1 0 1 0 2 n IRREDUCIBLE REPRESENTATIONS OF THECPTGROUPSIN QED a J 5 B. Carballo Pe´rez1, M.Socolovsky2 ] h InstitutodeCienciasNucleares, UniversidadNacionalAuto´nomadeMe´xico, p Circuitoexterior, CiudadUniversitaria, 04510, Me´xicoD.F.,Me´xico - h 1e-mail: [email protected] at 2e-mail: [email protected] m [ 3 Abstract: Weconstructtheinequivalentirreduciblerepresentations(IIR’s)oftheCPT v groupsoftheDiracfieldoperatorψˆandtheelectromagneticquantumpotentialAˆ . The µ 1 results are valid both for free and interacting (QED) fields. Also, and for the sake of 8 3 completeness,weconstruct theIIR’s oftheCPT groupoftheDiracequation. 2 . 6 0 9 AMSSubject Classification: 20C30, 20C35. 0 : v i X KeyWords: CPT groups;Dirac-Maxwellfields; irreduciblerepresentations. r a 1Introduction 2 1 Introduction The CPT group G (ψˆ) of the Dirac quantum field ψˆ is isomorphic to the Θˆ direct product of the quaternion group Q and the cyclic group of two elements Z ([1] 2 and [2]): G (ψˆ) ∼= Q×Z . (1) Θˆ 2 Θˆ is the product Cˆ ∗ Pˆ ∗ Tˆ of the three operators: Cˆ (charge conjugation), Pˆ (spaceinversion),Tˆ (timereversal);theoperationAˆ∗Bˆ,whereAˆandBˆ areanyofthe operators Cˆ, Pˆ, Tˆ, is givenby (Aˆ∗Bˆ)·ψˆ = (AˆBˆ)†ψˆ(AˆBˆ). G (ψˆ) is one of the nine Θˆ non abelian groups of a total of fourteen groups with sixteen elements [3]; only three ofthemhavethreegenerators. TheisomorphismG (ψˆ) → Q×Z isgivenbytherelations(2): Θˆ 2 1 7→ (1,1) −1 7→ (−1,1) Cˆ 7→ (1,−1) −Cˆ 7→ (−1,−1) Pˆ 7→ (ι,1) −Pˆ 7→ (−ι,1) Tˆ 7→ (γ,1) −Tˆ 7→ (−γ,1) Cˆ ∗Pˆ 7→ (ι,−1) −Cˆ ∗Pˆ 7→ (−ι,−1) Cˆ ∗Tˆ 7→ (γ,−1) −Cˆ ∗Tˆ 7→ (−γ,−1) ˆ ˆ ˆ ˆ P ∗T 7→ (κ,1) −P ∗T 7→ (−κ,1) Θˆ 7→ (κ,−1) −Θˆ 7→ (−κ,−1); (2) whereι, γ, κare thethree imaginaryunitsdefining thequaternionnumbers. On the otherhand, the action of theCˆ, Pˆ and Tˆ operators on the Maxwell electro- magnetic 4-potential Aˆ is given by (3) ([4] and [5]), with the Minkowski space-time µ metricdiag(1,−1,−1,−1): PˆAˆµ(x,t)Pˆ−1 = Aˆ (−x,t), µ CˆAˆµ(x,t)Cˆ−1 = −Aˆµ(x,t), TˆAˆµ(x,t)Tˆ−1 = Aˆ (x,−t). (3) µ ThisleadstotheCPT groupoftheelectromagneticfieldoperator,G (Aˆ ),which Θˆ µ is an abelian group of eight elements with three generators, isomorphic to Z × Z × 2 2 Z = Z3. Geometrically, Z3 is isomorphic to the group D , the symmetry group of 2 2 2 2h theparallelepiped. 1Introduction 3 Theisomorphism G (Aˆ ) → Z ×Z ×Z (4) Θˆ µ 2 2 2 isgivenby 1 7→ (e ,e ,e ) 1 2 3 Cˆ 7→ (a ,e ,e ) 1 2 3 ˆ P 7→ (e ,a ,e ) 1 2 3 Tˆ 7→ (e ,a ,a ) 1 2 3 Pˆ ∗Tˆ 7→ (e ,a ,a ) 1 2 3 Cˆ ∗Pˆ 7→ (a ,a ,e ) 1 2 3 Cˆ ∗Tˆ 7→ (a ,e ,a ) 1 2 3 Θˆ 7→ (a ,a ,a ); (5) 1 2 3 where e and a (for i = 1,2,3) represent, respectively, the identity and the “minus 1” i i ofeach oftheZ groups. 