A somewhat gentle introduction to differential graded commutative algebra Kristen A. Beck and Sean Sather-Wagstaff Abstract Differential graded (DG) commutative algebra provides powerful tech- niques for proving theorems about modules over commutative rings. These notes are a somewhat colloquial introduction to these techniques. In order to provide some motivation for commutative algebraists who are wondering about the benefits of learning and using these techniques, we present them in the context of a recent result of Nasseh and Sather-Wagstaff. These notes were used for the course “Differ- ential Graded Commutative Algebra” that was part of the Workshop on Connections Between Algebra and Geometry at the University of Regina, May 29–June 1, 2012. Dedicated with much respect to Tony Geramita 1 Introduction Convention 1.1 The term “ring” is short for “commutative noetherian ring with identity”, and “module” is short for “unital module”. Let R be a ring. These are notes for the course “Differential Graded Commutative Algebra” that was part of the Workshop on Connections Between Algebra and Geometry held at the University of Regina, May 29–June 1, 2012. They represent our attempt to provide a small amount of (1) motivation for commutative algebraists who are won- dering about the benefits of learning and using Differential Graded (DG) techniques, and (2) actual DG techniques. Kristen A. Beck Department of Mathematics, The University of Arizona, 617 N. Santa Rita Ave., P.O. Box 210089, Tucson, AZ 85721, USA, e-mail: [email protected] Sean Sather-Wagstaff Sean Sather-Wagstaff, Department of Mathematics, NDSU Dept # 2750, PO Box 6050, Fargo, ND 58108-6050 USA e-mail: [email protected] 1 2 Kristen A. Beck and Sean Sather-Wagstaff DG Algebra DG commutative algebra provides a useful framework for proving theorems about rings and modules, the statements of which have no reference to the DG universe. For instance, a standard theorem says the following: Theorem 1.2 ([20, Corollary 1]) Let (R;m) ! (S;n) be a flat local ring homo- morphism, that is, a ring homomorphism making S into a flat R-module such that mS ⊆ n. Then S is Gorenstein if and only if R and S=mS are Gorenstein. Moreover, there is an equality of Bass series IS(t) = IR(t)IS=mS(t). (See Definition 9.2 for the term “Bass series”.) Of course, the flat hypothesis is very important here. On the other hand, the use of DG algebras allows for a slight (or vast, depending on your perspective) improvement of this: Theorem 1.3 ([9, Theorem A]) Let (R;m) ! (S;n) be a local ring homomorphism of finite flat dimension, that is, a local ring homomorphism such that S has a bounded resolution by flat R-module. Then there is a formal Laurent series Ij (t) with non-negative integer coefficients such that IS(t) = IR(t)Ij (t). In particular, if S is Gorenstein, then so is R. In this result, the series Ij (t) is the Bass series of j. It is the Bass series of the “homotopy closed fibre” of j (instead of the usual closed fibre S=mS of j that is L used in Theorem 1.2) which is the commutative DG algebra S⊗R R=m. In particular, when S is flat over R, this is the usual closed fibre S=mS =∼ S⊗R R=m, so one recovers Theorem 1.2 as a corollary of Theorem 1.3. Furthermore, DG algebra comes equipped with constructions that can be used to replace your given ring with one that is nicer in some sense. To see how this works, consider the following strategy for using completions. To prove a theorem about a given local ring R, first show that the assumptions ascend to the completion Rb, prove the result for the complete ring Rb, and show how the conclusion for Rb implies the desired conclusion for R. This technique is useful since frequently Rb is nicer then R. For instance, Rb is a homomorphic image of a power series ring over a field or a complete discrete valuation ring, so it is universally catenary (whatever that means) and it has a dualizing complex (whatever that is), while the original ring R may not have either of these properties. When R is Cohen-Macaulay and local, a similar strategy sometimes allows one to mod out by a maximal R-regular sequence x to assume that R is artinian. The regular sequence assumption is like the flat condition for Rb in that it (sometimes) allows for the transfer of hypotheses and conclusions between R and the quotient R := R=(x). The artinian hypothesis is particularly nice, for instance, when R contains a field because then R is a finite dimensional algebra over a