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

局部化和取商是交换代数里面最常用的技巧. 取商能够给出包含理想的理想, 而局部化则能给出素理想内部的理想. 可以通过saturation 来一般地刻画S1RS^{-1}R 中的理想. 而对于module 而言, S1()S^{-1}(-) 这一个函子的正合性十分重要, 这给出了很多局部化module 的商的结果. 同样的, 也可以通过saturation 来刻画S1MS^{-1}M 的子模. 局部环在几何上是在一点处的连续函数germ 空间, 这为我们研究局部环提供了理由.

Localization of Ring

First we give the universal property of localization.

Theorem 1. Let RR be a ring, SS is a multiplicative subset. For any homomorphism ψ:RR\psi:R\to R', ψ(S)R×\psi(S)\subset R'^{\times} if and only if there exists a ring homomorphism ρ:S1RR\rho: S^{-1}R\to R' s.t. ρφS=ψ\rho\varphi_{S}=\psi, where φS\varphi_{S} is the canonical homomorphism: φS:RS1R\varphi_{S}:R\to S^{-1}R.

RS¡1RR0'Sý

If so ,ρ\rho is unique and Kerρ=KerψS1R\mathrm{Ker}\rho=\mathrm{Ker}\psi S^{-1}R.

Proof

Proof. If ρ\rho exists, for sSs\in S, ψ(s)=ρ(s/1)\psi(s)=\rho(s /1) so ψ(s)ρ(1/s)=ρ(1)=1\psi(s)\rho(1 /s)=\rho(1)=1. ψ(S)R×\psi(S)\subset R'^{\times }.

Conversely, if ψ(S)R×\psi(S)\subset R'^{\times }, take ρ(x/s)=ψ(x)ψ(s)1\rho(x /s)=\psi(x)\psi(s)^{-1}. It’s easy to check ρ\rho is well-defined homomorphism. And xR\forall x\in R, ρφS(x)=ρ(x/1)=ψ(x)ψ(1)1=ψ(x)\rho\varphi_{S}(x)=\rho(x /1)=\psi(x)\psi(1)^{-1}=\psi(x).

Lastly, KerρKerψS1R\mathrm{Ker}\rho\supset \mathrm{Ker}\psi S^{-1}R is trivial. Conversely, x/sKerρ\forall x/s\in\mathrm{Ker}\rho, ψ(x)ψ(s)1=0ψ(x)=0\psi(x)\psi(s)^{-1}=0\Rightarrow \psi(x)=0. So x/sKerψS1Rx /s\in\mathrm{Ker}\psi S^{-1}R. ◻

Corollary 1. Let RR be a ring, SS be a multiplicative subset. The canonical homomorphism φS:RS1R\varphi_{S}:R\to S^{-1}R is an isomorphism if and only if SS consists of units.

Proof

Proof. Omitted. ◻

We can localize a ring at any element fRf\in R.

Definition 1. Let RR be a ring, Sf={fnn0}S_{f}=\{f^{n}|n\geq 0\}. Then we call Sf1RS_{f}^{-1}R the localization of RR at ff, denoted by RfR_{f}.

Localization at an element can be easily charactorized.

Theorem 2. Let fRf\in R, XX is an indeterminate, then RfR[X]/(1fX)R_{f}\cong R[X] /(1-fX).

Proof

Proof. We use the universal property to prove it. Let xR[X]/(1fX)x\in R[X] /(1-fX) be the image of XX. Then 1φ(f)X=01-\varphi(f)X=0. So φ(f)\varphi(f) is a unit. φ(Sf)(R[X]/(1fX))×\varphi(S_{f})\subset (R[X] /(1-fX))^{\times }. Let any ψ:RR\psi:R\to R' with ψ(f)\psi(f) a unit in RR'. Define θ:R[X]R\theta:R[X]\to R' by θR=ψ\theta|_{R}=\psi and θ(X)=ψ(f)1\theta(X)=\psi(f)^{-1}. Then θ(1fX)=0\theta(1-fX)=0, (1fX)Kerθ(1-fX)\subset \mathrm{Ker}\theta, we have unique homomorphism ρ:R[X]/(1fX)R\rho:R[X] /(1-fX)\to R' s.t. ψ=ρφ\psi=\rho\varphi. Thus R[X]/(1fX)R[X] /(1-fX) satisfies the universal property, i.e. R[X]/(1fX)RfR[X] /(1-fX)\cong R_{f}. ◻

Proposition 1. Let RR be a ring, SS multiplicative subset, a\mathfrak{a} an ideal of RR.