2 AstheCPT transformationpropertiesoftheinteractingψˆ−Aˆ fieldsarethesame µ as for the free fields ([5] and [6]), it is then clear that the complete CPT group for QED,G (QED),isthedirectproductofthetwoabovementionedgroups,G (ψˆ)and Θˆ Θˆ G (Aˆ ), i.e., Θˆ µ G (QED) = G (ψˆ)×G (Aˆ ), (6) Θˆ Θˆ Θˆ µ whichturnsout tobeagroupoforder|G (QED)| = 16×8 = 128. Θˆ Thus, G (QED) ∼= (Q×Z )×Z3. (7) Θˆ 2 2 Since, as will be shown below, G (ψˆ) has ten inequivalent irreducible represen- Θˆ tations (IIR’s), while G (Aˆ ) has only eight IIR’s, then, the total number of IIR’s of Θˆ µ G (QED) iseighty: sixtyfourofthem1-dimensionaland sixteen2-dimensional. Θˆ It is interesting at this point to make a comment about the geometrical content of G (QED),whichisasetofpointsinspheres. Infact,each factorZ isa0-sphereS0, Θˆ 2 while Q can be thought as a subset of eight points in the 3-sphere S3. Thus, topologi- cally G (QED) ⊂ SU(2)×(U(1))4. (8) Θˆ In section 2 we construct by tensor products the ten IIR’s and character tables of G (ψˆ). In section 3 we construct the eight IIR’s of G (Aˆ ); in this case, since all Θˆ Θˆ µ representations are 1-dimensional, the characters coincide with the representations. In ˆ 2IIR’s ofGΘˆ(ψ) 4 section 4 we construct the IIR’s of the CPT group corresponding to the Dirac equa- tion for the wave function ψ, G(2)(ψ) in [1], and compare the result with the case Θ for the Dirac field operator ψˆ. Finally, in section 5, we discuss the relation between the ¨fermionic” and ¨bosonic” CPT groups and comment about the relation of these groupswithsomeapproaches to quantumparadoxes. ˆ G 2 IIR’s of (ψ) Θˆ The irreducible representations (irrep’s) of the direct product of two groups, G × H, are the tensor products of the irrep’s of each of the factors, namely the set {r ⊗r }1 forallr′ s,irrep’sofGandallr′ s,irrep’sofH [7]. IfGisthequaternion G H G H group, which, as is well known, has five IIR’s ϕ (with ϕ ,...,ϕ 1-dimensional and µ 1 4 ϕ 2-dimensional)[8] and characters tablegivenby table1;and H is Z (withIIR’s ψ 5 2 1 and ψ ), then G (ψˆ) hasten IIR’s: 2 Θˆ φ = ϕ ⊗ψ , φ = ϕ ⊗ψ , α = 1,2,3,4, (9) α α 1 α+4 α 2 φ = ϕ ⊗ψ , φ = ϕ ⊗ψ . (10) 9 5 1 10 5 2 φ ,...,φ are1-dimensional,whileφ and φ are2-dimensional. 