First we give the universal property of localization.
Theorem 1. Let R be a ring, S is a multiplicative subset. For any homomorphism ψ:R→R′, ψ(S)⊂R′× if and only if there exists a ring homomorphism ρ:S−1R→R′ s.t. ρφS=ψ, where φS is the canonical homomorphism: φS:R→S−1R.
If so ,ρ is unique and Kerρ=KerψS−1R.
Proof
Proof. If ρ exists, for s∈S, ψ(s)=ρ(s/1) so ψ(s)ρ(1/s)=ρ(1)=1. ψ(S)⊂R′×.
Conversely, if ψ(S)⊂R′×, take ρ(x/s)=ψ(x)ψ(s)−1. It’s easy to check ρ is well-defined homomorphism. And ∀x∈R, ρφS(x)=ρ(x/1)=ψ(x)ψ(1)−1=ψ(x).
Lastly, Kerρ⊃KerψS−1R is trivial. Conversely, ∀x/s∈Kerρ, ψ(x)ψ(s)−1=0⇒ψ(x)=0. So x/s∈KerψS−1R. ◻
Corollary 1. Let R be a ring, S be a multiplicative subset. The canonical homomorphism φS:R→S−1R is an isomorphism if and only if S consists of units.
Proof
Proof. Omitted. ◻
We can localize a ring at any element f∈R.
Definition 1. Let R be a ring, Sf={fn∣n≥0}. Then we call Sf−1R the localization of R at f, denoted by Rf.
Localization at an element can be easily charactorized.
Theorem 2. Let f∈R, X is an indeterminate, then Rf≅R[X]/(1−fX).
Proof
Proof. We use the universal property to prove it. Let x∈R[X]/(1−fX) be the image of X. Then 1−φ(f)X=0. So φ(f) is a unit. φ(Sf)⊂(R[X]/(1−fX))×. Let any ψ:R→R′ with ψ(f) a unit in R′. Define θ:R[X]→R′ by θ∣R=ψ and θ(X)=ψ(f)−1. Then θ(1−fX)=0, (1−fX)⊂Kerθ, we have unique homomorphism ρ:R[X]/(1−fX)→R′ s.t. ψ=ρφ. Thus R[X]/(1−fX) satisfies the universal property, i.e. R[X]/(1−fX)≅Rf. ◻
Proposition 1. Let R be a ring, S multiplicative subset, a an ideal of R.
aS−1R={a/s∣a∈a,s∈S}.
a∩S=∅ if and only if aS−1R=S−1R if and only if φS−1(aS−1R)=R.
Proof
Proof.
Let N={a/s∣a∈a,s∈S}. Obviously, N⊂aS−1R. Conversely, for a,b∈a, x/s,y/t∈S−1R. Then φ(a)x/s+φ(b)y/t=ax/s+by/t=(axt+bys)/st, so aS−1R⊂N.
If s∈a∩S, then 1S−1R=s/s∈aS−1R, so aS−1R=S−1R. φS−1(aS−1R)=R. Conversely, if φS−1(aS−1R)=R, then 1∈aS−1R. So ∃a∈a, s∈S s.t. a/s=1, so ∃t∈S with at=st. Since at∈a, st∈S, a∩S=∅. ◻
To describe the ideals of S−1R more precisely, we use the saturation: aS={a∈R∣∃s∈S s.t. as∈a}, i.e. aS=∩s∈S(a:s). And it’s easy to show KerφS=(0)S and a⊂aS. aS is an ideal.
Proposition 2. Let R be a ring, S a multiplicative subset,
Let b be an ideal of S−1R, then φS−1(b)=(φS−1(b))S, b=φS−1(b)S−1R.
Let a be an ideal of R, then aS−1R=aSS−1R, φS−1(aS−1R)=aS.
Let p be a prime ideal of R, p∩S=∅, then p=pS and pS−1R is prime in S−1R.
Proof
Proof.
First, ∀a∈(φS−1(b))S, ∃s∈S s.t. as∈φS(b). Then as/1∈b, so a/1∈b. Thus a=φS−1(a/1)∈φS−1(b)⇒φS−1(b)=φS−1(b). Second, ∀a/s∈b, a/1∈b. So a∈φS−1(b)⇒a/s∈(φS−1(b))S−1R. The converse is obvious.
First, ∀a∈aS, ∃s∈S, as∈a so a/1=as/1⋅1/s∈aS−1R. Thus aSS−1R⊂aS−1R. So aSS−1R=aS−1R. Second, φS−1(aS−1R)=φS−1(aSS−1R)⊃aS. Conversely, ∀x∈φS−1(aSS−1R), x/1∈aSS−1R=aS−1R. So x/1=a/s for a∈a, s∈S. Thus (xs−a)t=0 for some t∈S⇒x∈aS.
First, if sa∈p for s∈S, s∈/p⇒a∈p. So pS=p. Second, for any a/s⋅b/t∈pS−1R, we have ab∈φS−1(pS−1R)=pS=p. So a∈p or b∈p. Then a/s∈S−1R or b/t∈pS−1R. Hence pS−1R is prime. ◻
From the above theorem, we can conclude that φS gives a bijection between ideals a of R with a=aS and ideals of S−1R.
For the prime ideals p∩S=∅, we have p=pS so there is bijection between prime ideal p∩S=∅ and prime ideals of S−1R.
Definition 2. Let R be a ring, p a prime ideal of R. Let Sp=R−p. We call the ring Sp−1R the localization of R at p, and denoted by Rp.
Rp is the local ring with the maximal ideal pRp.
Theorem 3. Let R be a ring, S multiplicative subset, Let T′ be a multiplicative subset of S−1R, T=φS−1(T′). Assume S⊂T, then (T′)−1(S−1R)=T−1R.
