Branched coverings of \(\mathbb{C}\mathbb{P}^2\) and invariants of symplectic 4-manifolds (Q5926386)

\textit{S. K. Donaldson}'s remarkable recent work [Doc. Math., J. DMV, Extra Vol. ICM Berlin 1998, vol. II, 309-314 (1998; Zbl 0909.53018)] shows that every symplectic closed 4-manifold admits a Lefschetz pencil. By a result of Gompf every finitely presented group can be realized as the fundamental group of a symplectic 4-manifold. The work of \textit{J. Amorós, F. Bogomolov, L. Katzarkov} and \textit{T. Pantev} [Symplectic Lefschetz fibrations with arbitrary fundamental groups, J. Differ. Geom. 54, 489-545 (2000)] gives a purely topological construction of symplectic Lefschetz fibrations on symplectic 4-manifolds. Using Donaldson's approximately holomorphic sections in the above mentioned paper, \textit{D. Auroux} [Symplectic 4-manifolds as branched coverings of \(\mathbb{C}\mathbb{P}^2\), Invent. Math. 139, No. 3, 551-602 (2000)] gave another description of a symplectic 4-manifold as an approximately holomorphic branched covering of \(\mathbb{C}\mathbb{P}^2\). The symplectic 4-manifold \((X, \omega)\) with \(\frac{1}{2\pi i} [\omega] \in H^2(X, Z)\) admits a line bundle \(L\) with \(c_1(L) = \frac{1}{2\pi i} [\omega]\) and Hermitian metric, connection, and for \(k\) large enough, the bundle \(L^k \to X\) admits approximately holomorphic sections. The main result in Auroux's above paper is that there are three chosen approximately holomorphic sections which determine \(X\) as an approximately holomorphic branched covering of \(\mathbb{C}\mathbb{P}^2\). The branched locus \(D\) is a symplectic and approximately holomorphic singular curve in \(\mathbb{C}\mathbb{P}^2\) with finitely many singular points consisting of double points, tangential points and cuspidal points. Using Moishezon and Teicher's braid monodromy techniques, the branched curve \(D\) is everywhere transverse to the fibers of \(\pi: \mathbb{C}\mathbb{P}^2 \setminus [0:0:1]\) except for finitely many non-degenerately tangential points with \([0:0:1]\) a generic point, the cuspidal points are not tangent to the fiber of \(\pi\), \(D\) is transverse to itself, all singular points are distinct and lie in different fibers of the projection \(\pi\) and the 1-parameter of the branched curved \(D_t\) only has creation or cancellation of a pair of transverse double points with opposite orientations. Define such a branched cover as a quasiholomorphic covering if the branched locus \(D\) has the above property. The braid monodromy describing \(D\) is given by a braid factorization of the central element \(\Delta_d^2\) of the braid group \(B_d\). There are four types of factorizations of \(\Delta_d^2\) which are (1) holomorphic (from projection of algebraic surface to \(\mathbb{C}\mathbb{P}^2\)), (2) geometric (from the permuting factors in the braid factorization as Hurwitz equivalence, and all factors are simultaneously conjugated by some braid element as global conjugation), (3) cuspidal (from the degree 1, 2, 3 for factors in \(\Delta_d^2\)), and (4) cuspidal negative (from \(-2\), 1, 2, 3 for factors in \(\Delta_d^2\)). A geometric monodromy representation is a surjective group homomorphism \(\theta: F_d \to S_n\) such that \(\theta (\gamma_i)\) is transposition, \(\theta (\gamma_1, \dots,\gamma_d)=1\), acts in the same manner for degree 1 factor, \(\theta(\gamma_i)\) acts distinctly and commutes for degree \(\pm 2\) factors and does not commute for degree 3 factors, where \(F_d\) is the free group of \(d\) generators and \(S_n\) is the symmetric group of \(n\) letters where \(n\) is the number of branched covering sheets. Two pairs \((F_i, \theta_i)\) (\(i=1, 2\)) of factorizations and geometric monodromy representations are \(m\)-equivalent if there is a finite sequence of type (2) and a creation or cancellation to turn one into another. The main results of the paper under review are (i) Theorem 1 shows that every compact symplectic 4-manifold is a quasiholomorphic branched covering of \(\mathbb{C}\mathbb{P}^2\), and it is unique up to isotopy for \(k\) large enough in choosing the branched loci by Theorem 5 in the paper. (ii) Theorem 2 shows that every symplectic 4-manifold with \(\frac{1}{2\pi i}[\omega]\in H^2(X, Z)\) is uniquely characterized by the finite sequence of type (2) and type (4) to the quasiholomorphic branched coverings of \(\mathbb{C}\mathbb{P}^2\) canonically up to \(m\)-equivalence. Conversely, for any quasiholomorphic curve \(D\) and a geometric monodromy representation, there are type (4) factorizations corresponding to \(D\) canonically up to smooth isotopy, a symplectic 4-manifold \(X\) quasiholomorphic branched covering of \(\mathbb{C}\mathbb{P}^2\) with branched locus \(D\) and the symplectic structure on \(X\) is canonical up to symplectomorphism and depends only on the smooth isotopy class of the curve \(D\) by Theorem 3. Theorem 1 can be thought of as a refinement of a previous result of the first author. Theorem 2 and Theorem 3 give a complete classification of symplectic 4-manifolds in terms of algebraic classification of \(m\)-equivalent words in the braid group among the geometric monodromy representations. Furthermore, the authors prove that the quasiholomorphic coverings determine Lefschetz pencils. The methods in the paper are from Donaldson's approximately holomorphic curves and from D. Auroux's previous work as well as Moishezon and Teicher's braid monodromy techniques. A purely topological version is given in section 5. Due to the restrictions of the techniques used in the paper, the authors pose several conjectures (to improve their results) and questions (to their explicit examples).