1 8 9 10 Ch Q [1] [−1] 2[ι] 2[γ] 2[κ] χ 1 1 1 1 1 1 χ 1 1 1 −1 −1 2 χ 1 1 −1 1 −1 3 χ 1 1 −1 −1 1 4 χ 2 −2 0 0 0 5 Table1: Characters tableofQ. Thecorrespondingcharacters arethefunctions κ = χ ϕ : Q×Z → C, κ (q,h) = χ (q)ϕ (h), (11) α α 1 2 α α 1 κ = χ ϕ : Q×Z → C, κ (q,h) = χ (q)ϕ (h), (12) α+4 α 2 2 α+4 α 2 κ = χ ϕ : Q×Z → C, κ (q,h) = χ (q)ϕ (h), (13) 9 5 1 2 9 5 1 κ = χ ϕ : Q×Z → C, κ (q,h) = χ (q)ϕ (h). (14) 10 5 2 2 10 5 2 1⊗=⊗C isthetensorproductoverthecomplexnumbers 3IIR’s ofGΘˆ(Aˆµ) 5 Theserepresentationsand characters (for conjugateclasses)are explicitlygivenby the tables 2 and 3, respectively, where λ = trφ , and the conjugate classes are given i i bytherelations(15): [Iˆ] = [(1,e)] = {(1,e)}, ˆ [C] = [(1,a)] = {(1,a)}, [−Iˆ] = [(−1,e)] = {(−1,e)}, [−Cˆ] = [(−1,a)] = {(−1,a)}, [Pˆ] = [(ι,e)] = {(ι,e),(−ι,e)}, [Cˆ ∗Pˆ] = [(ι,a)] = {(ι,a),(−ι,a)}, [Tˆ] = [(γ,e)] = {(γ,e),(−γ,e)}, [Cˆ ∗Tˆ] = [(γ,a)] = {(γ,a),(−γ,a)}, ˆ ˆ [P ∗T] = [(κ,e)] = {(κ,e),(−κ,e)}, [Θˆ] = [(κ,a)] = {(κ,a),(−κ,a)}; (15) whereeand aare theelementsofZ . 2 The character table of G (ψˆ) is the same as the character table of D , the group Θˆ 4h ofsymmetriesofaprismwithsquarebase, thoughQ×Z ≇ D . 2 4h ˆ G A 3 IIR’s of ( ) Θˆ µ Since G (Aˆ ) ∼= Z3 is an abelian group, all its irrep’s are 1-dimensional. Θˆ µ 2 Since |G (Aˆ )| = 8, then G (Aˆ ) has eight IIR’s, which can be identified with the Θˆ µ Θˆ µ corresponding characters. Using the isomorphism between G (Aˆ ) and Z3 (eq. (5)), Θˆ µ 2 andthecharactertableofZ ,weobtainthetableofrepresentations(φ )orcharacters 2 ijk (χ )forG (Aˆ ), whereφ = χ = ψ ψ ψ . See table4. ijk Θˆ µ ijk ijk i j k G(2) 4 IIR’s of (ψ) Θ In [1] it was shown that at the level of the Dirac equation, for the 4-spinor ψ, exists two CPT groups, G(1)(ψ) and G(2)(ψ), whose elements are constructed with Θ Θ products of Dirac γ-matrices. Only G(2)(ψ), which turns out isomorphic to the semi- Θ direct productD ⋊Z , whereD ≡ DH isthedihedral groupof eightelements(the 4 2 4 8 groupofsymmetriesofthesquare), iscompatiblewithG (ψˆ). Θˆ 4IIR’s ofG(2)(ψ) 6 Θ ItisthenofinteresttoconstructtheIIR’sofG(2)(ψ),bothforcompletnessandalso Θ forcomparisonwiththeIIR’s oftheoperatorgroup. The action of Z ∼= {1,−1} on D as a subgroup of S (the symmetric group of 4 2 4 4 elements)isgivenby λ : Z → Aut(D ), 2 4 1 7→ λ(1) = Id , D4 λ(−1)(I) = I, λ(−1)(1234) = (1234), λ(−1)(24) = (13), λ(−1)(13) = (24), λ(−1)((12)(34)) = (14)(23), λ(−1)((14)(23)) = (12)(34), λ(−1)((13)(24)) = (13)(24), λ(−1)(1432) = (1432). (16) Here D = {I;(1234),(1432);(13)(24);(12)(34),(14)(23);(24),(13)}, (17) 4 wherebetween