Projective and Injective Model Structure on Chr

Projective and Injective Model Structure on Chr

Projective and injective model structure on ChR Shlomi Agmon Thursday, July 2nd, 2012 The material in sections 1 and 3 is taken mostly from [Rotman, 7.4]. The rest of the material is from [DS95, Section 7], unless mentioned otherwise. 1 Reminder - projectives Reminder: Both Hom functors are generally only left exact, while tensor prod- uct M ⊗ − is only right exact. An R-module P is called projective if the following equivalent conditions are satised: f • Given a surjection B ! C ! 0, any R-map P ! C lifts to B: P f B / / C / 0 • HomR (P; −) is exact. • Every short exact sequence 0 ! A ! B ! P ! 0 ending with P splits. • P is a direct summand of a free R module. An abelian category A is said to have enough projectives if every object B 2 A is a quotient of a projective object, meaning there exists a projective P 2 A such that P ! B ! 0 is exact. Note that all of the above could have actually been done in an arbitrary abelian category. For a general (not necessarily abelian) category C, P 2 C is called a projective object if HomC (P; −) sends epimorphisms (in C) to epimor- phisms (in Sets). In the following we specialize to R-modules. Example. A free module is projective. R-Mod has enough projectives because every module is a quotient of a free module. In particular, Ab = Z − Mod has enough projectives. 1 There are modules which are projective but not free1: Let . Then is a projective -module because it is a direct R = Z3 ⊕ Z2 Z3 R summand of the free R-module R, but it is not free. p p A more interesting example: R := Z −5 , then the ideal I = 3; 2 + −5 is a non-principal ideal, therefore cannot be free. However, I is a direct sum- mand of R2 .2 2 Projective model structure on ChR Let R be an associative ring with unit. ModR be left R-modules, ChR be (non- negatively graded) chain complexes of R-modules. Hk : ChR ! ModR the homology functors (k ≥ 0). We will prove portions of the following standard projective model structure on ChR: Theorem 1 ([DS95, 7.2]). Dene a map f : M ! N in ChR to be: (i) a weak equivalence = f induces isomorphisms HkM ! HkN (k ≥ 0), (ii) a cobration if 8k ≥ 0, fk : Mk ! Nk is a monomorphism + cokernel a projective R-module, (iii) a bration if 8k > 0, fk : Mk ! Nk is an epimorphism. Then with these choices ChR is a model category. 2.1 Reminder - basics of MC 3 classes that include identities, closed under composition, and: MC1 All small (co+)limits exist. MC2 2 out of 3 for w.e.. MC3 All 3 classes are closed under retracts. MC4 A lift B 99K X exists for A / X i p B / Y whenever i is a cobration, p a bration, and at least one of them acyclic. ∼ MC5 Any map f can be factored (functorially) in two ways: • ,!• • and ∼ • ,!• • Remarks: 1Over a PID R, projective , free. Because M is a projective R-module =) it is a direct summand of a free R-module. But over a PID, any submodule of a free module is free. u p 2Take the surjection R2 ! I; 7! u · 3 + v · 2 + −5 in one direction, the other is a v direct calculation 2 • Initial/terminal object, ;=∗ : the zero complex. ∼ c ∼ f • (co)brant; (co)brant replacement of X : ; ,! X X, X ,! X ∗. • In particular, Fibrant: any object is brant in this model structure. So we can take Xf = X. Cobrant = chain complexes of projectives, meaning chain complexes which are level-wise projective. 3 Warning: A projective object in ChR is a chain complex of projectives , but 4 a chain complex of projectives need not be a projective object in ChR . • Xc = a projective resolution (at least when X is concentrated in dim 0), ::: / P2 / P1 / P0 / 0 ∼ 0 / X0 / 0 For a chain complex concentrated in dim 0, the fact that R-Mod has enough projectives implies that a cobrant replacements exist. For a general chain complex this follow from MC5, which we will also prove using the same fact. 2.2 Proof of projective model structure - MC1 to MC4 Includes identities + closed under composition = easy5. MC1 Dimension-wise. p MC2 (2 out of 3) MC3 Technical and easy. 3Denote by K (A; n) the chain complex which is A at dimension n, and zero otherwise. We have the adjunction ∼ , where sends HomChR (M; K (A; n)) = HomR−Mod (Mn;A) M 7! Mn a chain complex M to the R-module at it's n-th dimension. Since the right adjoint K (−; n) preserves epomirohisms, we have that the left adjoint M 7! Mn preserves projectives. 