed.
Theorem 7.2. Let be 5 or 7, and suppose that is smoothly embedded in a surface . Then separates into two pieces, , where has oriented boundary and is negative definite. The groups and are zero, and the intersection form of is isomorphic to .
Proof. As in the proof of Proposition 2.2, the manifold is an -space, and one of the two pieces, , is negative definite. By (27), the correction terms of the two spin structures on are and . Both are positive, so the inequality of [18], Theorem 9.6 shows that the oriented boundary of is and that . By (25), the group has order 1 or 2. If the order were 2, then would be even and unimodular, and its rank would be divisible by 8. The rank is at most 7, so would be a rational homology ball. The correction term of the spin structure inherited from would then be zero, which it is not. So , and Theorem 1.1 shows that is isomorphic to . The group is zero by Proposition 7.1.
References
- [1]A. Daemi, T. Lidman and M. Miller Eismeier, 3-manifolds without any embedding in symplectic 4-manifolds, Geom. Topol. 28 (2024), 3357–3372.DOI
- [2]A. Daemi and M. Miller Eismeier, Instantons and rational homology spheres, preprint, arXiv:2210.14071. (Section numbers refer to version 2.)arxiv.org/abs/2210.14071
- [3]S. K. Donaldson, Connections, cohomology and the intersection forms of 4-manifolds, J. Differential Geom. 24 (1986), 275–341.DOI
- [4]S. K. Donaldson, Floer homology groups in Yang–Mills theory, Cambridge Tracts in Mathematics, vol. 147, Cambridge University Press, 2002.DOI
- [5]S. K. Donaldson and P. B. Kronheimer, The geometry of four-manifolds, Oxford Mathematical Monographs, Oxford University Press, 1990.
- [6]P. M. N. Feehan, Critical-exponent Sobolev norms and the slice theorem for the quotient space of connections, Pacific J. Math. 200 (2001), 71–118.
- [7]P. M. N. Feehan and T. G. Leness, The SO(3) monopole cobordism and superconformal simple type, Advances in Mathematics (2019), article 106817; arXiv:1408.5307, version 5. arxiv.org/abs/1408.5307. See Equation (1.2), Theorem 1.1, Hypothesis 2.5, and Remark 2.8.
- [8]V. Gritsenko, K. Hulek and G. K. Sankaran, Abelianisation of orthogonal groups and the fundamental group of modular varieties, J. Algebra 322 (2009), 463–478.arxiv.org/abs/0810.1614
- [9]G. O. Helle, Equivariant instanton Floer homology and calculations for the binary polyhedral spaces, preprint, arXiv:2203.09471.arxiv.org/abs/2203.09471
- [10]P. B. Kronheimer, Four-manifold invariants from higher-rank bundles, J. Differential Geom. 70 (2005), 59–112.DOI
- [11]P. B. Kronheimer and T. S. Mrowka, Embedded surfaces and the structure of Donaldson’s polynomial invariants, J. Differential Geom. 41 (1995), 573–734.DOI
- [12]M. Mariño, G. Moore, and G. Peradze, Four-manifold geography and superconformal symmetry, Mathematical Research Letters 6 (1999), 429–437; arXiv:math/9812042. arXiv.org/abs/math/9812042.DOI
- [13]M. Miller Eismeier, Equivariant instanton homology, preprint, arXiv:1907.01091. (Section numbers refer to version 3.)
- [14]J. Milnor and D. Husemoller, Symmetric bilinear forms, Ergebnisse der Mathematik und ihrer Grenzgebiete, vol. 73, Springer, 1973.DOI
- [15]L. Moser, Elementary surgery along a torus knot, Pacific J. Math. 38 (1971), 737–745.DOI
- [16]A. Mukherjee, A note on embeddings of 3-manifolds in symplectic 4-manifolds, preprint, arXiv:2010.03681.
- [17]L. I. Nicolaescu, On the Cappell–Lee–Miller gluing theorem, Pacific J. Math. 206 (2002), 159–185.DOI
- [18]P. Ozsváth and Z. Szabó, Absolutely graded Floer homologies and intersection forms for four-manifolds with boundary, Adv. Math. 173 (2003), 179–261.DOI
- [19]P. Ozsváth and Z. Szabó, On the Floer homology of plumbed three-manifolds, Geom. Topol. 7 (2003), 185–224.DOI
- [20]J.-P. Serre, A course in arithmetic, Graduate Texts in Mathematics, vol. 7, Springer, 1973.DOI
- [21]C. H. Taubes, Path-connected Yang–Mills moduli spaces, J. Differential Geom. 19 (1984), 337–392.DOI
- [22]C. H. Taubes, Self-dual connections on 4-manifolds with indefinite intersection matrix, J. Differential Geom. 19 (1984), 517–560.DOI
- [23]C. H. Taubes, A framework for Morse theory for the Yang–Mills functional, Invent. Math. 94 (1988), 327–402.DOI