Approximations and endomorphism algebras of modules. (Q2497141)

The book presents two complementary subjects of module theory by two leading experts: (1) Approximation theory which describes modules by approximating them with well understood classes of modules. The central concept is a cotorsion pair. (2) Realizing algebras as endomorphism algebras of modules, which shows that modules can have pathological and complex structure. Several related constructions are presented, which heavily rely on combinatorial set theoretical tools. A particular strong tool is the Black Box, which appears here in book form for the first time. The book requires only basic knowledge of algebra, homological algebra (exact sequences for Ext and Tor) and set theory (cardinal arithmetic, regular cardinals, closed unbounded sets). It develops the theories from scratch explaining and illustrating the concepts on the way. Complicated arguments are presented first in simpler forms, leading to weaker results but demonstrating clearly the main idea. This eases the way of understanding the whole argument later. Set theory is heavily used throughout the book, so the book is a good demonstration that minimal set theoretic knowledge can be a great help in algebra -- it even suffices for proving independence results. On the one hand, the book is well-suited for self-study at a graduate level, for making oneself familiar with the subject and for learning the techniques and tools to conduct one's own research. On the other hand, the book is also a survey: it shows the state-of-the-art of the subject by mentioning additional results. Below, a short, very incomplete summary of the main ideas of the book is presented. Chapter 1 revises the special modules playing important role during the book. Chapter 1.1 is essential for the second part of the book (realizing algebras), in particular the S-notions like S-reduced, S-complete, S-torsion-free, S-pure and S-cotorsion-free modules. Approximation theory makes the first half of the book (Chapters 2-8, 10 and 11). Chapters 2-4 recall the basic material on cotorsion pairs. Chapter 2 introduces the basic concepts of approximations. It first introduces the approximations: precovers and preenvelopes and their minimal versions, covers and envelopes. Then it introduces the notion of a cotorsion pair, which is a pair of classes of modules mutually determining each other. One of the classes is supposed to contain covers, the other class is supposed to contain envelopes. It appears to be no name for these classes, so in this review, these will be called as ``class for covers'' and ``class for envelopes''. The usefulness of a cotorsion pair for approximation is measured by the following notions. A cotorsion pair is `complete' if it contains a special preenvelope and a special precover for every module. A cotorsion pair is `perfect' if it contains covers and envelopes for every module. The most important result, the full proof of which is included, is Enochs' theorem (Corollary 2.3.7) stating that if a cotorsion pair is complete and closed (i.e., its class containing the precovers is closed under direct limits) than it is perfect. This is the fundamental way of proving the perfectness of a cotorsion pair. Chapter 3 provides useful criteria for completeness and perfectness of cotorsion pairs: (1) cotorsion pairs generated by a set are complete (Theorem 3.2.1), and (2) cotorsion pairs cogenerated by a class of pure-injective modules are perfect (Theorem 3.2.9). Chapter 4 provides the basic and historical examples of complete and perfect cotorsion pairs. It also includes decomposition properties of cotorsion pairs such as Kaplansky's theorem (Theorem 4.2.11): a cotorsion pair is generated by a set if and only if every module of its class for precovers has a filtration with small factors from the class. Section 4.3 contains a detailed proof that tilting (Corollary 4.3.11) and cotilting (Theorem 4.3.23(b)) cotorsion pairs are prefect. Chapter 5 is an introduction to the theory of tilting modules, including infinitely generated tilting modules. Tilting modules give rise to tilting cotorsion pairs. The main result (Theorem 5.2.20) is that tilting cotorsion pairs are of finite type. Chapter 6 specializes to \(1\)-tilting modules. It shows that the theory in Chapter 5 significantly simplifies. A representing set of \(1\)-tilting modules is provided for Prüfer, Dedekind and valuation domains. Finally, section 6.3 contains an interesting application (Theorem 6.3.16): if \(S\) is a multiplicative subset of non-zero divisors of a commutative ring \(R\), then the localization \(S^{-1}R\) has projective dimension at most \(1\) (Matlis localization) if and only if \(S^{-1}R/R\) is a direct sum of countably presented \(R\)-modules. Chapter 7 applies tilting theory to finitistic dimensions of rings. The link is provided by Theorem 7.1.10, which states that the little finitistic dimension is finite if and only if the cotorsion pair generated by the class of finitely generated modules of finite projective dimension is tilting. However, the tilting module for this cotorsion pair does not need to be finitely generated even for Artinian rings. Three applications are provided: (1) both the little and big finitistic dimension of \(n\)-Iwanaga-Gorenstein rings are \(n\) (Theorem 7.1.12); (2) the little finitistic dimension of an Artinian ring is the maximum projective dimension of a special precover (for the cotorsion pair generated by finitely generated modules of finite projective dimension) of all simple modules (Theorem 7.2.4); (3) The little and the big finitistic dimensions are equal for Artinian algebras for which the class of finitely generated modules of finite projective dimension are contravariantly finite (Theorem 7.3.8). Chapter 8 discusses cotilting theory, the dual of tilting theory. This chapter dualizes many results of Chapters 5 and 6. Another important result is that cotilting modules are pure-injective. The dual of a tilting module is always cotilting. Cotilting modules of this type are said to be of cofinite type. The \(1\)-cotilting cotorsion pairs over Dedekind domains are classified in Theorem 8.2.11. For Prüfer domains, only