3.2, The Dirac Equation 43
To see that the Dirac equation implies the Klein-Gordon equation, act on the
left with (-i7"9, —_m):
0 = (~in*O, — m)(i7"O, — mp
= ("9 Oy + m? Jes
= (bfyH, 0,0, + mw
= (8 + m?)y.
To write down a Lagrangian for the Dirac theory, we must figure out how
to multiply two Dirac spinors to form a Lorentz scalar. The obvious guess,
wy, does not work. Under a Lorentz boost this becomes wt‘, Ay qd; if the
boost matrix were unitary, we would have Al = Aj and everything would be
fine. But A. is not unitary, because the generators (3.26) are not Hermitian.
The solution is to define _
=yty. (3.32)
Under an infinitesimal Lorentz transformation parametrized by wiv, we have
wh wt (1+ dup. (S"”)')7°. The sum over ys and v has six distinct nonzero
terms. In the rotation terms, where p and v are both nonzero, (S#”)t = St”
and S#” commutes with 7°. In the boost terms, where jor v is 0, (S#”)7 =
—(S#”) but S#” anticommutes with 7°. Passing the 7° to the left therefore
removes the dagger from S$”, yielding the transformation law
Br vAy, (3.33)
and therefore the quantity 7y is a Lorentz scalar. Similarly you can show
(with the aid of (3.29)) that wy"v is a Lorentz vector.
The correct, Lorentz-invariant Dirac Lagrangian is therefore
iy", — mv. (3.34)
The Euler-Lagrange equation for & (or w") immediately yields the Dirac equa-
tion in the form (3.31); the Euler-Lagrange equation for y gives the same
equation, in Hermitian-conjugate form:
Loirac
ty" — mip (3.35)
Weyl Spinors
From the block-diagonal form of the generators (3.26) and (3.27), it is apparent
that the Dirac representation of the Lorentz group is reducible.t We can form
two 2-dimensional representations by considering each block separately, and
writing
b= Ge (3.36)
YR,
entation of the gamma matrices, ibility
this is essentially the reason for using the chiral representation.
4If we had used a different repr
would not be mani