Splitting deformations of degenerations of complex curves. Towards the classification of atoms of degenerations. III. (Q2498062)

This is a 590 pages book on deformation theory, using mostly topological methods, but also ``translated'' to algebraic geometry and using algebraic methods. The book contains all computational details, which it has to, so that it is not as lengthy as it may look. The book makes a systematic study of degenerations of complex curves, and their splitting deformations. The principal aim is to develop a deformation theory of degenerations of complex curves. The construction of these deformations uses special subdivisors of singular fibers such that these subdivisors are barked (peeled) off from the singular fiber. The construction of these barking deformations is related to deformations of singularities as well as the mapping class groups of Riemann surfaces (complex curves) via monodromies. Moreover, the positions of the singularities of a singular fiber appearing in a barking deformation are described in terms of the zeros of a certain polynomial expressed in terms of the Riemann theta function and its derivative. The book provides several applications: Construction of interesting examples of splitting deformations leading to the class number problem of such and the complete classification of absolute atoms of genus 1 to 5. This study reveals the close relationship between the monodromy of a degeneration and existence of certain deformations of the degeneration. By using a lot of visualization diagrams, the author explains how a singular fiber is deformed. These deformations are called barking deformations because some special subdivisor of the singular fiber looks like barked (peeled) of. Barking deformations have a remarkable cross-disciplinary nature; they are related to algebraic geometry, low dimensional topology and singularity theory. Notice that the present exposition mainly discusses one-parameter deformation theory. A degeneration of complex curves of genus \(g\) is a one-parameter family of smooth complex curves, which degenerates to a singular complex curve, that is \[ \pi:M\rightarrow\Delta \] where \(M\) is a smooth complex surface, \(\Delta\) a small disk in \(\mathbb{C}\) containing the origin, \(\pi\) is good enough, \(\pi^{-1}(0)\) is singular and \(\pi^{-1}(z)\) is smooth of genus \(g\) when \(z\neq 0\). To classify fibered surfaces, it is important to understand their local structure - degeneration - around each singular fiber. It is also important to know when the signature \(\sigma(S)\) (or some other invariant) of the fibered surface concentrates one singular fibres. Such questions motivate the study of degenerations and their invariants. Another important invariant is monodromy which already appeared in the early study of degenerations, notably in the work of Kodaira on the classification of degenerations of elliptic curves. He showed that there are eight degenerations and determined their monodromies. Namikawa and Ueno carried out the classification of degenerations of complex curves of genus 2: There are about 120 degenerations, the topological type of a degeneration is not necessarily determined by its singular fiber, nor by its monodromy. This is because of the non-triviality of the Torelli group \(T_g\). So monodromy is not powerful enough to classify degenerations. In the converting process from a topological monodromy to a monodromy, some information may be lost. This is the starting point of work of Matsumoto and Montesinos explained in this text. In particular, the topological monodromy of a degeneration is a very special homeomorphism; it is either periodic or pseudo-periodic. A Dehn twist \(\gamma\) along a loop \(l\) on \(C\) is an example of a pseudo-periodic homeomorphism, and Dehn twists generate some power of every pseudo-periodic homeomorphism. These can then be classified as either right or left. Using the theory of Teichmüller spaces, Earle-Sipe and Shiga-Tanigawa demonstrated that any topological monodromy is a right pseudo-periodic homeomorphism. Let \(\mathcal{P}_g\) be the set of periodic and right pseudo-periodic homeomorphisms of a complex curve of genus \(g\), \(\widehat{\mathcal{P}}_g\) the conjugacy classes of \(\mathcal{P}_g\) (under the action of \(\pi_1\)). Let \(\mathcal{D}_g\) denote the set of degenerations of complex curves of genus \(g\), and let \(\widehat{\mathcal{D}}_g\) be