Chapter 11

Spinor Formulation of Relativistic Quantum

11.1 The of the Dirac

We will provide in the following a new formulation of the in the chiral representation defined through (10.225–10.229). Starting is the Lorentz transformation ˜(~w, ϑ~) for the bispinor Ψ˜ in the chiral representation as given by (10.262). ThisS transformation can be written

a(~z) 0 ˜(~z) =   (11.1) S 0 b(~z) 1 a(~z) = exp  ~z ~σ  (11.2) 2 · 1 b(~z) = exp  ~z ∗ ~σ  (11.3) −2 · ~z = ~w i ϑ~ . (11.4) − We have altered here slightly our notation of ˜(~w, ϑ~), expressing its dependence on ~w, ϑ~ through a complex ~z, ~z C3. S ∈ Because of its block- form each of the diagonal components of ˜(~z), i.e., a(~z) and b(~z), must be two-dimensional irreducible representations of the Lorentz .S This fact is remarkable since it implies that the representations provided through a(~z) and b(~z) are of lower then 1 ~ µ the four-dimensional natural representation L(~w, ϑ) ν. The lower dimensionality of a(~z) and b(~z) implies, in a sense, that the corresponding representation of the is more than the natural representation and may serve as a building block for all representations, in particular, may be exploited to express the Lorentz- equations of relativistic . This is, indeed, what will be achieved in the following. We will proceed by building as much as possible on the results obtained sofar in the chiral repre- sentation of the Dirac equation. We will characterize the on which the transformations a(~z)

1We will see below that the representations a(~z) and b(~z) are, in fact, isomorphic to the natural representation, ~ µ i.e., different L( ~w, ϑ) ν correspond to different a(~z) and b(~z).

355 356 Formulation

~ µ and b(~z) act, the so-called spinor space, will establish the between L(~w, ϑ) ν and a(~z), b(~z), µ express 4-vectors A ,Aν, the ∂µ and the ~σ in the new representation and, finally, formulate the Dirac equation, equation, and the Klein–Gordon equation in the spinor representation.

A First Characterization of the Bispinor States We note that in case ~w = 0 the Dirac transformations are pure . In this case a(~z) and b(~z) are identical and read

1 a(i ϑ~) = b(i ϑ~) = exp  ϑ~ ~σ  , θ R3 . (11.5) −2 · ∈

The transformations in this case, i.e., for ~z = iϑ,~ ϑ~ R3, are elements of SU(2) and correspond, in ∈ ( 1 ) fact, to the rotational transformations of - 1 states as described by D 2 (ϑ~), usually expressed 2 m m0 as of rotations and of functions of α, β, γ (see Chapter 5). For ϑ~ = (0, β, 0)T the transformations are

β β ( 1 ) 2 cos 2 sin 2 a(i β eˆ2) = b(i β eˆ2) = d (β) =  −  . (11.6)  mm0  β β sin 2 cos 2 as given by (5.243). This characterization allows one to draw conclusions regarding the state space in which a(~z) and b(~z) operate, namely, a space of vectors state1 for which holds state2 state 1 1 , + 1  ∼ | 2 2 i  ~z = iϑ,~ ϑ~ R3 (11.7) state 2 1 , 1 ∈ ∼ | 2 − 2 i where “ ” stands for “transforms like”. Here 1 , 1 represents the familiar spin– 1 states. ∼ | 2 ± 2 i 2 Since a(~z) acts on the first two components of the solution Ψ˜ of the Dirac equation, and since b(~z) acts on the third and fourth component of Ψ˜ one can characterize Ψ˜

µ 1 1 Ψ˜ 1(x ) , + ∼ | 2 2 i  Ψ˜ (xµ) 1 , 1  2 2 2 ~ ~ R3 ˜ µ ∼ | 1 − 1 i ~z = iϑ, ϑ . (11.8)  Ψ3(x ) 2 , + 2  ∈  µ ∼ | 1 1 i   Ψ˜ 4(x ) ,  ∼ | 2 − 2 i We like to stress that there exists, however, a distinct difference in the transformation behaviour of µ µ µ µ Ψ˜ 1(x ), Ψ˜ 2(x ) and Ψ˜ 3(x ), Ψ˜ 4(x ) in case ~z = ~w + iϑ~ for ~w = 0. In this case holds a(~z) = b(~z) µ µ 6 µ µ 6 and Ψ˜ 1(x ), Ψ˜ 2(x ) transform according to a(~z) whereas Ψ˜ 3(x ), Ψ˜ 4(x ) transform according to b(~z).

Relationship Between a(~z) and b(~z) The transformation b(~z) can be related to the conjugate complex of the transformation a(~z), i.e., to 1 a∗(~z) = exp  ~z ∗ ~σ ∗ (11.9) 2 · 11.1: Lorentz Transformation of Dirac Bispinor 357

where ~σ ∗ = (σ1∗, σ2∗, σ3∗). One can readily verify [c.f. (5.224)]

σ1 = σ∗ , σ2 = σ∗ , σ3 = σ∗ . (11.10) 1 − 2 3 From this one can derive 1 b(~z) =  a∗(~z) − (11.11) where 0 1 0 1  = ,  1 = . (11.12)  1 0  −  1− 0  − 1 To prove (11.11) one first demonstrates that for  and − as given in (11.12) does, in fact, hold 1 1 1  − = 11. One notices then, using f(a)− = f(a− ),