Proof
Proof. Idea: use the universal property. Omitted. ◻
Corollary 2. Let R be a ring, p⊂q prime ideals. Then Rp is the localization of Rq at the prime ideal pRq.
Proof
Proof. Let S=R−q, T′=Rq−pRq. Set T=φS−1(T′). Then T=R−p. By above theorem, we have Rp the localization of Rq at pRq. ◻
Localization of Module
Theorem 4. Let R be a ring, S multiplicative subset. Then M has a compatible S−1R-module structure if and only if ∀s∈S, the homomorphism μs:m↦sm is an isomorphism. If so , the S−1R-module structure is unique.
Proof
Proof. Assume M has a compatible S−1R-module structure, take s∈S. Then μs=μs/1. So μs⋅μ1/s=1. Similarly, μ1/s⋅μs=1. μs is bijective.
Conversely,if μs is bijective for all s∈S. Then μ:R→End(M) given by μ:s↦μs sends S to units of End(M). Hence there’s unique MS−1R:S−1R→End(M), there’s unique compatible S−1R-module structure on M. ◻
Now we give the universal property of localized modules.
Theorem 5. Let R be a ring. S a multiplicative subset, M an R-module. Then for any S−1R-module N with R-module homomorphism ψ:M→N, there’s a S−1R-module homomorphism: ρ:S−1M→N s.t. ρφS=ψ.
Proof
Proof. Similar as the analogy in ring. ◻
We can show that S−1(−) is actually a functor. Moreover, it’s the left adjoint to the functor of restriction of scalars, i.e. for R-module M and S−1R-module N, we can regard N as an R-module. Then HomS−1R(S−1M,N)≅HomR(M,N) by universal property.
Corollary 3. Let R be a ring, S be a multiplicative subset. Then S−1(−) and S−1R⊗R− are naturally isomorphic.
Similar as the localization of rings, we use the saturation of a submodule to charactorize φS−1. For N⊂M be R-modules, the saturation NS={m∈M∣∃s s.t. ms∈N}.
Proposition 3. Let R be a ring, M be an R-module, N, P submodules. Let S,T be multiplicative subsets, K an S−1R-submodule of S−1M. Then
φS−1(K)=(φS−1(K))S and K=S−1(φS−1(K)).
φS−1(S−1N)=NS, S−1N=S−1NS, so KerφS=0S.
(NS)T=NST, S−1(S−1N)=S−1N.
Assume N⊂P, then NS⊂PS and S−1N⊂S−1P.
(N∩P)S=NS∩PS and S−1(N∩P)=S−1N∩S−1P.
(N+P)S⊃NS+PS and S−1(N+P)=S−1N+S−1P.
Assume S⊂T, then NS⊂NT.
Proof
Proof.
First for a∈(φS−1(K))S, ∃s∈S, sa/1∈K. Then a/1=1/s⋅sa/1∈K. So a∈φS−1(K). Second, K⊂S−1(φS−1(K)) is obvious. For any a/s∈S−1(φS−1(K)), a/1∈K⇒a/s=1/s⋅a/1∈K. So S−1(φS−1K)⊂K.
First, for any a∈NS, ∃s∈Ss.t.sa∈N. So sa/1∈S−1N, a/1=1/s⋅sa/1∈S−1N. Thus a∈φS−1(S−1N). Conversely, ∀a∈φS−1S−1N, a/1∈S−1N, so t(sa−b)=0 for some b∈N, s,t∈S. Thus sta=tb∈N, a∈NS. Immediately, we have KerφS=0S. Second, S−1N⊂S−1NS is obvious. ∀a/t∈S−1NS, sa∈N for some s∈S. Then a/t=sa/ts∈S−1N.
First just note a∈(NS)T if and only if ∃t∈t, s∈S s.t. s(tn)=(st)n∈N. Second one is obvious.
Obvious.
First, it’s easy to see (N∩P)S⊂NS∩PS. Converselym if a∈NS∩PS, ∃s,t∈S, sa∈N, ta∈P, so sta∈N∩P. Thus a∈(N∩P)S. Second, N∩P⊂N,P, so S−1(N∩P)=⊂S−1N∩S−1P. Conversely, if a/s=b/t∈S−1N∩S−1P with a∈N, b∈P, s,t∈S.Then ∃w∈S s.t. wta=wsb∈N∩P. Thus wta/wts=a/s∈S−1(N∩P).
First, for a∈NS, b∈PS, ∃s,t∈S s.t. sa∈N, tb∈P, so st(a+b)∈N+P. a+b∈(N+P)S. Second, N,P⊂N+P, so S−1(N+P)⊃S−1N+S−1P. The converse is obvious.
Obvious. ◻
Theorem 6. The functor S−1(−) is exact.
Proof
Proof. For M′fMgM′′ exact at M, it’s easy to see S−1g∘S−1f=0. So ImS−1f⊂KerS−1g.
Conversely, for a/s∈KerS−1g, g(a)/s=0 so there is t∈S s.t. tg(a)=g(ta)=0. So ta∈Kerg=Imf. ta=f(b) for some b∈M′. Hence a/s=S−1f(b/st)∈ImS−1f. Thus KerS−1g=ImS−1f. ◻
Corollary 4. S−1R is flat over R.
Proof
Proof. Notice that S−1R⊗RM≅S−1M. ◻
Corollary 5. Let M be an R-module. Then S−1(M/aM)≅S−1M/S−1(aM)≅S−1M/aS−1M.
Proof
Proof. It’s easy to see S−1(aM)=aS−1M. To show the first equation, just apply S−1(−) to 0→aM→M→M/aM. ◻