Abstract. Differential Graded Algebras can be studied through their Differential Graded modules. Among these, the compact ones attract particular attention. This paper proves that over a suitable chain Differential Graded Algebra R, each compact Differential Graded module M satisfies amp M ≥ amp R, where amp denotes amplitude which is defined in a straightforward way in terms of the homology of a DG module.In other words, the homology of each compact DG module M is at least as long as the homology of R itself. Conversely, DG modules with shorter homology than R are not compact, and so in general, there exist DG modules with finitely generated homology which are not compact.Hence, in contrast to ring theory, it makes no sense to define finite global dimension of DGAs by the condition that each DG module with finitely generated homology must be compact.
IntroductionDifferential Graded Algebras (DGAs) play an important role in both ring theory and algebraic topology. For instance, if M is a complex of modules, then the endomorphism complex Hom(M, M) is a DGA with multiplication given by composition of endomorphisms, and this can be used to prove ring theoretical results, see [12] and [13]. Another example is that over a commutative ring, the Koszul complex on a series of elements is a DGA, see [17, sec. 4.5], and again, ring theoretical results ensue, see [10].Likewise, DGAs occur naturally in algebraic topology, where the canonical example is the singular cochain complex C * (X) of a topological space X. Other constructions also give DGAs; for instance, if G is a topological monoid, then the singular chain complex C * (G) is a DGA whose multiplication is induced by the composition of G; see [4].Just as rings can be studied through their modules, DGAs can be studied through their Differential Graded modules (DG modules), and this is the subject of the present paper.The main results are a number of "amplitude inequalities" which give bounds on the amplitudes of various types of DG modules. Such results have been known for complexes of modules over rings since Iversen's paper [9], and it is natural to seek to extend them to DG modules.Another main point, implied by one of the amplitude inequalities, is that, in contrast to ring theory, it appears to make no sense to define finite global dimension of DGAs by the condition that each DG module with finitely generated homology must be compact. This is of interest since several people have been asking how one might define finite global dimension for DGAs.First main Theorem. To get to the first main Theorem of the paper, recall from [13] that if R is a DGA then a good setting for DG modules over R is the derived category of DG left-R-modules D(R).A DG left-R-module is called compact if it is in the smallest triangulated subcategory of D(R) containing R, or, to use the language of topologists, if it can be finitely built from R. The compact DG left-Rmodules form a triangulated subcategory D c (R) of D(R), and play the same important role as finitely presented modules of fi...