Sorry, your browser cannot access this site
This page requires browser support (enable) JavaScript
Learn more >

一些很简单的projective module的内容, 比较有意思的是最后一个定理. 这个这个定理给出了local ring上的finite generated projective module一定是free的. 这是Kaplanski定理的一个特殊情况, 在这里的证明使用了Nakayama引理数维数.

(头图快不够了, 估计以后要用美少女来代替了 (ㄒoㄒ)/)

Projective Module

Definition 1. An RR-module PP is projective if for any exact sequence ApB0A\xrightarrow{p}B\to 0 and f:PBf:P\to B, there exists a homomorphism g:PAg:P\to A s.t. the following diagram commutes:

PAB0gfp

The homomorphism p:ABp:A\to B will induce a homomorphism from Hom(P,B)\mathrm{Hom}(P,B) to Hom(P,A)\mathrm{Hom}(P,A), i.e. gg is a lifting of ff.

First, we have: free module FF is projective. Take a basis {fi}\{f_{i}\} of FF, from surjectivity of pp, aiA\exists a_{i}\in A s.t. p(ai)=f(fi)p(a_{i})=f(f_{i}). Then assign fif_{i} to aia_{i} and generate gg from it.

Theorem 1. The functor Hom(P,)\mathrm{Hom}(P,-) is exact if and only if PP projective.

Proof

Proof. For the exact sequence 0ABC00\to A\to B\to C\to 0, we have the sequence 0Hom(P,A)Hom(P,B)Hom(P,C)0\to \mathrm{Hom}(P,A)\to \mathrm{Hom}(P,B)\to \mathrm{Hom}(P,C) exact. BC0B\to C\to 0 exact, so Hom(P,B)Hom(P,C)\mathrm{Hom}(P,B)\to \mathrm{Hom}(P,C) is surjective if and only if PP projective. ◻

The projective module is a direct summand of a free module, i.e. PP is projective if and only if M\exists M such that PMP\oplus M is free module. We first show the theorem:

Theorem 2. An RR-module PP is projective if and only if the exact sequence 0ABP00\to A\to B\to P\to 0 splits.

Proof

Proof. Consider the exact sequence $$0\to A\xrightarrow{f}B\xrightarrow{g}P\to 0$$

If PP projective, take the exact sequence BP0B\to P\to 0 and idP:PPid_{P}:P\to P, it induces h:PBh:P\to B with gh=idPg\circ h=id_{P}. So the sequence splits.

Conversely, suppose every exact sequence 0ABP00\to A\to B\to P\to 0 splits. Consider any diagram

PBC0fp

take the free presentation 0KerπFπP00\to \mathrm{Ker}\pi\to F\xrightarrow{\pi} P\to 0 of PP, by assumption

0KerπFπP00\to \mathrm{Ker}\pi\to F\xrightarrow{\pi}P\to 0

splits. So h:PF\exists h:P\to F s.t. hπ=idPh\circ \pi=id_{P}. Since FF is projective, g:FB\exists g:F\to B s.t. fπ=pgf\circ \pi=p\circ g

FPBC0¼ghfp

Thus f=fπh=pghf=f\circ \pi\circ h=p\circ g\circ h, PP is projective. ◻

Theorem 3. PP is projective if and only if it’s a direct summand of a free RR-module FF.

Proof

Proof. If PP is projective, the projective presentation 0KerπFπP00\to \mathrm{Ker}\pi\to F\xrightarrow{\pi}P\to 0 splits, so F=PKerπF=P\oplus \mathrm{Ker}\pi.

Conversely, using similar diagram chasing skill for

FPBC0¼hfp

A direct corollary is

Theorem 4. Direct summands and sums of projective modules are projective modules.

Proof

Proof. Omitted. ◻

We have shown that for projective module PP and any module MM, ExtR1(P,M)=0\mathrm{Ext}_{R}^{1}(P,M)=0 in the Ext functor and Tor functor

Theorem 5. If ExtR1(P,M)=0\mathrm{Ext}_{R}^{1}(P,M)=0 for every RR-module MM, then PP is projective.

Proof

Proof. For a free(projective) resolution

F1F0P0\cdots\to F_{1}\to F_{0}\to P\to 0

Take M=Ker(F0P)M=\mathrm{Ker}(F_{0}\to P), then ExtR1(P,M)=0\mathrm{Ext}_{R}^{1}(P,M)=0. Take an element ξExtR1(P,M)\xi \in \mathrm{Ext}_{R}^{1}(P,M) given by the quotient of π:F1M\pi:F_{1}\to M. Since ξ\xi is zero, there is a map s:F0Ms:F_{0}\to M s.t. π=sd1\pi=s\circ d_{1}, d1:F1F0d_{1}:F_{1}\to F_{0}. Thus F0=KersKer(F0P)=PKer(F0P)F_{0}=\mathrm{Ker}s\oplus \mathrm{Ker}(F_{0}\to P)=P\oplus \mathrm{Ker}(F_{0}\to P). ◻

For projective module, it is not necessary to be free. For example, for the ring Z/6Z\mathbb{Z} /\mathbb{6Z}, the ideal (2)(2) is projective since (2)(3)Z/6Z(2)\oplus (3)\cong \mathbb{Z} /6\mathbb{Z}. Actually, any infinitely generated projective module over any Dedekind domain. We proof some weaker result:

Theorem 6. Any finite generated projective module MM of a local ring RR is free.

Proof

Proof. Take a minimal generating basis m1,mnm_{1},\dots m_{n} of MM, here minimal means the minimal number of generators. Let F=i=1nReiF=\bigoplus\limits_{i=1}^{n}Re_{i}, and φ:FM\varphi:F\to M with φ:rieirimi\varphi: \sum r_{i}e_{i}\mapsto \sum r_{i}m_{i}. Take K=KerφK=\mathrm{Ker}\varphi. If rimi=0M\sum r_{i}m_{i}=0\in M, pass it to the quotient M/mMM /\mathfrak{m}M, rimiˉ=0M/mM\sum r_{i}\bar{m_{i}}=0\in M /\mathfrak{m}M (Recall R/mRMM/mMR /\mathfrak{m}\otimes_{R} M\cong M /\mathfrak{m}M !). So rimr_{i}\in \mathfrak{m}. Hence KmFK\subset \mathfrak{m}F. Since MM is projective, the sequence

0KFM00\to K\to F\to M\to 0

splits. Thus FMKF\cong M\oplus K. Since FF and MM have same number of genenrators, dimR/m(R/mRF)=dimR/m(R/mRM)F/mFM/mM\dim_{R /\mathfrak{m}}(R/\mathfrak{m}\otimes_{R}F)=\dim_{R /\mathfrak{m}}(R /\mathfrak{m}\otimes_{R}M)\Rightarrow F /\mathfrak{m}F\cong M /\mathfrak{m}M.
We have

R/mRFF/mFR/mR(MK)(R/mRM)(R/mRK)M/mMK/mKR /\mathfrak{m}\otimes_{R} F\cong F /\mathfrak{m}F\cong R /\mathfrak{m}\otimes_{R}(M\oplus K)\cong (R /\mathfrak{m}\otimes_{R}M)\oplus (R /\mathfrak{m}\otimes_{R}K)\cong M /\mathfrak{m}M\oplus K/\mathfrak{m}K

Thus K/mK=0K /\mathfrak{m}K=0. Since KK is finite generated, by Nakayama lemma, K=0K=0. Thus F=MF=M. ◻

A more stronger result can be seen in Projective modules over a local ring.

评论