You are on page 1of 20
Homomorphisms Let G; and G» be groups. A surjective function f : G; —> Gy is said to be a homomorphism iff f(z - y) = f(x) - f(y) for all 2 and y in Gy. This last equation will be called the Homomorphism Test Equation and abbreviated HTE. The HTE is usually misunderstood by beginners. The group operator on the left of the equation is not the same as the group operator on the right. On the left we have f(xy). Since x and yare in Gj, this operator must be the operator in G;. On the right we have f(z) - f(y). Since both f(z) and f(y) are in G2, this operator must be the G» operator. A homomorphism not only converts the elements of G; into elements of Gz but must also convert the Gi operator into the G» operator. A graphic might be instructive f Gi @ The question mark in the diagram has to be f(x) - f(y) if f is a homomorphism. We say that group homomorphisms "preserve multiplication". As welll see, that's not all they "preserve" Theorem: If f : G, — Gis a group homomorphism then f(e;) = e2 where ey is the identity of G, and e is the identity of Gy Proof: Let x be an element of G). f(z) is an element of G2. Consider F(x-e1) = f(z): Fler) However, f(a - 1) also must equal f(x) since e, is the identity of G, J (2) = f(#)- Fler) Since ey is the identity of Gz and f(r) f(x) e2 = f(z): Fler) (by RHC) e2 = f(ei) inG: QED omomerhisms ni es Paget ‘Now we know that homomorphisms "preserve" identities. We also know that a chart like: Ds Zz e (0) o] 2 2 ab ab’ homomorphisms "link" identities. Theorem: If f ; G; — G2 isa group homomorphism then for every x in Gi, fla") = (f@)" Consider: f(x +2) = f(e:) = e2 However, f(z- 27) also equals f(a) - f(x") Therefore, f(x) - f(x) =e fe") = (F@)" QED This theorem can also be quite misunderstood. Consider the equation f(a) = (f(z). The inverse on the left is the Gy inverse of the element in Gi. The inverse on the right is the Gy inverse of the element f(r) in Gz. We now know that homomorphisms "preserve" inverses. Consider this diagram: Homomorphisms oe ExT, Page? This function cannot be a homomorphism. Let's consider one reason it can't because of our new theorem. f(i) = 6. Therefore Ff (é"') should equal (f@)" = 6". since ¢! = i and since b"! = b, we now know that ifé and d are linked then -i and 8% would be linked iff were a homomorphism. Thus f is not a homomorphism, Suppose f : Gi — Gy is a group homomorphism. Suppose an element 2 in Gis SclFinvertible. Suppose f(x) = y in Gs. Is it possible that y ix not self-invertible? Suppose y # v. We know fe) =y °. f(a) =y", However on Goes ey ue ae namie male the image of a selfinvertible element under a homomorphism is also ¢ self-invertible element, ‘A warning, however, is in order at this point. Suppose an element yin Gis self. invertible. If isa pre-image of y, «need not be selRinvertible, ‘The equations f(a") = (f(a) =y! = 1 do imply that the inverse of z in @, has to also be a pre- image of y. The following is a homomorphism Ds Kr e————> a P b a q ¥ ab r ab? a ab? 1. fle)=e 2, Since p,qand r are self-invertible, whenever an element is a pre-image of one of them its inverse is also a pre-image of the same one. Note that’ Definition: Let f : G: —* Go be a group homomorphism. The set of all elements in Gy that are pre-images of e (the identity of G) is called the kernel of the homomorphism and denoted Ky ie. Ky ={x| a € Gy and f(z) a) In our example of a homomorphism that linked D, and Ky, Kery = {e,b*} The following is a homomorphism: Homomorphisms Tay Howe, — k -k For this homomorphism, Kery = {1,-1, j,-J} Theorem: Let f : G; — G)be a group homomorphism. Ker, is a normal subgroup of G Proof: Let's first establish that Ker, is a subgroup of G1 Closure Let x € Kers Let y € Kery F(x-y) = F(z): fly) =e2-e2 = er c.e-ye Kery Inverse Closure Let x € Kery fle") = (F@)" =e" =e wal € Kery Now we know that Ker, is a subgroup of G,. Let's establish normalicy. Homomorphisms a Pages NTC Let x € Kery. Let q be an arbitrary element of G,. S(geq") = §(g)- Fle) f(g") = fq) -e2 f(g") ££") = 1) FM)" =e “gag! € Ker; KeryaGi QED Notice that in our proof we used the fact that f(qzq) = f(q")f(z)f(q). Itis easy to establish that the HTE can be extended to an arbitrary number of elements in a product, The following cannot be a homomorphism: Zs Ke 0] ————> e U} p 2} BI 4 14] Ca [5] E — m Ifit were, Kery = {[0], [3]}. However, {[0], [3]} is