Rings and modules Notation: A ⊂ B means A isasubsetof B, possiblyequalto B. 1. Revision Allringsarecommutativeringswithunity. 1.1. Let f:A → B bearinghomomorphism. Theorem on ring homomorphisms. The kernel I of f is an ideal of A, the image C of f isasubringof B. Thequotientring A/I isisomorphicto C. Proof. Considerthemap g:A/I → C, a+I (cid:55)→ f(a). Itiswelldefined: a+I = a(cid:48)+I implies a−a(cid:48) ∈ I implies f(a) = f(a(cid:48)). The element a+I belongs to the kernel of g iff g(a+I) = f(a) = 0, i.e. a ∈ I, i.e. a+I = I isthezeroelementof A/I. Thus, ker(g) = 0. Theimageof g is g(A/I) = {f(a) : a ∈ A} = C. Thus, g isanisomorphism. Theinversemorphismto g isgivenby f(a) (cid:55)→ a+I. Correspondence theorem. Let I be an ideal of a ring A. Then there is a bijection betweenthesetof allideals J of A suchthat I ⊂ J andthesetof allidealsof A/I: {J : I anidealof A,I ⊂ J} −→ {K : K anidealof A/I} J −→ J/I Proof. Denoteby h themorphism h:A → A/I, a (cid:55)→ a+I, itsimageis A/I andits kernelis I. For an ideal J of A, I ⊂ J, denote by h| :J → A/I,j (cid:55)→ j +I the restriction J of h to J. Its kernel is I. Similarly to the proof of the previous theorem we deduce that h| (J) isisomorphicto J/I whichisanidealof A/I. J For an ideal K of A/I define K(cid:48) = h−1(K) of A. Then K(cid:48) is an ideal of A, I ⊂ K(cid:48). Now we have two maps, J (cid:55)→ J/I and K (cid:55)→ h−1(K). They are inverse to each other, i.e. h−1(J/I) = J and h−1(K)/I = K. Thus, there is a one-to-one correspondencebetweentheideals. 2 Ringsandmodules 1.2. The intersection of ideals of A is an ideal of A. Given a subset S of A, one canspeakabouttheminimalidealof A whichcontains S. Thisidealisequalto {a s + ···+a s : a ∈ A,s ∈ S,m (cid:62) 1}. 1 1 m m i i Oftenitiscalledtheidealgeneratedby S. Let I,J beidealsofaring A. Their sum I +J is the minimal ideal of A which contains both I and J, more explicitly I +J = {i+j : i ∈ I,j ∈ J}. Certainly, I +(J +K) = (I +J)+K. Similarlyonedefinesthesumofseveralideals (cid:80) I . k Their product IJ is the minimal ideal which contains all ij : i ∈ I,j ∈ J, more explicitly IJ = {i j + ···+i j : n (cid:62) 1,i ∈ I,j ∈ J}. 1 1 n n m m Theproductisassociative: (IJ)K = I(JK) anddistributive: (I +J)K = IK +JK. Similarlyonedefinestheproductofseveralideals I ...I . 1 n Notethat (I+J )(I+J ) istheminimalidealwhichcontainsproducts (i +j )(i + 1 2 1 1 2 j ) = (i i +i j +i j )+j j , soitiscontainedin I +J J : 2 1 2 2 1 1 2 1 2 1 2 (I +J )(I +J ) ⊂ I +J J , 1 2 1 2 buttheinverseinclusiondoesnotholdingeneral. Foranelement a of A theprincipalidealgeneratedby a is (a) = aA = {ab : b ∈ A}. In particular, (0) = {0} is the smallest ideal of A and (1) = A is the largest ideal of A. Unless A = {0}, thesearetwodistinctidealsof A. Forseveralelements a , ...,a of A theidealgeneratedbythe a isdenoted 1 n i (a , ...,a ) = a A+ ···+a A = {a b + ···+a b : b ∈ A}. 1 n 1 n 1 1 n n i 1.3. Aring A isafieldifitcontainsanon-zeroelementandeverynon-zeroelement of A isinvertiblein A. Lemma. Anon-zeroringisafieldiff ithasexactlytwodifferentideals, (0) and (1). 