4 The projective objects of ChR are precisely the acyclic chain complexes of projectives. ·r So ···! 0 ! R ! R ! 0 for r 2 R non-unit is an example of a chain complex of projectives which is not a projective chain complex. Projective unbounded chain complexes have a slightly dierent character- ization. See more details at: http://mathoverflow.net/questions/103584/ on-the-difference-between-a-projective-chain-complex-and-a-level-wise-projective. 5The only point which requires some eort is that a composition of cobrations is a co- bration. Suppose 0 ! L ! M ! N is a series of monomorphisms, each with projective (level-wise) cokernels. Then M N N is exact, with the rst and last terms 0 ! L ! L ! M ! 0 being projective, thus also N is projective, as a direct sum of projective modules. L 3 MC4 Technical. There are, however, some interesting results en-route: Id Let A be an R-module. Dene Dn (A) to be A ! A in dimensions n; n − 1, zero otherwise. We have an adjunction ∼= (2.1) HomChR (Dn (A) ;M) ! HomModR (A; Mn) sending f 7! fn. From this adjunction it follows immediately that if A is a projective R- 6 module then Dn (A) is an acyclic and projective object in ChR. Obviously, a direct sum of such projective disks is projective and acyclic. Denote by ZkM the k-cycles of a chain complex M. We have a converse: Lemma ([DS95, 7.10]). Suppose that P in ChR is an acyclic chain complex of projectives (=level-wise projective). Then each module ZkP (k ≥ 0) is projec- tive, and is isomorphic as a chain complex to L . P k≥1 Dk (Zk−1P ) n Denote for short by Dn the chain complex Dn (R). Set S to be K (R; n), the chain complex which is R in dimension n, zero otherwise. There is the n−1 n obvious inclusion jn : S ! D . We have the following characterization of brations, which is very resemblant to that of Serre brations: Proposition ([DS95, 7.19]). A map f : X ! Y in ChR is (i) a bration i it has the RLP w.r.s.t. the maps 0 ! Dn for all n ≥ 1, and n−1 n (ii) an acyclic bration i it has the RLP w.r.s.t. the maps jn : S ! D for all n ≥ 0. The proof is a nice exercise in diagram chasing. 2.3 Proof of MC5 (factorization) A useful argument in order to produce factorizations of maps with lifting prop- erties is the small object argument. We will not present it, however, and will instead pursue a more direct approach. The reader is urged to look the small object argument in the literature ([DS95, 7.12], for example). The proof below is taken from [Riehl]. 2.3.1 Acyclic cobration - bration factorization Let M be an R-module, and choose a set of generators. M can be written as a quotient of the projective R-module L R, where we have taken one copy of R for each generator of M. The problem with this construction is that it is not functorial. 6A fancier way to show the same thing is the following: In an adjoint pair, if the right adjoint preserves epimorphisms, then the left adjoint preserves projective objects (direct proof). The functor M 7! Mn obviously preserves epimorphisms. 4 Let f : M ! N be a morphism in ChR. If we could make a functorial construction in ChR similar to the above of an acyclic chain complex P (N) of projectives, then we could factorize f in the form f M N 9/ i f⊕e M % M ⊕ P (N) which is clearly an acyclic cobration followed by a bration. Since the con- struction P (−) will be functorial, the obtained factorization will be functorial as well. Dene now M P (N) := Dk maps Dk!N with the obvious projection e : P (N) ! N, sending the copy of Dk indexed by g : Dk ! N to it's image under g. It is a direct sum of acyclic chain complexes of projectives, hence an acyclic complex of projectives. 2.3.2 Cobration - acyclic bration factorization For a chain map f : M ! N, consider the commutative diagram fn Mn / Nn @ @ Zn−1M / Zn−1N fn−1 induced by f. It induces a map to the pullback (omitting Mn from the corner of the last diagram) Mn ! Zn−1M × Nn (2.2) Zn−1N The proof of this factorization relies on the following technical lemma, which can be proven by standard diagram chasing methods. Lemma ([Riehl, 3.19]). f is a trivial bration i the map in (2.2) is a surjection for all n ≥ 0. Now that we are properly equipped, we can turn to the heart of the proof.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    9 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us