not a normal subgroup of 4s. Itisn't even a subgroup. Can you find other reasons that the above function could not be a homomorphism? Homo Tory Hove, Sao Our theorem limits the number of different homomorphisms for which a given group G can be the domain, The number is limited by the number of normal subgroups contained in G. For now, we do not know whether or not each normal subgroup of G can be a kernel for a homomorphism with G as the domain, We also don't know whether or not a given normal subgroup can generate more than one homomorphism. ‘Two rather trivial homomorphisms exist for any group G. The first is usually called the identity function and is defined by f(x) = z for every x in G. For example: Note that the kernel for the identity function for any group is {e}. To prove that the identity function is a homomorphism, consider: F (zy) = zy = f(x) f(y) for any x, y EG (ey) = F(x) f(y) A second trivial homomorphism that can be defined on any group G uses the group Z ; as the range group. 2 ; is the group with only one element [0] whose operating table is: 21 | (0) fo] | [0] For any group G, define the function f(z) = {0] for every x in G. For example Laity Toe, Cam Page? To confirm that F is a homomorphism, consider: F(@-y) = [0] = {0} - [0] = Fle) Fw) +. f(y) = f(z) fu) Note that the kernel of this homomorphism is G itself. Suppose G; is a finite group and o(G,) =n. Let f : G: > Gy be a function where G2 is also a finite group. In order to determine whether or not f is a homomorphism, we need to establish that f(y) = f (x) f(y) for every « and y in Gy How many equations would we need to consider to establish that f is a homomorphism by exhausting every possibility for x and y? We know can be any of n elements. Similarly, can be any of'n elements. Therefore, n? equations would have to be satisfied where each is of the type f(y) = f(x) f(u) Theorem: Let f : G; — G2 be a group homomorphism. Let Kery = K I(x) = fly) if Ka = Ky. Proof: Suppose that Ka = Ky. Since x = e;x and e, € K, we know that x € Kz. Since Kx = Ky, we can deduce that x € Ky. . x= hry for some ky € K. F(a) = F(kiy) = Fa) Fy) = ef) = FQ) f(z) = Fly). Now suppose that Kx # Ky. We know that if two cosets are not equal, then they are disjoint, Could f(x) = f(y)? Suppose so. Homomorphis Tay Fone, ae Pages Consider: F(ey') = f@)- FO") = F@UEWY' = FOE@" = e2 ayleK Let zy! =k c= ky weKy We know that x € Kx. °, Kx and Ky are not disjoint anes QED This theorem gives us a great amount of insight concerning the structure of a homomorphism. The set of all pre-images of a specific range element is precisely one of the cosets generated by the kernel of the homomorphism. Let's consider again the function taking D, to Ky that we itemized earlier in this chapter We claimed that this function was a homomorphism. In order to verify this we could consider all 64 equations of the form f(xy) = f(r) - f(y) where x and y are arbitrary elements of Dy, It is unnecessary for you to do this at this time, but all 64 equations are true. We should take a closer look at this function, The kernel of f is {e,b°} which indeed is a normal subgroup of Dy which illustrates one of our important theorems. Consider the cosets generated by the kernel {e, b?} which we will denote K. Homomorphisms Lary Tove. Rea Page Ke ={e,b"} Ka = {a,ab"} Kb = {b,b°} Kb = {Be} Kb! = {b8,b} Kab = {ab, ab*} Kab? = {ab’,a} Kab’ = {ab*, ab} The distinct cosets are {e,b?}, {a, ab*}, {b,%} and {ab, ab?}. The pre-image of each of the elements of Kis precisely one of these cosets. This is an illustration of another of our important theorems. It becomes clear that homomorphisms are highly structured. We will be interested in finding a way to resolve one key issue: What groups can a given group be linked to with a homomorphism where the given group is the domain? As we will see, the answer lies within the structure of the given group. Before we attack this issue, let's solve some example problems. Problem 1 Let G, be the group of all non-singular 2 x 2 matrices under matrix multiplication, Let Gy be the group of non-zero real numbers under the usual real number multiplication, Let f : G, — G2 be defined by s(e 3) = ad — be [Note that f is essentially the act of computing a determinant]. Is f a homomorphism? If so, what is the kernel of f? Solution: Consider first ab\ (ef)\_ ((24)-(G4))- ae+bg