3 Proof. If I is a non-zero ideal of a field F, then I contains a non-zero element a. Therefore it contains aa−1 = 1 and therefore it contains 1b = b for every b in F; so I = F. Conversely, if a non-zero ring has only two distinct ideals then it is a field: for every nonzero element aA must be equal to (1), hence a multiple of a is 1 and a is invertible. An ideal I of a ring A is called maximal if I (cid:54)= A and every ideal J such that I ⊂ J ⊂ A eithercoincideswith A orwith I. By1.1thisequivalentto: thequotient ring A/I has no proper ideals. By the previous lemma this is equivalent to A/I is a field. Soweproved Lemma. I isamaximalidealof A iff A/I isafield. 1.4. Aring A isanintegraldomainif A (cid:54)= 0 andforevery a,b ∈ A ab = 0 implies a = 0 or b = 0. Example: everyfieldisanintegraldomain: ab = 0 and a (cid:54)= 0 implies b = a−1ab = 0. Z is an integral domain. More generally, every non-zero subring of an integral domainisanintegraldomain. If A isanintegraldomain,onecanformthefieldoffractions F of A as {a/b : a ∈ A,b ∈ A\{0}}. Bydefinition a/b = c/d iff ad = bc. This is an equivalence relation: if a/b = c/d and c/d = e/f then ad = bc and cf = ed so adf = bcf = bed, d(af −be) = 0. As d isnotzero, af = be. Definetworingoperations a/b+c/d = (ad+bc)/(bd) and (a/b)(c/d) = (ac)/(bd). Thezeroof F is 0/1 = 0/a foranynon-zero a. Everynonzeroelement a/b of F is invertible: if a/b (cid:54)= 0 then (a/b)−1 = b/a. Thus F isafield. Theringhomomorphism A → F, a (cid:55)→ a/1 is injective: a/1 = 0/1 implies a = 0. Thus A can be identified withthesubring A/1 of F. Then a/b canbeidentifiedwith ab−1 givingthemeaning offractiontothesymbol a/b. Thus, every integral domain is a non-zero subring of a field, and the latter is an integral domain. So the class of integral domains coincides with the class of non-zero subringsoffields. 1.5. An ideal I of a ring A is called prime if I (cid:54)= A and for every a,b ∈ A the inclusion ab ∈ I impliesthateither a ∈ I or b ∈ I. Example: everyfieldhasaprimeideal: (0). Lemma. I isaprimeidealof A iff A/I isanintegraldomain. Proof. Let I be a prime ideal of A. Let (a+I)(b+I) = 0+I, then ab ∈ I. So at leastoneof a,b isin I whichmeansthateither a+I = 0+I or b+I = 0+I. Thus, A/I isanintegraldomain. 4 Ringsandmodules Conversely,let A/I beanintegraldomain. If ab ∈ I then (a+I)(b+I) = I = 0+I, hence either a+I = I and so a ∈ I, or b+I = I and so b ∈ I. Thus, I is a prime idealof A. Example: foraprimenumber p theideal pZ isaprimeidealof Z. Thezeroideal (0) isaprimeidealof Z. Corollary. Everymaximalidealisprime. Proof. Everyfieldisanintegraldomain. Remark. In general, not every prime ideal is maximal. For instance, (0) is a prime idealof Z whichisnotmaximal. 