semi-colonswehaveenclosedthefiveconjugationclasses. Then, thecompositionlawin D ⋊Z is 4 2 (g′,h′)(g,h) = (g′λ(h′)(g),h′h) : (18) (g′,1)(g,h) = (g′g,h) (19) (g′,−1)(g,h) = (g′λ(−1)(g),−h), (20) where, as aset, D ×Z = {(I,1),(I,−1),(−I,1),(−I,−1),(P,1),(P,−1),(−P,1), 4 2 (−P,−1),(CT,1),(CT,−1),(−CT,1),(−CT,−1),(Θ,1), (Θ,−1),(−Θ,1),(−Θ,−1)}, (21) withinverses (g,h)−1 = (λ(h)(g−1),h). (22) 4IIR’s ofG(2)(ψ) 7 Θ From eq. (55)in[1], thereis theisomorphism D → {I,−I,P,−P,CT,−CT,Θ,−Θ} (23) 4 givenby I 7→ I, (1234) 7→ P, (1432) 7→ −P, (13)(24) 7→ −I, (12)(34) 7→ Θ, (14)(23) 7→ −Θ, (24) 7→ −CT, (13) 7→ CT (24) whilefrom eq. (60)in [1]and eq. (17), theisomorphismbetween G(2) and D ⋊Z , Θ 4 2 G(2) → D ⋊Z , (25) Θ 4 2 isgivenby: I 7→ (I,1), −I 7→ (−I,1), C 7→ (−Θ,−1), −C 7→ (Θ,−1), P 7→ (P,1), −P 7→ (−P,1), T 7→ (P,−1), −T 7→ (−P,−1), CP 7→ (CT,−1), −CP 7→ (−CT,−1), CT 7→ (CT,1), −CT 7→ (−CT,1), PT 7→ (−I,−1), −PT 7→ (I,−1), Θ 7→ (Θ,1), −Θ 7→ (−Θ,1). (26) 4IIR’s ofG(2)(ψ) 8 Θ It iseasy to verifythatG(2) hasten conjugationclasses,namely Θ [I] = = {I}, [−I] = = {−I}, [C] = = {C,−C}, [T] = = {T}, [−T] = = {−T}, [P] = = {P,−P}, [CP] = = {CP,−CP}, [CT] = = {CT,−CT}, [PT] = = {PT,−PT}, [Θ] = = {Θ,−Θ}. (27) Then, being a finite group, G(2) has as many IIR’s as conjugation classes. Since Θ thesum ofthe squares ofthe dimensionsof theserepresentations mustbe 10, a simple calculationleadstotheexistenceofeight1-dimensionalirrep’sϕ ,k = 1,...8,andtwo k 2-dimensionalirrep’s,ϕ and ϕ . 9 10 ϕ is the trivial representation g 7→ 1, for all g ∈ G(2). The three 1-dimensional 1 Θ irrep’sϕ ,ϕ andϕ areobtainedtakingintoaccount: 2 3 4 1. D , C ×Z andQ areinvariantsubgroupsofD ⋊Z . 4 4 2 4 2 2. The well known theorem by which if H is an invariant subgroup of G and ϕ : G/H → K < GL(V) is a representation of G/H over the vector space V, then ϕ◦p is a degenerate representation of G, where p : G → G/H is the canonical projectionp(g) =< g >= gH [9]. Thus,wehavethecommutativediagram: K ϕ◦p ր տ ϕ . (28) G −→ G/H p D isgivenby ther.h.s. ofeq. (23)and C =< {P} >,so 4 4 C ×Z = {(I,1),(I,−1),(−I,1),(−I,−1),(P,1),(P,−1),(−P,1),(−P,−1)}, 4 2 (29) and Q ∼=< {C,P} >= {I,C,P,CP,−I −C,−P,−CP}. (30) In diagram (28), we choose G = D ⋊ Z and H equal to D , C × Z and Q, 4 2 4 4 2 respectively. Then 4IIR’s ofG(2)(ψ) 9 Θ D ⋊Z 4 2 ∼= Z = {e,a} (31) 2 H with e = D , a = {C,T,CP,PT,−C,−T,−CP,−PT}, 4 e = C ×Z , a = {C,CP,CT,Θ,−C,−CP,−CT,−Θ}, 4 2 e = Q, a = {T,CT,PT,Θ,−T,−CT,−PT,−Θ}, (32) respectively. Theelementsinthee’sarerepresentedby1,whiletheelementsinthea’s arerepresented by −1 (seetable5). As is well known, D has five IIR’s (one 2-dimensional and four 1-dimensional). 