Geometry of minimal energy Yang-Mills connections (Q531756)

Let \(G\) be a compact Lie group and \(E\) a principal \(G\)-bundle on a complete oriented Riemannian manifold \(M\). A Yang-Mills connection \(A\) is a critical point of the Yang-Mills functional \(\| F_A \|^2\). On compact oriented 4-manifolds an instanton (self-dual or anti-self-dual connection) is always an absolute minimizer of the Yang-Mills functional. This leads to the converse question: In four dimensions, are local minima for the Yang-Mills energy necessarily direct sums of instantons and abelian connections? Compare this question with the papers by \textit{L. Sadun} and \textit{J. Segert} [Bull. Am. Math. Soc., New Ser. 24, No.~1, 163--170 (1991; Zbl 0722.58014)] and by \textit{L. M. Sibner, R. J. Sibner} and \textit{K. Uhlenbeck} [Proc. Natl. Acad. Sci. USA 86, No.~22, 8610--8613 (1989; Zbl 0731.53031)]. In the case of partial positive results, see for example the paper by \textit{J.-P. Bourguignon, H. B. Lawson} and \textit{J. Simons} [Proc. Natl. Acad. Sci. USA 76, 1550--1553 (1979; Zbl 0408.53023)]. The main result is: Energy-minimizing Yang-Mills connections on compact homogeneous 4-manifolds are either instantons or split into sums of instantons when passing over to the adjoint bundle ad\((E)\). Two related results are given: On a Calabi Yau 3-fold \(M\), if a connection \(A\) is a stable critical point of the \((0,2)\)-part of the energy functional \(\| F_A^{0,2} \| ^2\), and in addition the \((0,2)\)-part of the curvature is in \(L^4\) for noncompact \(M\), then the \((0,2)\)-part of the curvature is covariant constant and takes values in an abelian subundle of ad\((E) \otimes \mathbb{C}\). If the holonomy of \(M\) is \(SU(3)\), then \(E\) is holomorphic. Finally, for a bundle \(E\) on a complete 3-dimensional manifold with nonnegative Ricci curvature, let \(A\) be a stable finite energy Yang-Mills connection. If \(M\) is noncompact and \(A\) is in \(L^4\), then \(\nabla_A F_A =0\). Moreover, the curvature takes values in a flat subbundle of ad\((E)\), and \(F_A=0\) unless \(M\) has local flat factors. Compare this with Chapter II, Corollary 2.3 in the paper by \textit{A. Jaffe} and \textit{C. Taubes} [Vortices and monopoles. Structure of static gauge theories. Progress in Physics, 2. Boston-Basel-Stuttgart: Birkhäuser (1980; Zbl 0457.53034)].