1.6. For rings A define their product A × ···×A as the set theoretical product i 1 n endowedwiththecomponentwiseadditionandmultiplication. Chinese Remainder Theorem. Let I , ...,I be ideals of A such that I +I = A 1 n i j forevery i (cid:54)= j. Then (cid:89) A/(I1...In) (cid:39) A/Ik, a+I1...In (cid:55)→ (a+Ik)1(cid:54)k(cid:54)n. 1(cid:54)k(cid:54)n Proof. Firstlet n = 2. Then I I ⊂ I ∩I = (I ∩I )A = (I ∩I )(I +I ) ⊂ (I ∩I )I +(I ∩I )I ⊂ I I . 1 2 1 2 1 2 1 2 1 2 1 2 1 1 2 2 1 2 So I I = I ∩I . Thekernelofthehomomorphism 1 2 1 2 (cid:89) A → A/I , a (cid:55)→ (a+I ,a+I ) k 1 2 1(cid:54)k(cid:54)2 is I ∩I = I I . Itissurjective: since I +I = A, thereareelements x ∈ I ,y ∈ I 1 2 1 2 1 2 1 2 such that x + y = 1 and hence bx + ay = a + (b − a)x ∈ a + I and similarly 1 bx+ay ∈ b+I . 2 Now proceed by induction on n. Denote J = I ,J = I ...I , so J J = 1 1 2 2 n 1 2 I ...I . Since I +I = A forall k > 1, wededuceusing1.2that 1 n 1 k J +J = I +I ...I ⊃ (I +I )...(I +I ) = A, 1 2 1 2 n 1 2 1 n so J +J = A. Nowinthesamewayasinthepreviousparagraphonegets A/(J J ) (cid:39) 1 2 1 2 (cid:81) (cid:81) A/J . Bytheinductionhypothesis A/J (cid:39) A/I . Thus, 1(cid:54)k(cid:54)2 k 2 2(cid:54)k(cid:54)n k (cid:89) A/(I ...I ) (cid:39) A/I . 1 n k 1(cid:54)k(cid:54)n 5 Example. Let p bedistinctprimesand r positiveintegers. Then i i (cid:89) Z/(pr1...prnZ) (cid:39) Z/priZ. 1 n i 6 Ringsandmodules 2. Modules over rings 2.1. Let A be a ring. An abelian group M is called an A-module if there is a multiplication A×M → M suchthat a(x+y) = ax+ay,(a+b)x = ax+bx,a(bx) = (ab)x,1x = x. Examples. Everyabeliangroupisa Z-module,sotheclassofabeliangroupscoincide withtheclassof Z-modules. Everyvectorspaceoverafield F isan F-module. 2.2. A map f:M → N is called a homomorphism of A-modules if f(x + y) = f(x)+f(y) forevery x,y ∈ M and f(ax) = af(x) forevery a ∈ A, x ∈ M. Aho- momorphism f of A-modulesiscalledanisomorphismof A-modules,oralternatively an A-isomorphism,if f isbijective. 2.3. Asubgroup N ofan A-module M iscalledan A-submoduleof M if an ∈ N forevery a ∈ A,n ∈ N. Example: Submodulesofthe A-module A areidealsof A. For an A-module M and its A-submodule N define the quotient module M/N as the quotient set of cosets m + N with the natural addition and multiplication by elementsof A. Similarly to 1.1 one proves: If M,N are A-modules and f:M → N is an A-module homomorphism, then the kernel of f is a submodule of M and the image of f isasubmoduleof N, and M/ker(f) is A-isomorphicto im(f). Similarlyto1.1submodulesofthequotientmodule M/N arein1–1correspondence withsubmodulesof M containing N. Inparticular,if f:M → N isan A-modulehomomorphism,and K isasubmodule of ker(f), then f inducesan A-modulehomomorphism g:M/K → N, m+K (cid:55)→ f(m). 2.4. For A-modules M,N the intersection M ∩N is an A-module. So if M,N arecontainedinalargermodule L, onecanspeakabouttheminimal A-modulewhich containsafixedsetofelementsrelatedto M and N. Then the M +N = {m+n : m ∈ M,n ∈ N} is the minimal A-module which containsallallelementsof M and N. Define the direct sum of modules as the set theoretical product with the natural additionandmultiplicationbyelementsof A. Lemma. Let N,K be A-submodulesofan A-module M. Amap f:N⊕K → N+K, f((n,k)) = n+k isasurjective A-modulehomomorpismwhosekernelis A-isomorphic tothesubmodule N∩K. Therefore,if N∩K = {0}, N⊕K isisomorphicto N+K. 7 Proof. Clearly f is surjective. Its kernel is {(n,k) : n+k = 0}. Then n = −k ∈ N ∩K. Amap {(n,k) : n+k = 0} → N ∩K, (n,−n) (cid:55)→ n isabijection. 