Dissolving 4-manifolds, covering spaces and Yamabe invariant (Q2017595)

Following a definition due to \textit{R. E. Gompf} [J. Differ. Geom. 34, No. 1, 93--114 (1991; Zbl 0751.14024)] a simply connected 4-manifolds \textit{dissolves} if it is diffeomorphic to either \(k\mathbb{C}\mathbb{P}^2\# \overline{\mathbb{C}\mathbb{P}^2}\) or \(k(K^3)\#\ell(S^2\times S^2)\), for some nonnegative integers \(k\) and \(\ell\), where \(\mathbb{C}\mathbb{P}^2\) denotes the complex projective plane and \(\overline{\mathbb{C}\mathbb{P}^2}\) denotes the underlying smooth 4-manifold \(\mathbb{C}\mathbb{P}^2\) equipped with the opposite orientation and \(K^3\) denotes a simply connected smooth complex surface with trivial canonical bundle. A famous result of \textit{C. T. C. Wall} [J. Lond. Math. Soc. 39, 141--149 (1964; Zbl 0131.20701)] affirms that any two simply connected oriented 4-manifolds with isomorphic intersections forms become diffeomorphic after some number of connected sums with \(S^2\times S^2\). This result together with Freedman's classification theorem [\textit{M. H. Freedman}, J. Differ. Geom. 17, 357--453 (1982; Zbl 0528.57011)] implies that for any simply connected 4-manifold \(M\), there exists an integer \(m\) such that the connected sum \(M\#m(S^2\times S^2)\) dissolves. The first goal of this paper is to construct new infinite families of spin and nonspin 4-manifolds with zero signature that dissolve after a connected sum with only one copy of \(S^2\times S^2\). In order to enunciate the main results in this direction the following notions and notations are necessary. By a paper of two of the authors [\textit{A. Akhmedov} and \textit{B. D. Park}, Math. Res. Lett. 17, No. 3, 483--492 (2010; Zbl 1275.57039)], for every integer \(k\geq 138\), there exists an infinite family of mutually nondiffeomorphic irreducible 4-manifolds \(\{Z(k)_n\mid n\in \mathbb{Z}\}\) that are all homeomorphic to \((2k-1)(S^2\times S^2)\). For a fixed integer \(k\geq 138\), \(Z(k)_n\) is symplectic if \(n\geq 0\) and nonsymplectic if \(n<0\). The \textit{Wall number} of \(M_1\) and \(M_2\), denoted by \(w(M_1,M_2)\), is the smallest non-negative integer \(m\) such that \(M_1\#m(S^2\times S^2)\approx M_2\# m(S^2\times S^2)\). \[ w_{k,n}:=w(Z(k)_n,(2k-1)(S^2\times S^2))\geq 1. \] For any integer \(m\geq 1\), \(Y_{k,n}(m)\) denotes the generalized fiber sum of \(m\) copies of \(Z(k)_n\) along a torus \(T\). \[ Y_{k,n}:=\overset{m}{\overbrace{Z(k)_n\#_T Z(k)_n\#_T\dots\#_T Z(k)_n}} \] Theorem 4. Let \(k\geq 138,m\geq 1\), and \(n\) be integers. Then \(Y_{k,n}(m)\) is a closed simply connected (symplectic if \(n\geq 0\)) spin 4-manifold with zero signature. \(Y_{k,n}(w_{k,n})\) dissolves after a connected sum with only one copy of \(S^2\times S^2\). More precisely, we have \[ Y_{k,n}(w_{k,n})\# (S^2\times S^2)\approx 2k w_{k,n}(S^2\times S^2). \] The first and last authors [\textit{A. Akhmedov} and \textit{B. D. Park}, J. Gökova Geom. Topol. GGT 2, 1--13 (2008; Zbl 1184.57017)] showed that for every integer \(q\geq 25\), there exists an irreductible symplectic 4-manifold \(M(q)\) that is homeomorphic but not diffeomorphic to \((2k-1)(\mathbb{C}\mathbb{P}^2\#\overline{\mathbb{C}\mathbb{P}^2})\). The Wall number of \(M(q)\) and \((2k-1)(\mathbb{C}\mathbb{P}^2\#\overline{\mathbb{C}\mathbb{P}^2})\) is denoted by \(w_q\geq 1\) and let \(\delta^q_{k,n}:max\{w_{k,n},w_q\}\geq 1\) and \(X(\delta^q_{k,n}):=M(q)\#_{T'}Y_{k,n}(\delta^q_{k,n})\), where the torus \(T'\) in \(M(q)\) is identified with a parallel copy of \(T\) in one of the \(Z(k)_n\) summands of \(Y_{k,n}(\delta^q_{k,n})\). Theorem 7. For any triple of integers \(q\geq 25,k\geq 138\) and \(n\geq 0,X(\delta^q_{k,n})\) is a closed simply connected symplectic irreducible nonspin 4-manifold with zero signature, and dissolves after a connected sum with only one copy of \(S^2\times S^2\), More precisely, we have \[ X(\delta^q_{k,n})\#(S^2\times S^2)\approx 2(q+k\delta^q_{k,n})(\mathbb{C}\mathbb{P}^2\#\overline{\mathbb{C}\mathbb{P}^2}). \] In the second part of the paper the authors give new counterexamples to a conjecture of \textit{J. Rosenberg} [Topology 25, 319--336 (1986; Zbl 0605.53020)] in dimension four using the above mentioned new families of dissolving 4-manifolds. \textbf{Rosenberg Conjecture}. Let \(M^n\) be a connected closed manifold of dimension \(n\) and whose fundamental group is finite of odd order. Then \(M^n\) admits a metric of positive curvature if and only if its universal cover does. In the case when \(n\geq 5\), Rosenberg's conjecture has been confirmed [Rosenberg, loc. cit.; \textit{S. Kwasik} and \textit{R. Schultz}, Comment. Math. Helv. 65, No. 2, 271--286 (1990; Zbl 0717.53026)], but, as the authors of the present papers affirm, no counterexample is known. In this sense the authors use their manifolds to construct new examples of 4-manifolds with negative Yamabe invariant and whose universal covers have positive Yamabe invariant. In particular, these provide new spin and nonspin counterexamples to Rosenberg's conjecture in the case of dimension four.