  1. aS1R={a/saa,sS}\mathfrak{a}S^{-1}R=\{a /s|a\in \mathfrak{a}, s\in S\}.

  2. aS=\mathfrak{a}\cap S=\varnothing if and only if aS1R=S1R\mathfrak{a}S^{-1}R=S^{-1}R if and only if φS1(aS1R)=R\varphi_{S}^{-1}(\mathfrak{a}S^{-1}R)=R.

Proof

Proof.

  1. Let N={a/saa,sS}N=\{a /s|a\in \mathfrak{a}, s\in S\}. Obviously, NaS1RN\subset \mathfrak{a}S^{-1}R. Conversely, for a,baa, b\in \mathfrak{a}, x/s,y/tS1Rx /s, y /t\in S^{-1}R. Then φ(a)x/s+φ(b)y/t=ax/s+by/t=(axt+bys)/st\varphi(a)x /s+\varphi(b)y /t=ax /s+by /t=(axt+bys) /st, so aS1RN\mathfrak{a}S^{-1}R\subset N.

  2. If saSs\in\mathfrak{a}\cap S, then 1S1R=s/saS1R1_{S^{-1}R}=s /s\in\mathfrak{a}S^{-1}R, so aS1R=S1R\mathfrak{a}S^{-1}R=S^{-1}R. φS1(aS1R)=R\varphi_{S}^{-1}(\mathfrak{a}S^{-1}R)=R.
    Conversely, if φS1(aS1R)=R\varphi_{S}^{-1}(\mathfrak{a}S^{-1}R)=R, then 1aS1R1\in \mathfrak{a}S^{-1}R. So aa\exists a\in\mathfrak{a}, sSs\in S s.t. a/s=1a /s=1, so tS\exists t\in S with at=stat=st. Since ataat\in\mathfrak{a}, stSst\in S, aS=\mathfrak{a}\cap S=\varnothing. ◻

To describe the ideals of S1RS^{-1}R more precisely, we use the saturation: aS={aRsS s.t. asa}\mathfrak{a}^{S}=\{a\in R|\exists s\in S \text{ s.t. }as\in\mathfrak{a}\}, i.e. aS=sS(a:s)\mathfrak{a}^{S}=\cap_{s\in S}(\mathfrak{a}:s). And it’s easy to show KerφS=(0)S\mathrm{Ker}\varphi_{S}=(0)^{S} and aaS\mathfrak{a}\subset \mathfrak{a}^{S}. aS\mathfrak{a}^{S} is an ideal.

Proposition 2. Let RR be a ring, SS a multiplicative subset,

  1. Let b\mathfrak{b} be an ideal of S1RS^{-1}R, then φS1(b)=(φS1(b))S\varphi_{S}^{-1}(\mathfrak{b})=(\varphi_{S}^{-1}(\mathfrak{b}))^{S}, b=φS1(b)S1R\mathfrak{b}=\varphi_{S}^{-1}(\mathfrak{b})S^{-1}R.

  2. Let a\mathfrak{a} be an ideal of RR, then aS1R=aSS1R\mathfrak{a}S^{-1}R=\mathfrak{a}^{S}S^{-1}R, φS1(aS1R)=aS\varphi_{S}^{-1}(\mathfrak{a}S^{-1}R)=\mathfrak{a}^{S}.

  3. Let p\mathfrak{p} be a prime ideal of RR, pS=\mathfrak{p}\cap S=\varnothing, then p=pS\mathfrak{p}=\mathfrak{p}^{S} and pS1R\mathfrak{p}S^{-1}R is prime in S1RS^{-1}R.

Proof

Proof.

  1. First, a(φS1(b))S\forall a\in (\varphi_{S}^{-1}(\mathfrak{b}))^{S}, sS\exists s\in S s.t. asφS(b)as\in \varphi_{S}(\mathfrak{b}). Then as/1bas /1\in\mathfrak{b}, so a/1ba /1\in\mathfrak{b}. Thus a=φS1(a/1)φS1(b)φS1(b)=φS1(b)\mathfrak{a}=\varphi_{S}^{-1}(a /1)\in \varphi_{S}^{-1}(\mathfrak{b})\Rightarrow \varphi_{S}^{-1}(\mathfrak{b})=\varphi_{S}^{-1}(\mathfrak{b}).
    Second, a/sb\forall a /s\in \mathfrak{b}, a/1ba /1\in\mathfrak{b}. So aφS1(b)a/s(φS1(b))S1Ra\in \varphi_{S}^{-1}(\mathfrak{b})\Rightarrow a /s\in(\varphi_{S}^{-1}(\mathfrak{b}))S^{-1}R. The converse is obvious.