1 1 1  a∗(~z)− = exp  ~z ∗ ~σ ∗−  . (11.13) 2 

Explicit using (5.224, 11.10, 11.12) yields

1 σ1− = σ1 = σ1∗ 1 − − σ2− = σ2 = σ2∗ (11.14) 1 − σ3− = σ3 = σ∗ , − − 3 or in short 1 ~σ− = ~σ ∗ . (11.15) − Similarly, one can show 1 − ~σ = ~σ ∗ , (11.16) − a result to be used further below. Hence, one can express

1 1  a∗(~z) − = exp  ~z ∗ ~σ = b(~z) . (11.17) −2 ·

We conclude, therefore, that the transformation (11.1) can be written in the form

˜ a(~z) 0 (~z) =  1  (11.18) S 0  a∗(~z) −

1 with a(~z) given by (11.2, 11.4) and , − given by (11.12). This demonstrates that a(~z) is the transformation which characterizes both components of ˜(~z). S Spatial Inversion One may question from the form of ˜(~z) why the Dirac equation needs to be four-dimensional, µ S µ µ µ featuring the components Ψ˜ 1(x ), Ψ˜ 2(x ) as well as Ψ˜ 3(x ), Ψ˜ 4(x ) even though these pairs of components transform independently of each other. The answer lies in the necessity that application of spatial inversion should transform a solution of the Dirac equation into another possible solution of the Dirac equation. The effect of inversion on Lorentz transformations is, however, that they alter ~w into ~w, but leave angles ϑ~ unaltered. − 358 Spinor Formulation

Let denote the representation of spatial inversion in the space of the wave functions Ψ.˜ Obviously, 2 =P 11,i.e., 1 = . The transformation ˜(~z) in the transformed space is then P P− P S b(~z) 0 ˜(~w + iϑ~) = ˜( ~w + iϑ~) =   , (11.19) P S P S − 0 a(~z) i.e., the transformations a(~z) and b(~z) become interchanged. This implies

Ψ˜ 1 Ψ˜ 3  Ψ˜   Ψ˜  2 = 4 . (11.20) P  Ψ˜   Ψ˜   3   1   Ψ˜ 4   Ψ˜ 2 

Obviously, the space spanned by only two of the components of Ψ˜ is not invariant under spatial inversion and, hence, does not suffice for particles like the which obey inversion . However, for particles like the which do not obey inversion symmetry two components of the wave function are sufficient. In fact, the Lorentz invariant equation for neutrinos is only 2–dimensional.

11.2 Relationship Between the Lie Groups SL(2,C) and SO(3,1)

We have pointed out that a(iϑ~), ϑ~ R3, which describes pure rotations, is an element of SU(2). However, a(~w + iϑ~) for ~w = 0 is an∈ element of 6 SL(2, C) = M,M is a complex 2 2–matrix, det(M) = 1 . (11.21) { × } One can verify this by evaluating the of a(~z)

1 ~z ~σ tr( 1 ~z ~σ) det( a(~z) ) = det  e 2 ·  = e 2 · = 1 (11.22) which follows from the fact that for any complex, non-singular matrix M holds2

det eM = etr(M) (11.23)  and from [c.f. (5.224)]

tr( σj ) = 0 , j = 1, 2, 3 . (11.24)

Exercise 11.2.1: Show that SL(2, C) defined in (11.21) together with as the binary forms a group.

2The proof of this important property is straightforward in case of hermitian M (see Chapter 5). For the general case the proof, based on the Jordan–Chevalley theorem, can be found in G.G.A. B¨auerleand E.A. de Kerf Lie , Part (Elsevier, Amsterdam, 1990), Exercise 1.10.3. 11.2: Lie Groups SL(2,C) and SO(3,1) 359

Mapping Aµ onto matrices M(Aµ)

We want to establish now the relationship between SL(2, C) and the group +↑ of proper, or- thochronous Lorentz transformations. Starting point is a bijective map betweenLR4 and the of two-dimensional hermitian matrices defined through µ µ M(A ) = σµ A (11.25) where   1 0 0 1 0 i 1 0 σµ =    ,   ,  −  ,    . (11.26)  0 1 1 0 i 0 0 1   −   σ σ σ σ  | {z0 } | {z1 } | {z2 } | {z3 } The quantity σµ thus defined does not transform like a covariant 4–vector. In fact, one wishes that the definition (11.25) of the matrix M(Aµ) is independent of the of reference, i.e., in a transformed frame should hold ρ µ L ν µ σµA σµA0 . (11.27) −→ Straightforward transformation into another would replace σµ by σµ0 . Using µ 1 µ ν A = (L− ) νA0 one would expect in a transformed frame to hold ρ µ L ν 1 µ ν σµA σµ0 (L− ) ν A0 . (11.28) −→ Consistency of (11.28) and (11.27) requires then µ Lν σµ0 = σν (11.29) µ where we used (10.76). Noting that for covariant vectors according to (10.75) holds aν0 = Lν aµ one realizes that σµ transforms inversely to covariant 4–vectors. We will prove below [cf. (11.135)] this transformation behaviour. M(Aµ) according to (11.25) can also be written A0 + A3 A1 iA2 M(Aµ) = . (11.30)  A1 + iA2 A0− A3  − Since the components of Aµ are real, the matrix M(Aµ) is hermitian as can be seen from inspection µ of (11.26) or from the fact that the matrices σ0, σ1, σ2, σ3 are hermitian. The function M(A ) is bijective, in fact, one can provide a simple expression for the inverse of M(Aµ)