af +bh\ _ (ere cf + ny ~ (ae + bg)(cf + dh) — (af + bh)(ce + dg) = accf + aedh + bgcf + badh — afce — afdg — bhce ~ bhdg = aedh + bgcf — afdg — bhce (Result 1) Now Consider i ) A ) = (ad — be) - (eh — fg) = adeh — adfg— beeh + befg (Result 2) ‘A comparison of results 1 and 2 reveal them to be equal. Therefore f is a homomorphism. Now we can consider the kernel of f. ‘The identity of the range group is 1. In order for a 2 x 2 matrix to be in the kernel, its image under f must be 1, The kernel is therefore the set of all 2 x 2 matrices whose determinants are 1. We now know that this Homomorphi Try Howe, amen Page To set is a normal subgroup of the group of all 2 2 non-singular matrices. This result can be established directly by considering closure, inverse closure and the NTC. Problem 2 Let Gy be the group of all positive real numbers under usual multiplication. Let Gp be the group of all reals under usual addition. Let f : G, —> G be defined by f(a) = Enz. Is f ahomomorphism? If'so, what is the kernel of f? Solution: Consider f(x - y) = én(ay) fn + try = f(z) +f) Since multiplication is the operator for G, and addition is the operator for Gy, we have shown that the group theoretic equation f(e#y) = f(x) f(y) has been satisfied. Therefore f is a homomorphism. The kernel of f is the set of pre-images of the identity of G» which is 0. Therefore if x € Kery then f(x) =0 = tnx Seine =r=1 ». Kery = {1} An injective (one-to-one) homomorphism is called an isomorphism. The domain group and the range group of such a function are said to be isomorphic. Not only is there a one-to-one correspondence of their elements, but there is a smooth transition from the domain operator to the range operator. As such, the groups are considered to be structurally the same but with (possibly) different labels on the elements. Two of the groups you have in your library are isomorphic. Consider: (23). ? (123) ab (132) ab? Homomorphisn ary How Paget Clearly f is injective. To establish that f is a homomorphism by exhaustion requires confirming 36 equations. We must note that the fact that S3 and Ds are isomorphic does not imply that other members of the Symmetric family are isomorphic to members of the Dihedral family. Indeed, S, is not isomorphic to D2. We could have gotten clues along the way that Ss and D3 were very similar. If you check your previous work, you will find that they both have the same number of subgroups. The orders of the subgroups match. The number of subgroups that are normal match. The number of order 2 and order 3 elements match. A final word about isomorphisms is in order. Given the set of all groups, the relation "is isomorphic to” is an equivalence relation. Theorem: Let f : G, — G bea group homomorphism. Let K’ be the kernel of f G)/K is isomorphic to Gz Proof: Let : G,/K — G, be defined by A(K'x) = f(z). Is h well defined? Suppose Ke =Ky, Then € Ky. ., 3k such that 2 = hry. . F(a) = Fk) = f(a) FW) = e2Fly) = Fv) Since f(x) = f(y), h(Kz) = h(icy). «his well defined Is h a homomorphism? Recall that in the last chapter we proved that Ka- Ky = Kzy. Using this, consider h(Ka- Ky) = h(Kzy) ‘ (zy) = f(e)f(y) = h(Ka)h(Ky) ©. his a homomorphism. Is h injective? Suppose (Kx) = h(Ky). Then we know f(x) = f(y). Consider: F(zy") = f(@)fw") = faim? = Fa)lf@)" =e say eK woe Ky Kr=Ky Therefore, if A(Kx) = h(Ky) it follows that Kx = Ky. This implies that h is injective QED Now we can find all groups to which a given group can be homomorphically linked. To achieve our goal we must 1, Find all normal subgroups of the given group 2. Factor the given group by each normal subgroup 3. Identify the factor groups Homomorphisms Tay Howe, Rowan Universi Page T2 Clearly f is injective. To establish that f is a homomorphism by exhaustion requires confirming 36 equations. We must note that the fact that Ss and D3 are isomorphic does not imply that other members of the Symmetric family are isomorphic to members of the Dihedral family Indeed, Sy is not isomorphic to. D2. We could have gotten clues along the way that Ss and D3 were very similar. If you check your previous ‘work, you will find that they both have the same number of subgroups. The orders of the subgroups match. The number of subgroups that are normal match. The number of order 2 and order 3 elements match. ‘A final word about isomorphisms is in order. Given the set of all groups, the relation “is isomorphic to" is an equivalence relation, a Theorem: Let f : Gy — Gz be a group homomorphism. Let K’ be the kernel of . G,/K is isomorphic to Gz Proof: Let A: Gi/K — Gz be defined by h(K=) = f(z). Is h well defined? Suppose Kr= Ky. Thenzé Ky. ©. 