  2. First, aaS\forall a\in \mathfrak{a}^{S}, sS\exists s\in S, asaas\in \mathfrak{a} so a/1=as/11/saS1Ra /1=as /1\cdot 1 /s\in\mathfrak{a}S^{-1}R. Thus aSS1RaS1R\mathfrak{a}^{S}S^{-1}R\subset \mathfrak{a}S^{-1}R. So aSS1R=aS1R\mathfrak{a}^{S}S^{-1}R=\mathfrak{a}S^{-1}R.
    Second, φS1(aS1R)=φS1(aSS1R)aS\varphi_{S}^{-1}(\mathfrak{a}S^{-1}R)=\varphi_{S}^{-1}(\mathfrak{a}^{S}S^{-1}R)\supset \mathfrak{a}^{S}. Conversely, xφS1(aSS1R)\forall x\in \varphi_{S}^{-1}(\mathfrak{a}^{S}S^{-1}R), x/1aSS1R=aS1Rx /1\in\mathfrak{a}^{S}S^{-1}R=\mathfrak{a}S^{-1}R. So x/1=a/sx /1=a /s for aaa\in \mathfrak{a}, sSs\in S. Thus (xsa)t=0(xs-a)t=0 for some tSxaSt\in S\Rightarrow x\in \mathfrak{a}^{S}.

  3. First, if sapsa\in \mathfrak{p} for sSs\in S, spaps\notin \mathfrak{p}\Rightarrow a\in \mathfrak{p}. So pS=p\mathfrak{p}^{S}=\mathfrak{p}.
    Second, for any a/sb/tpS1Ra /s\cdot b /t\in \mathfrak{p}S^{-1}R, we have abφS1(pS1R)=pS=pab\in \varphi_{S}^{-1}(\mathfrak{p}S^{-1}R)=\mathfrak{p}^{S}=\mathfrak{p}. So apa\in \mathfrak{p} or bpb\in\mathfrak{p}. Then a/sS1Ra /s\in S^{-1}R or b/tpS1Rb /t\in \mathfrak{p}S^{-1}R. Hence pS1R\mathfrak{p}S^{-1}R is prime. ◻

From the above theorem, we can conclude that φS\varphi_{S} gives a bijection between ideals a\mathfrak{a} of RR with a=aS\mathfrak{a}=\mathfrak{a}^{S} and ideals of S1RS^{-1}R.

For the prime ideals pS=\mathfrak{p}\cap S=\varnothing, we have p=pS\mathfrak{p}=\mathfrak{p}^{S} so there is bijection between prime ideal pS=\mathfrak{p}\cap S=\varnothing and prime ideals of S1RS^{-1}R.

Definition 2. Let RR be a ring, p\mathfrak{p} a prime ideal of RR. Let Sp=RpS_{\mathfrak{p}}=R-\mathfrak{p}. We call the ring Sp1RS_{\mathfrak{p}}^{-1}R the localization of RR at p\mathfrak{p}, and denoted by RpR_{\mathfrak{p}}.

RpR_{\mathfrak{p}} is the local ring with the maximal ideal pRp\mathfrak{p}R_{\mathfrak{p}}.

Theorem 3. Let RR be a ring, SS multiplicative subset, Let TT' be a multiplicative subset of S1RS^{-1}R, T=φS1(T)T=\varphi_{S}^{-1}(T'). Assume STS\subset T, then (T)1(S1R)=T1R(T')^{-1}(S^{-1}R)=T^{-1}R.

Proof

Proof. Idea: use the universal property. Omitted. ◻

Corollary 2. Let RR be a ring, pq\mathfrak{p}\subset \mathfrak{q} prime ideals. Then RpR_{\mathfrak{p}} is the localization of RqR_{\mathfrak{q}} at the prime ideal pRq\mathfrak{p}R_{\mathfrak{q}}.

Proof

Proof. Let S=RqS=R-\mathfrak{q}, T=RqpRqT'=R_{\mathfrak{q}}-\mathfrak{p}R_{\mathfrak{q}}. Set T=φS1(T)T=\varphi_{S}^{-1}(T'). Then T=RpT=R-\mathfrak{p}. By above theorem, we have RpR_{\mathfrak{p}} the localization of RqR_{\mathfrak{q}} at pRq\mathfrak{p}R_{\mathfrak{q}}. ◻

Localization of Module

Theorem 4. Let RR be a ring, SS multiplicative subset. Then MM has a compatible S1RS^{-1}R-module structure if and only if sS\forall s\in S, the homomorphism μs:msm\mu_{s}:m\mapsto sm is an isomorphism. If so , the S1RS^{-1}R-module structure is unique.