4 The2-dimensionalirrep isgivenby 1 0 I 7→ = 1 ≡ I (cid:18)0 1(cid:19) 2×2 −1 0 (13)(24) 7→ = −1 ≡ −I (cid:18) 0 −1(cid:19) 2×2 0 −1 (1234) 7→ = −iσ ≡ P (cid:18)1 0 (cid:19) 2 0 1 (1432) 7→ = iσ ≡ −P (cid:18)−1 0(cid:19) 2 −1 0 (12)(34) 7→ = −σ ≡ Θ (cid:18) 0 1(cid:19) 3 1 0 (14)(23) 7→ = σ ≡ −Θ (cid:18)0 −1(cid:19) 3 0 1 (24) 7→ = σ ≡ −CT (cid:18)1 0(cid:19) 1 0 −1 (13) 7→ = −σ ≡ CT, (33) (cid:18)−1 0 (cid:19) 1 whereinthelastidentificationweused therelations(24). WiththechoiceC = iσ , weobtain, 1 i 0 i 0 0 −i T = = iI, CP = = iσ , PT = = σ , (34) (cid:18)0 i(cid:19) (cid:18)0 −i(cid:19) 3 (cid:18)i 0 (cid:19) 2 5Final comments 10 which completes the 2-dimensionalirrep ϕ of G(2); whilewith the choice C = −iσ , 9 Θ 1 weobtain, −i 0 −i 0 0 i T = = −iI, CP = = −iσ , PT = = −σ , (cid:18) 0 −i(cid:19) (cid:18) 0 i(cid:19) 3 (cid:18)−i 0(cid:19) 2 (35) which completes the 2-dimensional irrep ϕ of G(2). It can be easily verified that ϕ 10 Θ 9 and ϕ are inequivalent, i.e., it does not exist a matrix S such that SMS† = M′ (or 10 SMS−1 = M′)forM ∈ ϕ and M′ ∈ ϕ . 9 10 The remaining four 1-dimensional IIR’s, ϕ , ϕ , ϕ and ϕ , are obtained from 5 6 7 8 the orthogonality between the columns of the characters table for conjugation classes (completeness relation). Then, the orthogonality between the rows of the complete characters tableisverified. TheIIR’s and characters forG(2)(ψ) ∼= D ⋊Z are summarizedintables5 and 6. Θ 4 2 ComparingwiththetablesforG (ψˆ), weseethatthe1-dimensionalirrep’scoincide: Θˆ φ = ϕ , φ = ϕ , φ = ϕ , φ = ϕ , 1 1 2 3 3 4 4 2 φ = ϕ , φ = ϕ , φ = ϕ , φ = ϕ . (36) 5 8 6 6 7 7 8 5 However,φ is notequivalenttoeitherϕ orϕ andthesameholdsforφ . 9 9 10 10 5 Final comments It is of interest to ask the question if, from the point of view of group theory, thebehaviourofthephotonfieldAˆ undertheC,P,T transformationisindependentof µ thebehaviouroftheelectron-positronfieldψˆ. Theanswertothisquestionisafirmative. In fact, in G (ψˆ) there is only one Z subgroup, namely that generated by Cˆ : Z ∼= Θˆ 2 2 {Iˆ,Cˆ}. Then,G (Aˆ ) ∼= Z3 is notasubgroupofG (ψˆ): Θˆ µ 2 Θˆ G (Aˆ ) 6< G (ψˆ). (37) Θˆ µ Θˆ ThesamehappensinrelationtoG(2)(ψ). InthisgrouptherearethreeZ -subgroups: Θ 2 Z(1) = {I,CT}, Z(2) = {I,PT}, Z(3) = {I,Θ}. (38) 2 2 2 However, the table of Z3 has eight 1’s, while the table of G(2)(ψ) has only seven 2 Θ 1’s. So G (Aˆ ) 6< G(2). (39) Θˆ µ Θ