µ µ 1 M 0 = M(A ) A = tr M 0 σµ . (11.31) ↔ 2 

Exercise 11.2.2: Show that σ0, σ1, σ2, σ3 provide a linear–independent for the space of µ µ hermitian 2 2–matrices. Argue why M(A ) = σµA provides a bijective map. Demonstrate that (11.31) holds.×

The following important property holds for M(Aµ) µ µ det ( M(A ) ) = A Aµ (11.32) which follows directly from (11.30). 360 Spinor Formulation

Transforming the matrices M(Aµ)

We define now a transformation of the matrix M(Aµ) in the space of hermitian 2 2–matrices × a M M 0 = a M a† , a SL(2, C) . (11.33) −→ ∈ This transformation conserves the hermitian property of M since

(M 0)† = ( a M a† )† = (a†)†M †a† = a M a† = M 0 (11.34) where we used the properties M † = M and (a†)† = a. Due to det(a) = 1 the transformation (11.33) conserves the determinant of M. In fact, it holds for the matrix M 0 defined through (11.33)

det(M 0) = det( a M a† ) = det(a) det(a†) det(M) = [det(a)]2 det(M) = det(M) . (11.35)

We now apply the transformation (11.33) to M(Aµ) describing the action of the transformation in µ µ µ terms of transformations of A . In fact, for any a SL(2, C) and for any A there exists an A0 such that ∈ µ µ M(A0 ) = a M(A ) a† . (11.36)

µ The suitable A0 can readily be constructed using (11.31). Accordingly, any a SL(2, C) defines the transformation [c.f. (11.31)] ∈

µ a µ 1 µ A A0 = tr a M(A ) a†σµ . (11.37) −→ 2  

Because of (11.32, 11.35) holds for this transformation

µ µ A0 Aµ0 = A Aµ (11.38) which implies that (11.37) defines actually a Lorentz transformation. The linear character of the µ transformation becomes apparent expressing A0 as given in (11.37) using (11.25)

1 A µ = tr a σ a σ Aν (11.39) 0 2  ν † µ  which allows us to express finally

1 A µ = L(a)µ Aν ; L(a)µ = tr a σ a σ . (11.40) 0 ν ν 2  ν † µ 

µ Exercise 11.2.3: Show that L(a) ν defined in (11.40) is an element of ↑ . L+ 11.2: Lie Groups SL(2,C) and SO(3,1) 361

µ L(a) ν provides a

µ We want to demonstrate now that the map between SL(2, C) and SO(3,1) defined through L(a) ν [cf. (11.40)] respects the group property of SL(2, C) and of SO(3,1), i.e.,

µ µ µ µ L¯ ρ = L(a1) ν L(a2) ρ = L( a1 a2 ) ρ (11.41) product | product{z } | {z } in SO(3,1) in SL(2,C)

For this purpose one writes using tr(AB) = tr(BA)

µ µ 1 1 L(a ) L(a ) = tr a σ a†σ tr a σ a†σ 1 ν 2 ρ X 2  1 ν 1 µ  2  2 ρ 2 ν  ν 1 1 = tr σ a†σ a tr a σ a†σ . (11.42) X 2  ν 1 µ 1  2  2 ρ 2 ν  ν

Defining

Γ = a1†σµa1 , Γ0 = a2σρa2† (11.43) µ and using the definition of L¯ ρ in (11.41) results in

1 1 L¯µ = (σ ) Γ Γ (σ ) = A Γ Γ (11.44) ρ 4 X ν αβ βα γδ ν δγ 4 X αβγδ βα γδ ν,α,β α,β γ,δ γ,δ where A = (σ ) (σ ) . (11.45) αβγδ X ν αβ ν δγ ν One can demonstrate through direct evaluation

2 α = β = γ = δ = 1   2 α = β = γ = δ = 2  Aαβγδ =  2 α = γ = 1 , β = δ = 2 (11.46) 2 α = γ = 2 , β = δ = 1   0 else  which yields

µ 1 1 L¯ ρ = Γ11Γ0 + Γ22Γ0 + Γ12Γ0 + Γ21Γ0 = tr ΓΓ0 2 11 22 21 12  2  1 1 = tr a†σ a a σ a† = tr σ a a σ a†a† 2  1 µ 1 2 ρ 2  2  µ 1 2 ρ 2 1  1 = tr a a σ (a a ) σ = L(a a )µ . (11.47) 2  1 2 ρ 1 2 † µ  1 2 ρ

µ This completes the proof of the homomorphic property of L(a) ν. 362 Spinor Formulation

Generators for SL(2, C) which correspond to K,~ J~ The transformations a SL(2, C) as complex 2 2–matrices are defined through four complex or, correspondingly, eight∈ real . Because of× the condition det(a) = 1 only six independent real numbers actually suffice for the definition of a. One expects then that six generators Gj and six real coordinates fj can be defined which allow one to represent a in the form