Proof

Proof. Assume MM has a compatible S1RS^{-1}R-module structure, take sSs\in S. Then μs=μs/1\mu_{s}=\mu_{s /1}. So μsμ1/s=1\mu_{s}\cdot \mu_{1 /s}=1. Similarly, μ1/sμs=1\mu_{1 /s}\cdot \mu_{s}=1. μs\mu_{s} is bijective.

Conversely,if μs\mu_{s} is bijective for all sSs\in S. Then μ:REnd(M)\mu:R\to \mathrm{End}(M) given by μ:sμs\mu:s\mapsto \mu_{s} sends SS to units of End(M)\mathrm{End}(M). Hence there’s unique MS1R:S1REnd(M)M_{S^{-1}R}:S^{-1}R\to \mathrm{End}(M), there’s unique compatible S1RS^{-1}R-module structure on MM. ◻

Now we give the universal property of localized modules.

Theorem 5. Let RR be a ring. SS a multiplicative subset, MM an RR-module. Then for any S1RS^{-1}R-module NN with RR-module homomorphism ψ:MN\psi:M\to N, there’s a S1RS^{-1}R-module homomorphism: ρ:S1MN\rho:S^{-1}M\to N s.t. ρφS=ψ\rho\varphi_{S}=\psi.

Proof

Proof. Similar as the analogy in ring. ◻

We can show that S1()S^{-1}(-) is actually a functor. Moreover, it’s the left adjoint to the functor of restriction of scalars, i.e. for RR-module MM and S1RS^{-1}R-module NN, we can regard NN as an RR-module. Then HomS1R(S1M,N)HomR(M,N)\mathrm{Hom}_{S^{-1}R}(S^{-1}M,N)\cong \mathrm{Hom}_{R}(M,N) by universal property.

Corollary 3. Let RR be a ring, SS be a multiplicative subset. Then S1()S^{-1}(-) and S1RRS^{-1}R\otimes_{R}- are naturally isomorphic.

Similar as the localization of rings, we use the saturation of a submodule to charactorize φS1\varphi_{S}^{-1}. For NMN\subset M be RR-modules, the saturation NS={mMs s.t. msN}N^{S}=\{m\in M|\exists s \text{ s.t. }ms\in N\}.

Proposition 3. Let RR be a ring, MM be an RR-module, NN, PP submodules. Let S,TS,T be multiplicative subsets, KK an S1RS^{-1}R-submodule of S1MS^{-1}M. Then

  1. φS1(K)=(φS1(K))S\varphi_{S}^{-1}(K)=(\varphi_{S}^{-1}(K))^{S} and K=S1(φS1(K))K=S^{-1}(\varphi_{S}^{-1}(K)).

  2. φS1(S1N)=NS\varphi_{S}^{-1}(S^{-1}N)=N^{S}, S1N=S1NSS^{-1}N=S^{-1}N^{S}, so KerφS=0S\mathrm{Ker}\varphi_{S}=0^{S}.

  3. (NS)T=NST(N^{S})^{T}=N^{ST}, S1(S1N)=S1NS^{-1}(S^{-1}N)=S^{-1}N.

  4. Assume NPN\subset P, then NSPSN^{S}\subset P^{S} and S1NS1PS^{-1}N\subset S^{-1}P.

  5. (NP)S=NSPS(N\cap P)^{S}=N^{S}\cap P^{S} and S1(NP)=S1NS1PS^{-1}(N\cap P)=S^{-1}N\cap S^{-1}P.

  6. (N+P)SNS+PS(N+P)^{S}\supset N^{S}+P^{S} and S1(N+P)=S1N+S1PS^{-1}(N+P)=S^{-1}N+S^{-1}P.

  7. Assume STS\subset T, then NSNTN^{S}\subset N^{T}.

Proof

Proof.

  1. First for a(φS1(K))Sa\in(\varphi_{S}^{-1}(K))^{S}, sS\exists s\in S, sa/1Ksa /1\in K. Then a/1=1/ssa/1Ka /1=1/s\cdot sa /1\in K. So aφS1(K)a\in \varphi_{S}^{-1}(K).
    Second, KS1(φS1(K))K\subset S^{-1}(\varphi_{S}^{-1}(K)) is obvious. For any a/sS1(φS1(K))a /s\in S^{-1}(\varphi_{S}^{-1}(K)), a/1Ka/s=1/sa/1Ka /1\in K\Rightarrow a /s=1 /s\cdot a /1\in K. So S1(φS1K)KS^{-1}(\varphi_{S}^{-1}K)\subset K.