the set of its topologically equivalent classes. The main result of Matsumoto and Montesinos is that the elements in \(\widehat{\mathcal{P}}_g\) are in one to one correspondence with the elements of \(\widehat{\mathcal{D}}_g\). This book mainly considers normally minimal degenerations: A degeneration \(\pi:M\rightarrow\Delta\) is called normally minimal if the singular fiber \(X\) satisfies (1) the reduced curve \(X_{\text{red}}:=\sum_i\Theta_i\) is a normal crossing, and (2) If \(\Theta_i\) is an exceptional curve, then \(\Theta_i\) intersects other irreducible components at at least three points. According to wether the topological monodromy is periodic or pseudo-periodic, the singular fiber \(X\) is stellar or constellar. \(X\) has a central irreducible component (core) and several chains of projective lines emanating from the core. Such a chain is called a branch of \(X\). A constellar singular fiber is obtained by bonding branches of a stellar fibers, and a resulting chain of projective lines after bonding is called a trunk. Thus the number of singular fibers of genus \(g\) grows rapidly as \(g\) grows higher, in fact there is an algorithm to carry out the topological classification of degenerations of given genus. This leads to classifying degenerations modulo deformations indicated by Morsification of singularities: When does an isolated singularity \(V\) admit a deformation \(\{V_t\}\) such that \(V_t\) for \(t\neq 0\) possesses only \(A_1\)-singularities? The book carefully explains Morsification of singular fibers. Let \(\Delta=\{s\in\mathbb{C}:| s| <\delta\},\) \(\Delta^\dag=\{t\in\mathbb{C}:| t| <\varepsilon\}.\) Let \(\mathcal{M}\) be a complex \(3\)-manifold. Then \(\Psi:\mathcal{M}\rightarrow\Delta\times\Delta^\dag\) is a deformation family of \(\pi_0:M_0\rightarrow\Delta_0=\Psi| _{\Psi^{-1}(\Delta\times\{0\})}:\Psi^{-1} (\Delta\times\{0\})\rightarrow\Delta\times\{0\}.\) If \(\pi_t\) for \(t\neq 0\) has at least two singular fibers, \(X_1,X_2,\dots,X_n\) then \(\Psi\) is called a splitting family of the degeneration \(\pi_0\), and \(X=\pi_{0}^{-1}(\Delta\times\{0\})\) is said to split into these. If a singular fiber \(X\) admits no splitting deformations, \(\pi:M\rightarrow\Delta\) is called atomic, and \(X\) is called an atomic fiber. A Morsification of a degeneration is a splitting family such that all fibers except the fiber in the origin are atomic. This is to strong, and the author considers a consept of stepwise Morsification. This finite-stage Morsification of a degeneration is useful for studying the topological types of fibered algebraic surfaces. The number of atomic generations of genus \(g\) must be much less than that of all degenerations of genus \(g\). This leads to classification of degenerations modulo deformations. The author explains the double covering method for constructing splitting families based on branched curves. This method cannot be used for degenerations of genus \(g\geq 3\), and this is the main reason for developing a new deformation theory, applicable to any degeneration. These are barking deformations obtained by barking off a special subdivisor of the singular fiber from the singular fiber. This way the author deduces criteria for the splittability of degenerations, provides interesting examples of splitting deformations which leads to the ``class number problem'' for degenerations and determines absolute atoms of genus 3,4 and 5. The class number problem is the question of how many topologically equivalent deformations a fixed splitting have. The classification of atomic degenerations is then explicitly given, linked closely to the theory of isolated singularities. In fact a weaker classification ``absolutely atomic'' is used. The main criteria for splittability is explicitly stated and explained, and finally, in the end of the book a list of representative crusts for a large class of singular fibers of genus from 1 to 5 is given, enough for the purpose of classifying absolute atoms. The book is explicit and contains all material needed to understand the subject. It is at a nice level and should be possible to read. Most commonly, algebraic geometers translate from differential geometry to solve problems. In this book the concept is vice versa: Algebraic methods are used to solve topological problems. Thus this book may at the first glance look elementary for an algebraist, but it is not.