6 a = exp  f G  . (11.48) X j j  j=1  We want to determine now the generators of the transformation a(~z) as defined in (11.2) which µ correspond to the generators K1,K2,K3,J1,J2,J3 of the Lorentz transformations L ν in the natural representations, i.e., correspond to the generators given by (10.47, 10.48). To this end we consider infinitesimal transformations and keep only terms of zero order and first order in the small variables. To obtain the generator of a(~z) corresponding to the generator K1, denoted below as κ1, we write (11.36) µ ν µ M(L νA ) = a M(A ) a† (11.49) µ assuming (note that g ν is just the familiar Kronecker δµν)

µ µ µ L ν = g ν +  (K1) ν (11.50)

a = 11 +  κ1 . (11.51)

µ µ Insertion of (K1) ν as given in (10.48) yields for the l.h.s. of (11.49), noting the linearity of M(A ),

µ µ ν µ 1 0 M(A +  (K1) ν A ) = M(A ) +  M(( A , A , 0, 0) ) µ 1− − 0 = M(A )  σ0A  σ1A (11.52) − − where we employed (11.25) in the last step. For the .h.s. of (11.49) we obtain using (11.51)

µ ( 11 +  κ1 ) M(A ) ( 11 +  κ1† ) µ µ µ 2 = M(A ) +  ( κ1M(A ) + M(A )κ1† ) + O( ) µ µ 2 = M(A ) +  ( κ1σµ + σµκ1† ) A + O( ) . (11.53) Equating (11.52) and (11.53) results in the condition

1 0 µ σ0A σ1A = ( κ1σµ + σµ κ† ) A . (11.54) − 1 This reads for the four cases Aµ = (1, 0, 0, 0), Aµ = (0, 1, 0, 0), Aµ = (0, 0, 1, 0), Aµ = (0, 0, 0, 1)

σ1 = κ1σ0 + σ0κ† = κ1 + κ† (11.55) − 1 1 σ0 = κ1σ1 + σ1κ1† (11.56)

0 = κ1σ2 + σ2κ1† (11.57)

0 = κ1σ3 + σ3κ1† . (11.58) One can verify readily that 1 κ1 = σ1 (11.59) −2 11.3: 363 obeys these conditions. Similarly, one can show that the generators κ2, κ3 of a(~z) corresponding to K2,K3 and λ1, λ2, λ3 corresponding to J1,J2,J3 are given by 1 i κ = ~σ, ~λ = ~σ. (11.60) −2 2 We can, hence, state that the following two transformations are equivalent

~w K~ + ϑ~ J~ 1 ( ~w iϑ~) ~σ L(~w, ϑ~) = e · · , a(~w iϑ~) = e− 2 − · (11.61) − | {z } | {z } SO(3,1) SL(2, C) acts on∈ 4–vectors Aµ acts on∈spinors φα C2 (characterized below)∈

This identifies the transformations a(~z = ~w iϑ~) as representations of Lorentz transformations, ~w describing boosts and ϑ~ describing rotations.−

Exercise 11.2.4: Show that the generators (11.60) of a SL(2, C) correspond to the generators µ ∈ K~ and J~ of L ν.

11.3 Spinors

Definition of contravariant spinors We will now further characterize the states on which the transformation a(~z) and its conjugate complex a∗(~z) act, the so-called contravariant spinors. We consider first the transformation a(~z) which acts on a 2-dimensional space of states denoted by

φ1 φα =   C2 . (11.62) φ2 ∈ According to our earlier discussion holds

φ1 transforms under rotations (~z = iϑ~) like a spin– 1 state 1 , + 1 2 | 2 2 i φ2 transforms under rotations (~z = iϑ~) like a spin– 1 state 1 , 1 . 2 | 2 − 2 i We denote the general (~z = ~w + iϑ~) transformation by

α α β def α 1 α 2 φ0 = a β φ = a 1φ + a 2φ , α = 1, 2 (11.63) where we extended the convention of 4-vectors to spinors. Here

1 1 α a 1 a 2 a(~z) = ( a β ) =  2 2  (11.64) a 1 a 2 describes the matrix a(~z). 364 Spinor Formulation

Definition of a product The question arises if for the states φα there exists a scalar product which is invariant under Lorentz transformations, i.e., invariant under transformations a(~z). Such a scalar product does, indeed, exist and it plays a role for spinors which is as central as the role of the scalar product µ A Aµ is for 4–vectors. To arrive at a suitable scalar product we consider first only rotational ~ α 1 transformations a(iϑ). In this case spinors φ transform like spin– 2 states and an invariant, which can be constructed from products φ1χ2, etc., is the singlet state. In the notation developed in Chapter 5 holds for the singlet state

1 1 1 1 1 1 , ; 0, 0 = (0, 0 , m; , m) , m 1 , m 2 (11.65) | 2 2 i X | 2 2 − | 2 i | 2 − i m= 1 ± 2 where 1 describes the spin state of “particle 1” and 2 describes the spin state of “particle 2” and| ·(0 · ·i, 0 1 , 1 ; 1 , 1 ) stands for the Clebsch–Gordon| · · coefficient. ·i Using (0, 0 1 , 1 ; 1 , 1 ) = | 2 ± 2 2 ∓ 2 | 2 ± 2 2 ∓ 2 1/√2 and equating the spin states of “particle 1” with the spinor φα, those of “particle 2” with ±the spinor χβ one can state that the quantity 1 Σ = φ1 χ2 φ2 χ1 (11.66) √2 −  should constitute a singlet spin state, i.e., should remain invariant under transformations a(iϑ~). In fact, as we will demonstrate below such states are invariant under general Lorentz transformations a(~z).