3 kysuch thats = ky. «- f(@) = f(b) = fle) FW) = cafe) = Fu) Since f(x) = f(y), h(Kz) = (Ky). e his well defined Is h. a homomorphism? Recall that in the last chapter we proved that Kz- Ky = Kzy. Using this, consider: A(Kz Ky) = hay) : = fav), . = F210) = h(Kz)h(Ky) A is a homomorphism. (Ky). Then we know f(x) = f(y). F(z)fr) a @ror =e oayleK vee Ky Kr=Ky ‘Therefore, if h(Kz) = h(Ky) it follows that Kz = Ky. This implies that A is injective QED Is h injective? Suppose h(K2) Consider: fly") = Now we can find all groups to which a given group can be homomorphically linked. To achieve our goal we must 1, Find all normal subgroups of the given group 2. Factor the given group by each normal subgroup 3. Identify the factor groups 4, Create the homomorphisms Example 1: Find all groups to which Qs can be homomorphically linked. Solution: The normal subgroups of Qs are : {1}, Qs, {AyD}, (1-1, 5)-a} and (1, -1, by -k} Qs/{1}is an isomorphic copy of Qs. .°. Qs can be homomorphically linked to Qs: The identity function will be the homomorphism (which is actually an isomorphism), Qs/ Qs has only one distinct coset. , Qs can be homomorphically linked to # The function needed is: Homomorphisms Tay owe, er RC zl aN Os i i k “k Gi (0] {1 Qs/{1,-1} has four distinct cosets: {1,1}, {i,-i}, {J,-7} and (k, -k}. Letting {1,-1} equal K, these correspond to K, Ki, Kj, and Kk. A table for Qs/K is Q/K | K_| Ki_| Kj | Kk K K_| Ki | Kj | kk Ki_ | Ki | K [Kk | Kj Kj_| Ki |Kk|K |Ki Kk | Kk | Ki | Ki | K This group is an isomorphic copy of Ky. To homomorphically link Qs and Ks, we could use this function: Homomorphisms Tany Tow, Scnea It is important to observe that each of the pre-images of elements in Kis @ coset induced by factoring Qs by {1,-1}. Is the function above the only one possible? We could have made i and ~i the pre-images of any one of the three elements p, q, 7. Once one of these twas selected, j and -j could have had either of the two remaining elements of Ks as their image. Using combinatorics, there are 6 different possible homomorphic linkages of Qs and Ky {. -Li. Qs/{is-1, 4, -i} isa element group. Since 2 is @ prime number, the only group of order 2 is Z 2. The function we use to create the homomorphic link is: Homomor Gary Hove, asin ra {Usk ici} and {1-1 =k} These two normal subgroups can also be used as kernels for homomorphisms between Qs and Z >. Therefore, Qs can be linked homomorphically to # » in three different ways, To sum up, we know that Qs has only #1, 2 2, Ky and Qeas homomorphic images. Example 2 Find all groups that are homomorphic images of Ds. Solution: Using the trivial subgroups {e} and Ds, we can link Dg to Dg and 4 1 respectively. Using any of the 6 element subgroups (they all have to be normal), we can link Dg to Z ». For example, using the 6 element subgroup {e, a, 6”, b', ab*, ab*} we can create: De e a b e {0} Ye wf , ab ab? ron] ab? ‘ ab’ ab? The only subgroup of order 3 in Dg is {e, b?, b}. Therefore this subgroup has to be normal. The cosets induced by this subgroup are: {e, b?,b*}, {a, ab®, ab*}, {b,d°, b°} and {ab,ab*,ab"}. If we call these K, Ka, Kb and Kab respectively, we can obtain the operating table. Kab This is an isomorphic copy of Ky. One possible homomorphism is Homomorphisns Lary Howe, Raa Page? None of the 4 clement subgroups of Dg is normal. Therefore, Dg can not be homomorphically linked to 5. However, {e, 6°} is a2 element normal subgroup. It is the center of Dg. The cosets it generates are {¢, 6}, {a, ab}, {b,b*}, {82,5}, {ab, ab*} and {ab?, ab°} . We'll call these respectively K, Ka, K6, Kb’, Kaband Kab. An operating table is: Homomorphisms Tiny Howe, Remeeby Page Ts Kab? K Ka Kab? Ka K Kab? | Kb | Kv _ Kb | Kab® K_ | Ka | Kab Ke | Ke | Kab | K | Kb | Kab?| Ka Kab | Kab | Kb? | Kab? | Ka | K | Kb Kab? | Kab? | Kb | Ka | Kab| KB | K TI pertinent homomorphism is: e a b ab ab abe ab! s abi is an isomorphic copy of Ds Therefore, Ds can be linked to Ds. The — 2 ‘ab oo o Therfore, Ds can have the following homomorphic images: #1, Ds, # 2, Kyand Ds Homomorphisms Tay Hove, acetae Page 19

You might also like