Geometry VI. Riemannian geometry. Transl. from the Russian by S. A. Vakhrameev (Q5928120)

It is a daunting task for any author nowadays to write an introductory textbook on Riemannian Geometry, whether destined for graduate students or for professional mathematicians. First, because Riemannian geometry has developed in so many different directions, it is virtually impossible to give due attention to every single topic. So, every author needs to select carefully those themes that he wants to pursue in depth while still conveying a sense of the vastness of the field as a whole. Second, there is already a whole arsenal of useful and worthwhile books on Riemannian geometry, both on the introductory level and as reference works. So, a new publication should bring something new to an already pampered audience, be it in its choice of material, its innovative approach or its original or refreshing style. The most striking feature of the present introductory book on Riemannian geometry is that the first third of the book is not about Riemannian geometry as such, but about the more general theory of manifolds with a linear connection. This sets it apart from most other works in the field, but in the reviewer's opinion, this is also one of the strengths of the book. Indeed, in this way a clear distinction is made possible between those notions and theorems which belong strictly to Riemannian geometry and those which are valid in a more general framework, a distinction which may be blurred if one starts in the Riemannian context straightaway. Starting from an arbitrary linear connection on a manifold, the author introduces geodesics, the exponential mapping, affine fields and various curvatures, all of which will later be specified in the presence of a Riemannian metric. The highlight of this part of the book is a full proof of Cartan's theorem on symmetric spaces, stating that any globally symmetric space is isomorphic to the quotient \({\mathcal G}/{\mathcal H}\) of a Lie group~\(\mathcal G\) by a subgroup~\(\mathcal H\) satisfying \(\text{Fix}(\sigma)_e\subset{\mathcal H}\subset \text{Fix}(\sigma)\) for some involutive automorphism~\(\sigma\) of~\(\mathcal G\). Also the equivalence with symmetric spaces in the sense of Loos is proved. Only afterwards does the author introduce a Riemannian metric. Metric properties of geodesics are considered, including the characterization of geodesics as `locally shortest curves'. The Hopf-Rinow theorem on the equivalence of geodesic completeness and metric completeness is proved. The study of surfaces (both abstract ones and those immersed in Euclidean three-space) recurs a few times: the Gauss-Bonnet theorem is proved (a higher-dimensional extension requiring characteristic classes is given without proof); some comments are given on minimal surfaces as well as a classification of surfaces with a conformal structure. For general submanifolds, the Gauss-, Codazzi- and Ricci-relations are derived. For hypersurfaces in (pseudo-)Euclidean space, the uniqueness (up to congruence) of the embedding for a given second fundamental form (as well as existence of such embeddings) is proved. Two applications of this result are given: the rigidity of hypersurfaces of the Euclidean space with at least three non-zero principal curvatures at each point and Riemann's theorem which states that spaces of constant curvature are locally isometric to the corresponding model spaces (realised as hypersurfaces in (pseudo-)Euclidean space). Next, the stronger Cartan-Killing theorem giving a classification of complete connected spaces of constant curvature is proved. After some chapters on the curvature and the topology of four-dimensional (Einstein) manifolds and on metric Lie groups, the study of conjugate points, Jacobi vector fields and cut loci leads to such important and well-known theorems as the Myers theorem and the Cartan-Hadamard theorem relating curvature and topology, and the Bochner theorem on the finiteness of the isometry group of compact Riemannian manifolds with negative definite Ricci tensor. This material is complemented with an appendix of several chapters on differentiable manifolds and the basics of vector bundle theory (without proofs), making the book essentially self-contained. Throughout the book, the author clearly prefers the coordinate approach even though the invariant approach gets some attention too. In the reviewer's opinion, a major shortcoming is the complete lack of references in the text itself. Also the bibliography is limited to a list of 13 recommended textbooks that admittedly are excellent. This may be due to the origin of the book as classroom lectures, but an additional effort here would have greatly enhanced the book's usefulness as a reference work. Another (minor) negative point is the high number of misprints, some of which are quite confusing for the reader. A special note concerning the many exercises scattered throughout the text is in order too. These are not so much applications or extensions of the theory as in other introductory texts, but they form an essential part of the theoretical content itself. Almost invariably, they complete proofs or they are used in an essential way later in the text. This may make it hard for students working through the book by themselves, as no solutions and few hints are provided. Actually, what is true for the exercises, holds for the book itself: it is structured in such a way that most, if not all, of the results are building blocks for one of the major theorems later on. There are hardly any loose ends. In this way, M.~M. Postnikov has written a well-structured and readable book with a satisfying sense of completeness to it. The reviewer believes this book deserves a place next to the already existing literature on Riemannian geometry, principally as a basis for teaching a course on abstract Riemannian geometry (after an introduction to differentiable manifolds) but also as a reference work.