Appendix A. Tensor products, direct and inverse limits

FC13 — Matsumura, Commutative Ring Theory.

This chapter contains 10 definitions, 2 theorems, 2 examples, 1 counterexample, 4 constructions and 8 formulas.

Nobody has checked this chapter against Lean yet, so the statements below say nothing about whether they are formalized.

Definition: bilinear map

FC13-CA-U001 · Appendix A, p. 266

For A-modules M,N,L, a map M×N→L is bilinear when it is A-linear in each argument separately; T_A(M,N;L) denotes the A-module of bilinear maps.

Definition/construction: tensor product

FC13-CA-U002 · Appendix A, p. 266 · after FC13-CA-U001

The tensor product M⊗_A N is characterized by a universal bilinear map M×N→M⊗_A N. It exists as the quotient of the free A-module on M×N by the bilinearity relations.

Formula: Formula 1 (tensor universal property)

FC13-CA-U003 · Appendix A, p. 266 · after FC13-CA-U002

Hom_A(M⊗_A N,L)≅T_A(M,N;L) naturally.

Construction/formula: multilinear tensor products and Formula 2

FC13-CA-U004 · Appendix A, pp. 266–267 · after FC13-CA-U002

The r-fold tensor product represents r-multilinear maps. Tensor product is associative: (M⊗M′)⊗M″≅M⊗M′⊗M″≅M⊗(M′⊗M″).

Formulas: Formulas 3–5

FC13-CA-U005 · Appendix A, p. 267 · after FC13-CA-U002

There are canonical isomorphisms M⊗N≅N⊗M, M⊗A≅M, and (⊕_λ M_λ)⊗N≅⊕_λ(M_λ⊗N).

Formula: Formula 6 (tensor product of maps)

FC13-CA-U006 · Appendix A, p. 267 · after FC13-CA-U002

Linear maps f:M→M′ and g:N→N′ induce f⊗g:M⊗N→M′⊗N′ with (f⊗g)(Σx_i⊗y_i)=Σf(x_i)⊗g(y_i).

Formula: Formula 7 (tensor quotient for two surjections)

FC13-CA-U007 · Appendix A, p. 267 · after FC13-CA-U006

For exact 0→K→M→M′→0 and 0→L→N→N′→0, M′⊗N′ is (M⊗N) modulo the sum of the images K⊗N and M⊗L.

Formula: Formula 8 (right exactness of tensor product)

FC13-CA-U008 · Appendix A, pp. 267–268 · after FC13-CA-U007

If M₁→M₂→M₃→0 is exact, then M₁⊗N→M₂⊗N→M₃⊗N→0 is exact.

Definition/lemma: split exactness and flat modules

FC13-CA-U009 · Appendix A, p. 268 · after FC13-CA-U008

Tensoring preserves split injections. An A-module N is flat precisely when −⊗_A N is exact.

Definition: two-sided A–B module and scalar action on tensor products

FC13-CA-U010 · Appendix A, p. 268

For an A–B-bimodule P, M⊗_A P is naturally a B-module by (Σy_i⊗x_i)b=Σy_i⊗(x_i b).

Formula: Formula 9 (Hom–tensor adjunction for bimodules)

FC13-CA-U011 · Appendix A, p. 268 · after FC13-CA-U010

For A–B-bimodule P, A-module M and B-module N, Hom_A(M,Hom_B(P,N))≅Hom_B(M⊗_A P,N).

Formula: Formula 10 (associativity under change of rings)

FC13-CA-U012 · Appendix A, p. 268 · after FC13-CA-U010

For compatible bimodules, (M⊗_A P)⊗_B N≅M⊗_A(P⊗_B N).

Definition: extension of scalars

FC13-CA-U013 · Appendix A, p. 268 · after FC13-CA-U010

For a ring homomorphism A→B and A-module M, M⊗_A B is the B-module obtained by extension of scalars, also denoted M_B.

Formula: Formula 11 (tensor product commutes with scalar extension)

FC13-CA-U014 · Appendix A, p. 268 · after FC13-CA-U012, FC13-CA-U013

For A-modules M,M′ and A→B, (M⊗_A B)⊗_B(M′⊗_A B)≅(M⊗_A M′)⊗_A B.

Construction: tensor product of A-algebras

FC13-CA-U015 · Appendix A, p. 269 · after FC13-CA-U002

For A-algebras B,C, the A-module B⊗_A C is an A-algebra with (b⊗c)(b′⊗c′)=bb′⊗cc′; it is generated by B⊗1 and 1⊗C.

Examples: quotient and polynomial base change

FC13-CA-U016 · Appendix A, p. 269 · after FC13-CA-U015

For an ideal a⊲A, B⊗_A(A/a)≅B/aB. Also B⊗_A A[X]≅B[X], and similarly in several variables.

Definition: directed set

FC13-CA-U017 · Appendix A, p. 269

A poset Λ is directed if any two elements admit a common upper bound; finite subsets ordered by inclusion are the standard non-total example.

Definition: direct system

FC13-CA-U018 · Appendix A, pp. 269–270 · after FC13-CA-U017

A direct system indexed by a directed set consists of objects M_λ and transition maps f_{μλ}:M_λ→M_μ for λ≤μ satisfying identity and composition laws; the definition applies to sets, modules, rings, and categories.

Definition: morphism of direct systems

FC13-CA-U019 · Appendix A, p. 270 · after FC13-CA-U018

A morphism of direct systems with the same index set is a compatible family of maps commuting with all transition maps.

Definition/construction: direct limit

FC13-CA-U020 · Appendix A, p. 270 · after FC13-CA-U018

The direct limit lim→M_λ is characterized by the universal property for compatible maps out of the system. It exists as the quotient of the disjoint union by eventual equality; module and ring structures descend when present.

Theorem: Theorem A1

FC13-CA-U021 · Appendix A, pp. 270–271 · after FC13-CA-U020

Tensor product commutes with direct limits: lim→(M_λ⊗_A N)≅(lim→M_λ)⊗_A N.

Theorem: Theorem A2

FC13-CA-U022 · Appendix A, p. 271 · after FC13-CA-U020

Direct limits of A-modules are exact: a levelwise exact sequence of three direct systems yields an exact sequence of their direct limits.

Construction: module as direct limit of finitely generated submodules

FC13-CA-U023 · Appendix A, p. 271 · after FC13-CA-U020

Every A-module is the direct limit of its finitely generated submodules ordered by inclusion.

Construction: ring as direct limit of Noetherian subrings

FC13-CA-U024 · Appendix A, p. 271 · after FC13-CA-U020

Every ring is a direct limit of subrings finitely generated over its minimal subring; these subrings are Noetherian.

Definition/construction: inverse system and inverse limit

FC13-CA-U025 · Appendix A, pp. 271–272 · after FC13-CA-U017

Inverse systems and inverse limits are the arrow-reversed notions. The inverse limit is the compatible-subset of the direct product ∏M_λ and inherits module/ring structure.

Example: p-adic integers and I-adic completion

FC13-CA-U026 · Appendix A, p. 272 · after FC13-CA-U025

The inverse limit lim← Z/(p^n) is the ring of p-adic integers. More generally lim← A/I^n is the I-adic completion of A.

Counterexample: inverse limit is left exact but not exact

FC13-CA-U027 · Appendix A, pp. 272–273 · after FC13-CA-U025

Inverse limit of modules is left exact in general but not right exact; the source gives an inverse system of exact rows whose inverse-limit map fails surjectivity.