Definition of covariant spinors Expression (11.66) is a , invariant and as such has the necessary properties of a scalar3 product. However, this scalar product is anti-symmetric, i.e., exchange of φα and χβ alters the sign of the expression. The existence of a scalar product gives rise to the definition of a dual α representation of the states φ denoted by φα. The corresponding states are defined through

1 2 2 1 1 2 φ χ φ χ = φ χ1 + φ χ2 (11.67) − It obviously holds 2 χ1 χ χα =   =  1  . (11.68) χ2 χ − α β We will refer to φ , χ ,... as contravariant spinors and to φα, χβ,... as covariant spinors. The relationship between the two can be expressed

1 φ1 φ   =   2  (11.69) φ2 φ 0 1  = (11.70)  1 0  − 3‘Scalar’ implies invariance under rotations and is conventionally generalized to invariance under other symmetry trasnformations. 11.3: Spinors 365 as can be verified from (11.68). The inverse of (11.69, 11.70) is

1 φ 1 φ1  2  = −   (11.71) φ φ2 0 1  1 = . (11.72) −  1− 0 

Exercise 11.3.1: Show that for any non-singular complex 2 2–matrix M holds × 1 1 T  M − = det(M) M − 

1 The matrices , − connecting contravariant and covariant spinors play the role of the metric µν gµν, g of the [cf. (10.10, 10.74)]. Accordingly, we will express (11.69, 11.70) and (11.71, 11.72) in a notation analogous to that chosen for contravariant and covariant 4–vectors [cf. (10.72)]

β φα = αβ φ (11.73) α αβ φ =  φβ (11.74)

11 12 0 1 αβ =   =   (11.75) 21 22 1 0 − 11 12 0 1 αβ = = (11.76)  21 22   1− 0 

The scalar product (11.67) will be expressed as

α 1 2 1 2 2 1 φ χα = φ χ1 + φ χ2 = φ χ φ χ . (11.77) − For this scalar product holds α α φ χα = χ φα . (11.78) − The transformation behaviour of φα according to (11.63, 11.73, 11.75) is given by

β γδ φα0 = αβ a γ  φδ (11.79) as can be readily verified.

α Proof that φ χα is Lorentz invariant We want to verify now that the scalar product (11.77) is Lorentz invariant. In the transformed frame holds α α γ δκ β φ0 χα0 = a β αγa δ φ χκ . (11.80) One can write in matrix notation

α γ δκ 1 T a β αγa δ =  a − a . (11.81) h  iκβ 366 Spinor Formulation

Using (11.2) and (11.14) one can write

1 1 ~z ~σ 1 1 ~z ~σ 1 1 ~z ~σ  a − =  e 2 · − = e 2 · − = e− 2 · ∗ (11.82) and with f(A)T = f(AT ) for f(A)

1 T 1 ~z (~σ )T 1 ~z ~σ 1  a − = e− 2 · ∗ = e− 2 · = a− (11.83)  T Here we have employed the hermitian property of ~σ, i.e., ~σ† = (~σ ∗) = ~σ. Insertion of (11.83) into (11.81) yields

α γ δκ 1 T 1 a β αγa δ =  a − a = a− a = δκβ (11.84) h  iκβ  κβ and, hence, from (11.80) α β φ0 χα0 = φ χβ . (11.85)

The spinors We consider now the conjugate complex spinors

1 α φ ∗ (φ )∗ =    . (11.86) φ2 ∗  A concise notation of the conjugate complex spinors is provided by

1˙ α α˙ φ (φ )∗ = φ = ! (11.87) φ2˙

k˙ k which we will employ from now on. Obviously, it holds φ = (φ )∗, k = 1, 2. The transformation behaviour of φα˙ is α˙ α β˙ φ0 = (a β)∗ φ (11.88) which one verifies taking the conjugate complex of (11.63). As discussed above, a∗(~z) provides a representation of the Lorentz group which is distinct from that provided by a(~z). Hence, the conjugate complex spinors φα˙ need to be considered separately from the spinors φα. We denote

α α˙ (a β)∗ = a β˙ (11.89) such that (11.88) reads α˙ α˙ β˙ φ0 = a β˙ φ (11.90) extending the summation convention to ‘dotted’ indices. We also define covariant versions of φα˙

φα˙ = (φα)∗ . (11.91) 11.4: Spinor Tensors 367

The relationship between contravariant and covariant conjugate complex spinors can be expressed in analogy to (11.73, 11.76)

β˙ φα˙ = α˙ β˙ φ (11.92) α˙ α˙ β˙ φ =  φβ˙ (11.93)

α˙ β˙ = αβ (11.94) ˙ α˙ β = αβ (11.95)

αβ α˙ where αβ and  are the real matrices defined in (11.75, 11.75). For the spinors φ and χα˙ thus defined holds that the scalar product

α˙ 1˙ 2˙ φ χα˙ = φ χ1˙ + φ χ2˙ (11.96) is Lorentz invariant, a property which is rather evident. The transformation behaviour of φα˙ is

β˙ γ˙ δ˙ φα0˙ = α˙ β˙ a γ˙  φδ˙ . (11.97)

