Moduli spaces of PU(2)-monopoles (Q5929568)

Generic regularity of the moduli spaces of \(PU(2)\)-monopoles is proved and their Uhlenbeck compactifications are constructed. Generic regularity was claimed to be proved in \textit{V. Pidstrigach} and \textit{A. Tyurin}, Russ. Acad. Sci., Izv. Math. 40, 267-351 (1993; Zbl 0796.14024)]. But in this paper, the author gives counterexamples of key statements in their proof, and gives a rigorous proof. It uses perturbation and properties of Fredholm maps [\textit{S. Smale}, Am. J. Math. 87, 861-866 (1965; Zbl 0143.35301)]. On the other hand, as in the instanton case [\textit{S. Donaldson} and \textit{P. B. Kronheimer}, The geometry of four-manifolds, Oxford (1990; Zbl 0820.57002) hereafter refered to as [1]], compactification is done according to the scheme \[ \text{Local estimates} \to\text{Regularity}\to \text{Removable Singularity} \to\text{Compactification.} \] But since monopole equations are only scale invariant and not conformal invariant, new arguments are introduced in this paper. The outline of the paper is as follows: In Section 2, the double covering group \(\text{Spin}^{U(2)} =\text{Spin} \times_{\mathbb{Z}_2} U(2)\) of \(SO\times PU(2) \times S^1\) is defined. The covering map is decomposed as \((\pi,\overline \delta, \det)\). The maps \(ad_\pm: \text{Spin}^{U (2)}(4)\to so(su(2))\), \(\sigma_\pm: \text{Spin}^{U(2)} (4)\to U(\mathbb{H}_\pm \otimes_\mathbb{C} \mathbb{C}^2)\) and the Clifford map \(\gamma:P \times_{SO(4)} \mathbb{R}^4\to P^u\times_\pi \mathbb{R}^4\) are also defined. Here \(P\) is an \(SO(4)\)-bundle whose second Stiefel-Whitney class admits integral lifts and \(P^u\) is the associated \(\text{Spin}^{U(2)}\)-bundle. Fixing connections \(a\) of \(\det(P^u)\) and \(A\) of \(\overline\delta (P^u)\), the Dirac operator \(\mathbb{D}_{A,a}\) is defined to be \(\gamma\nabla_{B_{a.a}}\), where \(B_{A,a}\) is the connection of \(P^u\) determined by \(A\) and \(a\). Then the \(PU(2) \)-monopole equations \((SW^\sigma_a)\) are the equations for a pair \((A,\Psi)\) of a \(PU(2)\)-connection \(A\) and a positive spinor \(\Psi\) such that \[ \mathbb{D}_{A,a} \Psi=0, \quad \Gamma(F_A^+)= (\Psi\overline \Psi)_0. \] The moduli space of \(PU(2) \)-connections is denoted by \({\mathcal M}^\sigma_a\) (Section 2). After giving counterexamples of the key statement of Pidstrigach-Tyurin, generic regularity of \({\mathcal M}^\sigma_a\) is proved in Section 3. For this purpose, first the \(PU(2)\)-monopole equations are perturbed as follows; \[ \mathbb{D}^0_{a,A} (\Psi)+\beta (\Psi) =0,\quad \Gamma(F_A^+)= K(\Psi \overline\Psi)_0. \] Here \(K\) is a matrix valued function. The map defined by the left hand sides of these equations is called the Seiberg-Witten map. Then it is shown if the Seiberg-Witten map extended to the parametrized moduli space is not a submersion in a point \((A,\Psi, \beta, K)\), then the spinor component \(\Psi\) must degenerate (Theorem 3.1). Next, varying the \({\mathcal C}^k\) Clifford map \(\gamma\), geometric properties of the Seiberg-Witten map are analyzed, and the author derives the Fredholm property of the projection to the space of \({\mathcal C}^k\)-Clifford maps on a neighborhood of a non-abelian point on an enlarged moduli space \(\widetilde {\mathcal M}^*_X\) (Lemma 3.15). Here \((A,\Psi)\) is said to be non-abelian in \(x\) if the second fundamental form of the line bundle generated by \(\Psi\) around \(x\) is non-zero. Then, since Fredholm maps are locally proper, the set of parameters for which there exists a non-abelian solution with non-trivial degenerate spinor component is of the first category (Corollary 3.16). Since the moduli space \(\widetilde{{\mathcal D}{\mathcal M}}^*_X\) of solutions with non-trivial but degenerate spinor component in the enlarged moduli space \(\widetilde{\mathcal M}^*_X\) has infinite codimension in every non-abelian point (Theorem 3.7), generic parameters are obtained by intersecting the component of this set with the set of regular values of the projection of \(\widetilde {\mathcal M}^*_X\setminus\widetilde{{\mathcal D}{\mathcal M}}^*_X\) on the parameter space (Theorem 3.19). It is also remarked by using the same argument also gives generic regularity theorem for \(\text{Spin}^C\)-moduli spaces (Theorem 3.21). On the other hand, the author remarks generic regularity is not enough, because the Kählerian parameters are all non-generic in the author's sense. The Uhlenbeck compactifications of the perturbed moduli spaces of \(PU(2)\)-monopoles are constructed in Section 4. Let \({\mathfrak p}=(\gamma,C,a,\beta,K)\) be a system of data. Then it is called a standard system if it takes the form \((\gamma_0, C_0, 0,0,K_0)\), \(\gamma_0\) is the standard identification \(\Lambda^1_B\to B\times \mathbb{H}\), \(C_0\) the flat \(SO(4)\)-connection and \(K_0\) is a constant automorphism. Then, applying gauge-fixing (Theorem 4.6, cf. [1], Theorem 2.3.7), local estimates for data close to the standard data are given (Theorem 4.7). Then rewriting these estimates in terms of the curvature (Corollary 4.8), local and global compactness results are derived (Corollary 4.9 and 4.10, cf. [1], Proposition 4.4.9). Next, analysing the elliptic operator \(\mathbb{D}_{\mathfrak q}+d^*_g +\Gamma_\gamma \circ d\), regularity of \(L^4\)-small \(L^2_1\)-almost solutions with connection component in Coulomb gauge is proved (Proposition 4.14). The third step (removable singularities, Theorem 4.21) of the construction of compactification is achieved by applying gauge fixing on the sphere (Theorem 4.20. cf. [1], Proposition 2.3.13). Then the compactification of the moduli space is constructed by using the embedding the moduli space to the space of ideal monopoles (Theorem 4.24). Here an ideal monopole means the pair of an element in a symmetric power of \(X\) and a monopole (Definition 4.22).