2.5. The submodule M generated by elements x is the minimal submodule which i containsallofthem,itconsistsoffinite A-linearcombinationsof x ; elements x ∈ M i i arecalledgeneratorsof M. Theminimalnumberofgenerators(ifitexists)of M iscalledtherankof M. M issaidtobeoffinitetypeifithasafinitenumberofgenerators. (cid:80) An A-module M is called free if M has generators x such that a x = 0 i i i implies a = 0 forall i. Thesetof x iscalledthenabasisof M. i i 2.6. Lemma. (1)Themodule An = ⊕1(cid:54)i(cid:54)nA isfreeof rank n. (2)Let M bean A-moduleof finitetypeandlet x , ...,x begeneratorsof M. 1 n Defineahomomorphism (cid:88) f:An → M, (a , ...,a ) (cid:55)→ a x . 1 n i i Itissurjective. If N isthekernelof f, then M isisomorphictothequotientmodule (A)n/N. Thus, every A-module of finite type is isomorphic to a quotient of a free module. (3)Everyfreemoduleof finiterank n isisomorphicto An. Proof. (1),(2)followfromthedefinitions. If M isfreeandthenumberofgenerators isfiniteequalto n, thenthehomomorphism (A)n → M issurjectiveandinjective. Elementsof N serveasrelationsforgeneratorsof M. Asacorollarywededucethatthedirectsumoffreemodulesisfree: An ⊕Am (cid:39) An+m. Examples. 1. Fromlinearalgebraitisknownthateverymoduleoffiniterankovera fieldhasabasisandisfree. 2. Let A = Z and M = Z/nZ for n > 1. Then M has rank 1 and is not a free A-module,sinceif M (cid:39) (Z)1 then M wouldhavebeeninfinite. 3. Polylinear constructions 3.1. Thesetof A-modulehomomorphismsfroman A-module M to N isan A-module: (af)(m) = a·f(m), (f +g)(m) = f(m)+g(m). Itisdenoted Hom (M,N). A 8 Ringsandmodules Example. Let A = F be a field, and let M be an F-vector space of dimension d 1 and N be an F-vector space of dimension d . Fix a basis {m } in M and a basis 2 i {n } in N. Let C be a matrix of order d ×d with entries in F. Define a map j 1 2 (cid:80) (cid:80) f:M → N, f( a m ) = b n where (b ,...,b ) = (a ,...,a )C. The map f i i j j 1 d2 1 d1 is an F-linear map. Conversely, every F-linear map M → N is determined by its values on {m : 1 (cid:54) i (cid:54) d }. Write f(m ) = (cid:80)c n and define C = (c ). This i 1 i ij j ij givesaninversemaptothepreviousmap. Thus, Hom (M,N) is F-isomorphictothe F F-moduleofmatricesoforder d ×d withentriesin F. 1 2 Examples–Exercises. Hom (0,N) = Hom (M,0) = 0. Hom (A,N) (cid:39) N, A A A Hom (M,N ⊕N ) (cid:39) Hom (M,N )⊕Hom (M,N ). A 1 2 A 1 A 2 3.2. Amap f:M ×N → R iscalled A-bilinearifforall m,m ,m ,n,n ,n ,a 1 2 1 2 f(m,n +n ) = f(m,n )+f(m,n ), f(m,an) = af(m,n) 1 2 1 2 f(m +m ,n) = f(m ,n)+f(m ,n), f(am,n) = af(m,n). 