1 The transformation, in matrix notation, is governed by the operator  a∗(~z) − which arises in ˜ 1 the Lorentz transformation (11.18) of the bispinor wave function Ψ,  a∗(~z) − accounting for the transformation behaviour of the third and fourth spinor component of Ψ.˜ A comparision of (11.18) and (11.97) implies then that φα˙ transforms like Ψ˜ 3, Ψ˜ 4, i.e., one can state

1 1 1 φ˙ transforms under rotations (~z = iϑ~) like a spin– state , + 1 2 | 2 2 i 1 1 1 φ˙ transforms under rotations (~z = iϑ~) like a spin– state , . 2 2 | 2 − 2 i The transformation behaviour of Ψ˜ (note that we do not include presently the space-time depen- dence of the wave function)

Ψ˜ 1 a(~z) ˜ ˜ ˜ ˜  Ψ2  Ψ0 = (~z) Ψ = ˜  (11.98) 1 Ψ3 S a∗(~z)− ˜  Ψ4  obviously implies that the solution of the Dirac equation in the chiral representation can be written in spinor form µ 1 µ Ψ˜ 1(x ) φ (x )  µ   2 µ  α µ Ψ˜ 2(x ) φ (x ) φ (x ) ˜ µ = µ =  µ  . (11.99)  Ψ3(x )   χ1˙ (x )  χβ˙ (x )  ˜ µ   µ   Ψ4(x )   χ2˙ (x ) 

11.4 Spinor Tensors

We generalize now our definition of spinors φα to tensors. A

˙ ˙ ˙ α1α2 αkβ1β2 β` t ··· ··· (11.100) 368 Spinor Formulation is a quantity which under Lorentz transformations behaves as

k ` ˙ ˙ ˙ ˙ ˙ ˙ α1α2 αkβ1β2 β` αm βn γ1 γkδ1 δ` t0 ··· ··· = a γ a ˙ t ··· ··· . (11.101) Y m Y δn m=1 n=1

An example is the tensor tαβ˙ which will play an important role in the spinor presentation of the Dirac equation. This tensor transforms according to

αβ˙ α β˙ γδ˙ t0 = a γa δ˙ t (11.102) This reads in matrix notation, using conventional matrix indices j, k, `, m,

tjk0 = a t a† = aj`akm∗ t`m . (11.103)  jk X `,m

Similarly, the transformation bevaviour of a tensor tαβ reads in spinor and matrix notation

αβ α β γδ T t0 = a γ a δ t , t0 = a t a = aj`akmt`m (11.104) jk jk X `,m Indices on tensors can also be lowered employing the formula

β˙ γβ˙ tα = αγt (11.105) and generalizations thereof. αβ An example of a tensor is  and αβ. This tensor is actually invariant under Lorentz transforma- tions, i.e., it holds αβ αβ 0 =  , αβ0 = αβ (11.106)

Exercise 11.4.1: Prove equation (11.106).

The 4–vector Aµ in spinor form We want to provide now the spinor form of the 4-vector Aµ, i.e., we want to express Aµ through a spinor tensor. This task implies that we seek a tensor, the elements of which are linear functions µ µ µ µ of A . An obvious candidate is [cf. (11.25] M(A ) = σµA . We had demonstrated that M(A ) transforms according to µ ν µ M 0 = M(L νA ) = a M(A ) a† (11.107) which reads in spinor notation [cf. (11.102, 11.103)

αβ˙ α β˙ γδ˙ A0 = a γa δ˙A . (11.108) Obviously, this transformation behaviour is in harmony with the tensor notation adopted, i.e., with contravariant indices αβ˙. According to (11.25) the tensor is explicitly

11˙ 12˙ 0 3 1 2 ˙ A A A + A A iA Aαβ = ! =  −  . (11.109) A21˙ A22˙ A1 + iA2 A0 A3 − 11.4: Spinor Tensors 369

αβ˙ One can express A also through Aµ

˙ A0 A3 A1 + iA2 Aαβ =  − −  . (11.110) A1 iA2 A0 + A3 − − µ The 4-vectors A ,Aµ can also be associated with tensors

γδ˙ Aαβ˙ = αγβ˙δ˙A . (11.111)

This tensor reads in matrix notation

11˙ 12˙ A11˙ A12˙ 0 1 A A ! 0 1   =   21˙ 22˙  −  A ˙ A ˙ 1 0 A A 1 0 21 22 − A22˙ A21˙ = − ! . (11.112) A12˙ A11˙ − Hence, employing (11.109, 11.110) one obtains

A0 A3 A1 iA2 A = (11.113) αβ˙  A1 +−iA2 − A0−+ A3  −

A0 + A3 A1 + iA2 Aαβ˙ =   . (11.114) A1 iA2 A0 A3 − − µ We finally note that the 4–vector scalar product A Bµ reads in spinor notation

1 ˙ AµB = AαβB . (11.115) µ 2 αβ˙

Exercise 11.4.2: Prove that (11.115) is correct.