the \(1\)-cotilting cotorsion pairs of cofinite types are classified. An example of a cotorsion pair over a valuation domain \textit{not} of cofinite type is presented. The last two chapters on approximation theory applies realization techniques from the second part of the book to cotorsion pairs. Thus they demonstrate a symbiosis between the two seemingly different subjects of the book. Chapter 10 shows that for some cotorsion pairs, it is independent of ZFC whether it is generated by a set. The main result of Section 10.1 is that under the Weak Diamond Principle, every cotorsion pair cogenerated by a set is also generated by a set (Theorem 10.1.2). Its proof is a good example of how one can apply set theoretic principles in algebra. Section 10.2 proves, assuming Shelah's Uniformization Principle, that some cotorsion pairs cogenerated by a set are \textit{not} generated by a set. This is again a nice application of set theory to algebra. Altogether, Chapter 10 demonstrates how one can prove independence results in algebra. Chapter 11 analyses the lattice of cotorsion pairs of Abelian groups. The main result is that every partially ordered set embeds into the lattice of cotorsion pairs, which provides a rich supply of cotorsion pairs. The building blocks are ultra-cotorsion-free modules constructed using the Black Box from the second part (realizing algebras) of the book. The second half of the book is about constructing modules with special properties, e.g., to have a prescribed endomorphism algebra. Chapter 9 contains the set theoretic background for module constructions together with their simplest application. Section 9.1 introduces prediction principles, the algebraic versions of which are statements of the kind that a small set of functions defined on a subset contains approximates for all of the functions on the total set. One of them, the Diamond Principle, is used to prove a baby version (Theorem 9.1.19) of algebra realization based on a step lemma (Lemma 9.1.16). These demonstrate important techniques returning later. Section 9.2 is for the most versatile tool: the Black Box, which does not need any additional set theoretic assumption. Typically, every application needs its own version of the Black Box. Thus, many variants of it are described to help the reader create new versions for her/his own purposes. The Strong Black Box has more restriction on the cardinality of the constructed modules but it is technically easier to handle and hence it is sufficient for most applications. Three versions are presented: (1) for realizing algebras as endomorphism algebras, (2) for constructing \(E(R)\)-algebras and (3) for constructing ultra-cotorsion-free modules for cotorsion pairs. The General Black Box, which was the first version historically, allows more cardinalities but its underlying combinatorial structure is more involved. Section 9.3 presents Shelah's Elevator, which is the basis of realizing algebras with few prescribed submodules to be discussed in Chapter 14. The main realization theorems are the subject of Chapter 12, where a large class of algebras is realized. The main restriction on the algebra to be realized is to be S-cotorsion-free. Subsections 12.1 and 12.2 contain the easier construction for small algebras. Subsection 12.3 explains the harder construction using the Black Boxes, and provides a wide selection of applications. To appreciate the realization theorems, Chapter 15 presents interesting algebras. When realized, they lead to modules with strange decomposition properties. Chapter 14 realizes algebras as the algebra of endomorphisms of a module leaving some submodules invariant. This even works for rings (e.g. fields) where realization of algebras are not possible. Over rings having enough primes, submodules can be made invariant by making them infinitely divisible by a prime, hence this even leads to algebra realization on ordinary modules. The basic tool is Shelah's Elevator. An easy result (Theorem 14.1.6) is realizing algebras with 5 invariant submodules, which are given explicitly. A hard result is that even 4 invariant submodules are enough (Theorem 14.2.12). Section 14.4 constructs `absolutely' indecomposable modules. The word ``absolutely'' means that the endomorphism algebras remain the same in every generic extension of the universe. As usual, this is based on an absolute construction in combinatorial set theory: a construction of a family of absolutely rigid trees. A noteworthy aspect is that there is a large cardinal upper bound (the first Erdős cardinal) on the size of the possible module. Section 14.5 realizes algebras `absolutely' as endomorphism algebras of modules. The construction uses the absolutely indecomposable modules constructed in the previous section. Chapter 13 deals with generalized \(E(R)\)-algebras, which are \(R\)-algebras isomorphic to their own endomorphism algebras. An \(E(R)\)-algebra is a commutative generalized \(E(R)\)-algebra. This chapter comes with many surprises. First, \(E(R)\)-algebras are not isolated objects. Localization functors on the category of Abelian groups produce \(E(R)\)-algebras (Theorem 13.1.6). Also, pure-invertible \(E(R)\)-algebras are u\-nique\-ly transitive modules, so they provide an easy way for constructing uniquely transitive modules (Observation 13.3.7). Second, \(E(R)\)-algebras (especially which are not torsion-free) have strong structures. Yet there are many of them, though they are not easy to construct without the tools of this book. However, the techniques of this book adapt well to constructing them but one has to take into account the structure theorems. Third, (non-commutative) generalized \(E(R)\)-algebras are even harder to construct. The reason is that their multiplication has an unusual structure: it is like function composition in \(\lambda\)-calculus. Thus, first the necessary version of \(\lambda\)-calculus is developed, and then the monoid of the terms of this calculus is used to provide the multiplicative structure of the constructed algebras. All in all, I highly recommend the book to everyone interested in cotorsion pairs, approximation theory, realization of algebras or application of set theory to algebra.