1 2 1 2 So for every m the map N → R,n (cid:55)→ f(m,n) is a homomorphism of A-modules andforevery n themap M → R,m (cid:55)→ f(m,n) isahomomorphismof A-modules. Note that an A-bilinear map f does not induce a homomorphism of A-modules M ⊕N → R, since f(a(m,n)) = f(am,an) = a2f(m,n) isnotequalto af(m,n) in general. Denote the set of all A-bilinear maps f:M ×N → R by Bil (M,N;R). The A latter is an A-module with respect to the sum of maps and multiplication of a map by anelementof A. Similarlyonecandefine A-n-linearmaps. Example. Let A = F be a field, and let M be an F-vector space of dimension d and N be an F-vector space of dimension d . Fix a basis {m } in M and a 1 2 i basis {n } in N. Let C be a matrix of order d ×d with entries in F. Define j 1 2 a map f:M × N → F, f(m,n) = mCn◦ where m is written as a row and n◦ as a column. The map f is an F-bilinear map. Conversely, every F-bilinear map M × N → F is determined by its values on {(m ,n ) : 1 (cid:54) i (cid:54) d ,1 (cid:54) j (cid:54) i j 1 (cid:80) (cid:80) (cid:80) d }: f( a m , b n ) = a b f(m ,n ). Now form a matrix C whose entries 2 i i j j i j i j are f(m ,n ). Thus, there is a one-to-one correspondence between bilinear maps i j M ×N → F andmatricesoforder d ×d withentriesin F. 1 2 3.3. To study A-bilinear maps from M ×N to R it is useful to introduce another A-module T andabilinearmap g:M×N → T suchthatbilinearmaps f:M×N → R are in one-to-one correspondence with homomorphisms of A-modules T → R via g. In other words, we define an isomorphism of A-modules Bil (M,N;R) (cid:39) A Hom (T,R); T doesnotdependon R butonlyon A-modules M and N. A Todefine T firstdenoteby L thefree A-modulewithabasisconsistingofelements l indexed by elements of M ×N. So an arbitrary element of L is a finite sum m,n 9 (cid:80) a l with a ∈ A, m ∈ M and n ∈ N. Let K be the A-submodule of L i mi,ni i i i generatedbyelements l −l −l , l −l −l , m1+m2,n m1,n m2,n m,n1+n2 m,n1 m,n2 l −al , l −al am,n m,n m,an m,n (forall a ∈ A, m ∈ M and n ∈ N). Denote T = L/K. The image of l in T, i.e. the coset l +K is usually m,n m,n denotedby m⊗n. Since L isgeneratedby l , themodule T isgeneratedby m⊗n, i.e. m,n (cid:8)(cid:88) (cid:9) T = a m ⊗n : a ∈ A,m ∈ M,n ∈ N . i i i i i i Thesesatisfyrelations: (m +m )⊗n = m ⊗n+m ⊗n, (am)⊗n = a(m⊗n), 1 2 1 2 m⊗(n +n ) = m⊗n +m⊗n , m⊗(an) = a(m⊗n) 1 2 1 2 (forall a ∈ A, m ∈ M and n ∈ N). The module T is denote M ⊗ N and is called the tensor product of M and N A over A. If M, N arefinitelygenerated A-modules,withgenerators m ,n , then M⊗ N i j A isafinitelygenerated A-modulewithgenerators m ⊗n . i j Wehave n⊗0 = 0(n⊗1) = 0. Nowdefineamap g:M×N → M⊗ N by (m,n) (cid:55)→ m⊗n. Itisan A-bilinear A map. 3.4. Theorem. For an A-bilinear map f:M ×N → R define f(cid:48):M ⊗ N → R A as f(cid:48)((cid:80)a m ⊗n ) = (cid:80)a f(m ,n ). It is a well defined map and it is a homomor- i i i i i i phism of A-modules. The correspondence f (cid:55)→ f(cid:48) is an isomorphism of A-modules Bil (M,N;R) and Hom (M ⊗ N,R). A A A Proof. Extend f to a homomorphism L → R defined on elements of the basis of L by l (cid:55)→ f(m,n). m,n Since f is bilinear, all generators of K are mapped to zero, so we get f(cid:48) = α(f):M ⊗ N → R, f(cid:48)((cid:80)a m ⊗n ) = (cid:80)a f(m ,n ). Themap α isahomomor- A i i i i i i phismof A-modules. Conversely, if f(cid:48):M ⊗ N → R is a homomorphism of A-modules, then define A f = β(f(cid:48)):M ×N → R as f(m,n) = f(cid:48)◦g(m,n). Then f isan A-bilinearmap. Now α◦β(f(cid:48)) = α(f(cid:48)◦g) andso α◦β(f(cid:48))((cid:80)a m ⊗n ) = (cid:80)a f(cid:48)◦g(m ,n ) = i i i i i i f(cid:48)((cid:80)a m ⊗n ). We also have β ◦α(f)(m,n) = α(f)◦g(m,n) = α(f)(m⊗n) = i i i f(m,n). Thus, α and β areisomorphisms. 10 Ringsandmodules Thus,usingthetensorproductonecanreducethestudyofbilinearmapstothestudy oflinearmaps. Example. Let A = F be a field. Let M,N be two F-vector spaces of dimensions d and d . In accordance with the previous theorem the vector space of linear maps 1 2 M ⊗N → F is isomorphic to the vector space of bilinear maps M × N → F. In F accordance with Example in 3.2 the dimension of the space Bil(M,N;F) is d d ; if 1 2 m , ...,m is a basis of M and n , ...,n is a basis of N, then every bilinear 1 d1 1 d2 map f:M ×N → F isdeterminedbyitsvalueson {(m ,n )}. i j Therefore, the dimension of the vector space Hom (M ⊗ N,F) is d d . It F F 1 2 is known from linear algebra that the dimension of a vector space V equals to the dimensionof Hom (V,F). Sothedimensionof M ⊗N is d d ; the F-vectorspace F 1 2 M ⊗ N hasabasis m ⊗n , 1 (cid:54) i (cid:54) d ,1 (cid:54) j (cid:54) d . F i j 1 2 Note that in the particular case of M = N the space N ⊗ N has dimension F equaltothesquareofthedimensionof N. Inphysics, N over F = C representsthe statevectorofaparticle,and N ⊗C N representsthestatevectorsoftwoindependent particlesofthesamekind. 3.5. Firstpropertiesofthetensorproduct: Lemma. (i) M ⊗ A (cid:39) M, A (ii) M ⊗ N (cid:39) N ⊗ M, A A (iii) (M ⊗ N)⊗ R (cid:39) M ⊗ (N ⊗ R), A A A A (iv) M ⊗ (N ⊕R) (cid:39) (M ⊗ N)⊕(M ⊗ R), A A A (v) Hom (M⊗ N,K) (cid:39) Hom (M,Hom (N,K)) (cid:39) Hom (N,Hom (M,K)). A A A A A A Proof. To prove (i) we first define an A-homomorphism f:L → M,l (cid:55)→ am m,a where L is a free A-module with a basis l , m ∈ M,a ∈ A. Then K (which m,a is the submodule of L defined as in 3.3) is in the kernel of f. So f induces an A-homomorphism g:M ⊗ A = L/K → M, m ⊗ a (cid:55)→ am. Define h:M → A M ⊗ A,m (cid:55)→ m⊗1. Then g and h areinversetoeachother. A To prove (ii) use an A-homomorphism f:M ⊗N → N ⊗M,m⊗n (cid:55)→ n⊗m whichcorrespondstoamap l (cid:55)→ n⊗m andan A-homomorphism g:N ⊗M → m,n M ⊗N,n⊗m (cid:55)→ m⊗n. f and g areinversetoeachother. Toprove(iii)use m⊗(n⊗r) (cid:55)→ (m⊗n)⊗r,(m⊗n)⊗r (cid:55)→ m⊗(n⊗r). For (iv) use m⊗(n,r) (cid:55)→ (m⊗n,m⊗r), (m ⊗n,m ⊗r) (cid:55)→ m ⊗(n,0)+ 1 2 1 m ⊗(0,r). 2 For(v)use h ∈ Hom (M⊗ N,K) (cid:55)→ h(cid:48) ∈ Hom (M,Hom (N,K)), h(cid:48)(m)(n) = A A A A h(m⊗n) and h(cid:48) ∈ Hom (M,Hom (N,K)) (cid:55)→ h ∈ Hom (M⊗ N,K), h(m⊗n) = A A A A h(cid:48)(m)(n). 3.6. Examples. (1) An⊗ Am = Anm. A