∂µ in spinor notation

µ αβ˙ The relationship between 4–vectors A ,Aµ and tensors t can be applied to the partial differential operator ∂µ. Using (11.110) one can state

˙ ∂0 ∂3 ∂1 + i∂2 ∂αβ =  − −  . (11.116) ∂1 i∂2 ∂0 + ∂3 − − Similarly, (11.114) yields

∂0 + ∂3 ∂1 + i∂2 ∂αβ˙ =   . (11.117) ∂1 i∂2 ∂0 ∂3 − − 370 Spinor Formulation

σµ in Tensor Notation µ We want to develop now the tensor notation for σµ (11.26) and its contravariant analogue σ

1 0 0 1 0 i 1 0 σ = , , , µ  0 1   1 0   i −0   0 1  − 1 0 0 1 0 i 1 0 σµ = , , , (11.118)  0 1   1− 0   i 0   −0 1  − − µ For this purpose we consider first the transformation behaviour of σ and σµ. We will obtain the µ µ transformation behaviour of σ , σµ building on the known transformation behaviour ofγ ˜ . This is µ µ possible sinceγ ˜ can be expressed through σ , σµ. Comparision of (10.229) and (11.118) yields

0 σµ γ˜µ =   . (11.119) σµ 0

Using (11.15) one can write µ 1 σµ =  (σ )∗ − (11.120) and, hence, µ µ 0 σ γ˜ =  µ 1  . (11.121)  (σ )∗ − 0 One expects then that the transformation properties of σµ should follow from the transformation properties established already for γµ [c.f. (10.243)]. Note that (10.243) holds independently of the representation chosen, i.e., holds also in the chiral representation. To obtain the transformation properties of σµ we employ then (10.243) in the chiral representation η µ expressing (L ξ) by (11.18) and γ by (11.121). Equation (10.243) reads then S µ 1 a 0 0 σ a− 0 ν  1   µ 1   1  L µ = 0 a∗− (σ )∗− 0 0 a∗− 0 σν  ν 1  . (11.122) (σ )∗− 0 The l.h.s. of this equation is

µ 1 1 0 aσ − (a∗)−  ν  µ 1 1  L µ (11.123) a∗(σ )∗− a− 0 and, hence, one can conclude

µ 1 1 ν ν aσ − (a∗)−  L µ = σ (11.124) µ 1 1 ν ν 1 a∗(σ )∗− a− L µ = (σ )∗− . (11.125)

Equation (11.125) is equivalent to

µ 1 1 ν ν a∗(σ )∗− a−  L µ = (σ )∗ (11.126) which is the complex conjugate of (11.124), i.e., (11.125) is equivalent to (11.124). Hence, (11.124) constitutes the essential transformation property of σµ and will be considered further. 11.4: Spinor Tensors 371

One can rewrite (11.124) using (11.16, 11.2)

1 1 1 1 ~z ~σ 1 ~z  1~σ  1 ~z ~σ − (a∗)−  = − e− 2 ∗· ∗  = e− 2 ∗· − ∗ = e 2 ∗· . (11.127)

T Exploiting the hermitian property of ~σ, e.g., (~σ ∗) = ~σ yields using (11.2)

1 T 1 T 1 1 ~z ∗ (~σ ∗) ~z ∗ ~σ ∗ T − (a∗)−  = e 2 · = h e 2 · i = [a∗] . (11.128)

One can express, therefore, equation (11.124)

µ T ν ν a σ [a∗] L µ = σ . (11.129)

µ T We want to demonstrate now that the expression aσ [a∗] is to be interpreted as the transform of σµ under Lorentz transformations. In fact, under rotations the Pauli matrices transform like (j = 1, 2, 3) T † σj a(iϑ~) σj a(iϑ~) = a(iϑ~) σj a∗(iϑ~) . (11.130) −→     We argue in analogy to the logic applied in going from (11.107) to (11.108) that the same trans- formation behaviour applies then for general Lorentz transformations, i.e., transformations (11.2, 11.4) with ~w = 0. One can, hence, state that σµ in a new reference frame is 6 µ µ σ0 = a σ a† (11.131) where a is given by (11.2, 11.4). This transformation behaviour, according to (11.102, 11.103) identifies σµ as a tensor of type tαβ˙ . It holds according to (11.118)

(σµ)11˙ (σµ)12˙ ! = (σµ)21˙ (σµ)22˙

  1 0 0 1 0 i 1 0    ,  −  ,   ,  −  ,  (11.132)  0 1 1 0 i 0 0 1   − −   µ = 0 µ = 1 µ = 2 µ = 3  | {z } | {z } | {z } | {z } and

11˙ 12˙ (σµ) (σµ)  µ µ  = (σ )21˙ (σ )22˙

  1 0 0 1 0 i 1 0    ,   ,  −  ,   ,  . (11.133)  0 1 1 0 i 0 0 1   −   µ = 0 µ = 1 µ = 2 µ = 3  | {z } | {z } | {z } | {z } Combining (11.129, 11.131) one can express the transformation behaviour of σµ in the succinct form ν µ ν L µ σ0 = σ . (11.134) 372 Spinor Formulation

Inverting contravariant and covariant indices one can also state

µ Lν σµ0 = σν . (11.135) µ This is the property surmised already above [cf. (11.29)]. We can summarize that σ and σµ transform like a 4–vector, however, the transformation is inverse to that of ordinary 4–vectors. Each of the 4 4 = 16 matrix elements in (11.132) and (11.133) is characterized through a 4-vector index µ, µ =× 0, 1, 2, 3 as well as through two spinor indices αβ˙. We want to express now σµ and σµ also with respect to the 4–vector index µ in spinor form employing (11.114). This yields 1 0 0 1       σ0 + σ3 σ1 + iσ2 0 0 0 0 σαβ˙ =   = 2 (11.136) σ1 iσ2 σ0 σ3  0 0 0 0  − −        1 0 0 1  where on the rhs. the submatrices correspond to tαβ˙ spinors. We can, in fact, state