field. The DG universe contains a construction Re that is similar R, with an advantage and a disadvantage. The advantage is that it is more flexible than R because it does not require the ring to be Cohen-Macaulay, and it produces a finite dimensional al- gebra over a field regardless of whether or not R contains a field. The disadvantage DG commutative algebra 3 is that Re is a DG algebra instead of just an algebra, so it is graded commutative (al- most, but not quite, commutative) and there is a bit more data to track when working with Re. However, the advantages outweigh the disadvantages in that Re allows us to prove results for arbitrary local rings that can only be proved (as we understand things today) in special cases using R. One such result is the following: Theorem 1.4 ([32, Theorem A]) A local ring has only finitely many semidualizing modules up to isomorphism. Even if you don’t know what a semidualizing module is, you can see the point. Without DG techniques, we only know how to prove this result for Cohen-Macaulay rings that contain a field; see Theorem 2.13. With DG techniques, you get the un- qualified result, which answers a question of Vasconcelos [41]. What These Notes Are Essentially, these notes contain a sketch of the proof of Theorem 1.4; see 5.32, 7.38, and 8.17 below. Along the way, we provide a big-picture view of some of the tools and techniques in DG algebra (and other areas) needed to get a basic under- standing of this proof. Also, since our motivation comes from the study of semid- ualizing modules, we provide a bit of motivation for the study of those gadgets in Appendix 9. In particular, we do not assume that the reader is familiar with the semidualizing world. Since these notes are based on a course, they contain many exercises; sketches of solutions are contained in Appendix 10. They also contain a number of exam- ples and facts that are presented without proof. A diligent reader may also wish to consider many of these as exercises. What These Notes Are Not These notes do not contain a great number of details about the tools and techniques in DG algebra. There are already excellent sources available for this, particularly, the seminal works [4, 6, 10]. The interested reader is encouraged to dig into these sources for proofs and constructions not given here. Our goal is to give some idea of what the tools look like and how they get used to solve problems. (To help readers in their digging, we provide many references for properties that we use.) 4 Kristen A. Beck and Sean Sather-Wagstaff Notation When it is convenient, we use notation from [11, 31]. Here we specify our conven- tions for some notions that have several notations: pdR(M): projective dimension of an R-module M idR(M): injective dimension of an R-module M lenR(M): length of an R-module M Sn: the symmetric group on f1;:::;ng. sgn(i): the signum of an element i 2 Sn. Acknowledgements We are grateful to the Department of Mathematics and Statistics at the University of Regina for its hospitality during the Workshop on Connections Between Algebra and Geometry. We are also grateful to the organizers and participants for providing such a stimulating environment, and to Saeed Nasseh for carefully reading parts of this manuscript. We are also thankful to the referee for valuable comments and corrections. Sean Sather-Wagstaff was supported in part by a grant from the NSA. 2 Semidualizing Modules This section contains background material on semidualizing modules. It also con- tains a special case of Theorem 1.4; see Theorem 2.13. Further survey material can be found in [35, 38] and Appendix 9. Definition 2.1 A finitely generated R-module C is semidualizing if the natural ho- R mothety map cC : R ! HomR(C;C) given by r 7! [c 7! rc] is an isomorphism and i ExtR(C;C) = 0 for all i > 1. A dualizing R-module is a semidualizing R-module such that idR(C) < ¥. The set of isomorphism classes of semidualizing R-modules is denoted S0(R). Remark 2.2 The symbol S is an S, as in \mathfrak{S}. Example 2.3 The free R-module R1 is semidualizing. Fact 2.4 The ring R has a dualizing module if and only if it is Cohen-Macaulay and a homomorphic image of a Gorenstein ring; when these conditions are satisfied, a dualizing R-module is the same as a “canonical” R-module. See Foxby [19, Theorem 4.1], Reiten [34, (3) Theorem], and Sharp [40, 2.1 Theorem (i)]. Remark 2.5 To the best of our knowledge, semidualizing modules were first intro- duced by Foxby [19].