  2. First, for any aNSa\in N^{S}, sSs.t.saN\exists s\in S s.t. sa \in N. So sa/1S1Nsa /1\in S^{-1}N, a/1=1/ssa/1S1Na /1=1 /s\cdot sa /1\in S^{-1}N. Thus aφS1(S1N)a\in \varphi_{S}^{-1}(S^{-1}N).
    Conversely, aφS1S1N\forall a\in \varphi_{S}^{-1}S^{-1}N, a/1S1Na /1\in S^{-1}N, so t(sab)=0t(sa-b)=0 for some bNb\in N, s,tSs,t\in S. Thus sta=tbNsta=tb\in N, aNSa\in N^{S}. Immediately, we have KerφS=0S\mathrm{Ker}\varphi_{S}=0^{S}.
    Second, S1NS1NSS^{-1}N\subset S^{-1}N^{S} is obvious. a/tS1NS\forall a /t\in S^{-1}N^{S}, saNsa\in N for some sSs\in S. Then a/t=sa/tsS1Na /t=sa /ts\in S^{-1}N.

  3. First just note a(NS)Ta\in (N^{S})^{T} if and only if tt\exists t\in t, sSs\in S s.t. s(tn)=(st)nNs(tn)=(st)n\in N.
    Second one is obvious.

  4. Obvious.

  5. First, it’s easy to see (NP)SNSPS(N\cap P)^{S}\subset N^{S}\cap P^{S}. Converselym if aNSPSa\in N^{S}\cap P^{S}, s,tS\exists s,t\in S, saNsa\in N, taPta\in P, so staNPsta\in N\cap P. Thus a(NP)Sa\in (N\cap P)^{S}.
    Second, NPN,PN\cap P\subset N, P, so S1(NP)=S1NS1PS^{-1}(N\cap P)=\subset S^{-1}N\cap S^{-1}P. Conversely, if a/s=b/tS1NS1Pa /s=b/ t\in S^{-1}N\cap S^{-1}P with aNa\in N, bPb\in P, s,tSs,t\in S.Then wS\exists w\in S s.t. wta=wsbNPwta=wsb\in N\cap P. Thus wta/wts=a/sS1(NP)wta /wts=a /s\in S^{-1}(N\cap P).

  6. First, for aNSa\in N^{S}, bPSb\in P^{S}, s,tS\exists s,t\in S s.t. saNsa\in N, tbPtb\in P, so st(a+b)N+Pst(a+b)\in N+P. a+b(N+P)Sa+b\in (N+P)^{S}.
    Second, N,PN+PN,P\subset N+P, so S1(N+P)S1N+S1PS^{-1}(N+P)\supset S^{-1}N+S^{-1}P. The converse is obvious.

  7. Obvious. ◻

Theorem 6. The functor S1()S^{-1}(-) is exact.

Proof

Proof. For MfMgMM'\xrightarrow{f} M\xrightarrow{g}M'' exact at MM, it’s easy to see S1gS1f=0S^{-1}g\circ S^{-1}f=0. So ImS1fKerS1g\mathrm{Im}S^{-1}f\subset \mathrm{Ker}S^{-1}g.

Conversely, for a/sKerS1ga /s\in \mathrm{Ker}S^{-1}g, g(a)/s=0g(a) /s=0 so there is tSt\in S s.t. tg(a)=g(ta)=0tg(a)=g(ta)=0. So taKerg=Imfta\in \mathrm{Ker}g=\mathrm{Im}f. ta=f(b)ta=f(b) for some bMb\in M'. Hence a/s=S1f(b/st)ImS1fa /s=S^{-1}f(b /st)\in \mathrm{Im}S^{-1}f. Thus KerS1g=ImS1f\mathrm{Ker}S^{-1}g=\mathrm{Im}S^{-1}f. ◻

Corollary 4. S1RS^{-1}R is flat over RR.

Proof

Proof. Notice that S1RRMS1MS^{-1}R\otimes_{R}M\cong S^{-1}M. ◻

Corollary 5. Let MM be an RR-module. Then S1(M/aM)S1M/S1(aM)S1M/aS1MS^{-1}(M /\mathfrak{a}M)\cong S^{-1}M /S^{-1}(\mathfrak{a}M)\cong S^{-1}M /\mathfrak{a}S^{-1}M.

Proof

Proof. It’s easy to see S1(aM)=aS1MS^{-1}(\mathfrak{a}M)=\mathfrak{a}S^{-1}M. To show the first equation, just apply S1()S^{-1}(-) to 0aMMM/aM0\to \mathfrak{a}M\to M\to M/\mathfrak{a}M. ◻

评论