11˙ 12˙ 11˙ 12˙ (σ11˙ ) (σ11˙ ) (σ12˙ ) (σ12˙ )  ˙ ˙ ! ˙ ˙ !  γδ˙ 21 22 21 22 (σ11˙ ) (σ11˙ ) (σ12˙ ) (σ12˙ ) σ ˙ =   . (11.137)  αβ  11˙ 12˙ 11˙ 12˙   (σ21˙ ) (σ21˙ ) ! (σ22˙ ) (σ22˙ ) !   21˙ 22˙ 21˙ 22˙   (σ21˙ ) (σ21˙ ) (σ22˙ ) (σ22˙ )  Equating this with the r.h.s. of (11.136) results in the succinct expression

1 γδ˙ σ = δ δ . (11.138) 2  αβ˙  αγ β˙δ˙

Note that all elements of σαβ˙ are real and that there are only four non-vanishing elements. In (11.138) the ‘inner’ covariant spinor indices, i.e., α, β˙, account for the 4–vector index µ, whereas the ‘outer’ contravariant spinor indices. i.e., γ, δ˙, account for the elements of the individual Pauli µ matrices. We will now consider the representation of σ , σµ in which the contravariant indices are moved ‘inside’, i.e., account for the 4-vector µ, and the covariant indices are moved outside. The αβ˙ desired change of representation(σµ) (σµ)αβ˙ corresponds to a transformation of the basis of spin states −→ f g f     =    (11.139) g −→ f g − and, hence, corresponds to the transformation

11˙ 12˙ (σµ)11˙ (σµ)12˙ 0 1 (σµ) (σµ) ! 0 1   =   21˙ 22˙  −  (11.140) (σµ) ˙ (σµ) ˙ 1 0 (σ ) (σ ) 1 0 21 22 − µ µ 1 αβ˙ where we employed the expressions (11.12) for  and − . Using (11.110) to express σ in terms of σµ yields together with (5.224) 0 0 0 0       ˙ σ0 σ3 σ1 + iσ2 0 1 1 0 σαβ =  −  = 2 − (11.141) σ1 iσ2 σ0 + σ3  0 1 1 0  − −   −      0 0 0 1  11.4: Spinor Tensors 373 and, employing then transformation (11.140) to transform each of the four submatrices, which in αβ˙ (11.137) are in a basis ( ) to a basis ( ) ˙ results in ··· ··· αβ

11˙ 11˙ 12˙ 12˙  (σ )11˙ (σ )12˙ ! (σ )11˙ (σ )12˙ !  11˙ 11˙ 12˙ 12˙ ˙ (σ ) ˙ (σ ) ˙ (σ ) ˙ (σ ) ˙ σαβ =  21 22 21 22   γδ˙  21˙ 21˙ 22˙ 22˙   (σ )11˙ (σ )12˙ ! (σ )11˙ (σ )12˙ !   21˙ 21˙ 22˙ 22˙   (σ )21˙ (σ )22˙ (σ )21˙ (σ )22˙  1 0 0 1   0 0   0 0   = 2 . (11.142)  0 0 0 0         1 0 0 1 

This can be expressed

1 αβ˙ σ = δαγδ ˙ ˙ . (11.143) 2  γδ˙ βδ Combined with (11.138) one can conclude that the following property holds

γδ˙ 1 1 αβ˙ σ ˙ = σ = δαγδ ˙ ˙ . (11.144) 2  αβ 2  γδ˙ βδ

The Dirac Matrices γµ in spinor notation

We want to express now the Dirac matricesγ ˜µ in spinor form. For this purpose we start from the µ µ µ 1 expression (11.121) ofγ ˜ . This expression implies that the element ofγ ˜ given by  σ − is in the µ µ αβ˙ basis ˙ whereas the element ofγ ˜ given by σ is in the basis . Accordingly, we write |αβ |

µ αβ˙ µ 0 (σ ) ! γ˜ = µ . (11.145) ((σ )αβ˙ )∗ 0

Let Aµ be a covariant 4–vector. One can write then the scalar product using (11.115)

µ αβ˙ µ 0 (σ ) Aµ ! γ˜ Aµ = µ ((σ )αβ˙ )∗Aµ 0 ˙ 0 1 (σ )αβAγδ˙ = 2 γδ˙ ! . (11.146) 1 γδ˙ 2 ((σ )αβ˙ )∗Aγδ˙ 0

Exploiting the property (11.144) results in the simple relationship

αβ˙ µ 0 A γ˜ Aµ = ! . (11.147) Aαβ˙ 0 374 Spinor Formulation

11.5 Lorentz Invariant Equations in Spinor Form

Dirac Equation (11.147) allows us to rewrite the Dirac equation in the chiral representation (10.226)

αβ˙ µ µ 0 i ∂ µ µ i γ ∂µΨ(˜ x ) = ! Ψ(˜ x ) = m Ψ(˜ x ) . (11.148) i ∂αβ˙ 0

Employing Ψ(˜ xµ) in the form (11.99) yields the Dirac equation in spinor form

αβ˙ α i ∂ χβ˙ = m φ (11.149) α i ∂αβ˙ φ = m χβ˙ . (11.150) The simplicity of this equation is striking.