Dirac Spinors as the Minimal Left Ideal of a Clifford Algebra


The energy mass shell equation is

(E0c)2=(Ec)2−||p→||2

Where E0=mc2 is the rest energy, E is the total energy and p→ is the momentum vector.

Replacing energy and momentum with their respective wave equation operators E=iℏ∂∂t and p=−iℏ⟨∂∂x,∂∂y,∂∂z⟩ gives the Klein-Gordon equation.

(E0c)2=ℏ2(−1c22∂2∂t2+∂2∂x2+∂2∂y2+∂2∂z2)
−1ℏ2(E0c)2=(1c22∂2∂t2−∂2∂x2−∂2∂y2−∂2∂z2)

Assume there exists some expression that squares to the operators on the right.

(γt1c∂∂t+γx∂∂x+γy∂∂y+γz∂∂z)2=1c22∂2∂t2−∂2∂x2−∂2∂y2−∂2∂z2

Multiplying the left side out gives the below

γt21c2∂2∂t2+γtγx1c∂2∂t∂x+γtγy1c∂2∂t∂y+γtγz1c∂2∂t∂z
+γxγt1c∂2∂x∂t+γx2∂2∂x2+γxγy∂2∂x∂y+γxγz∂2∂x∂z
+γyγt1c∂2∂y∂t+γyγx∂2∂y∂x+γy2∂2∂y2+γyγz∂2∂y∂z
+γzγt1c∂2∂z∂t+γzγx∂2∂z∂x+γzγy∂2∂z∂y+γz2∂2∂z2
=1c22∂2∂t2−∂2∂x2−∂2∂y2−∂2∂z2

Assuming the derivative operators are commutative then then grouping

γt21c2∂∂t2+γx2∂∂x2+γy2∂∂y2+γz2∂∂z2
+(γtγx+γxγt)1c∂∂t∂x
+(γtγy+γyγt)1c∂∂t∂y
+(γtγx+γzγt)1c∂∂t∂z
+(γxγy+γyγx)∂∂x∂y
+(γxγz+γzγx)∂∂x∂z
+(γyγz+γzγy)∂∂y∂z
=1c22∂2∂t2−∂2∂x2−∂2∂y2−∂2∂z2

Assuming the derivatives are independent the only solution can be that

γt2=1
γx2=−1
γy2=−1
γz2=−1
γtγx+γxγt=0
γtγy+γyγt=0
γtγx+γzγt=0
γxγy+γyγx=0
γxγz+γzγx=0
γyγz+γzγy=0

These just so happen to be basis elements of a Clifford algebra Cl(V,Q) over a vector space with basis elements

tγt+xγx+yγy+zγz∈V=ℝ4

and quadratic form

Q(tγt+xγx+yγy+zγz)=t2−x2−y2−z2

By "square rooting" both sides of the Klein-Gordon equation the resulting equation, the Dirac equation, can then be written as

±iℏE0c=γt1c∂∂t+γx∂∂x+γy∂∂y+γz∂∂z

(The choice of plus minus will just switch matter and antimatter.) After applying the operator to the wave field ψ(t,x,y,z) we get

±iℏE0cψ(t,x,y,z)=γt1c∂ψ(t,x,y,z)∂t+γx∂ψ(t,x,y,z)∂x+γy∂ψ(t,x,y,z)∂y+γz∂ψ(t,x,y,z)∂z

Whatever the elements of the field ψ(t,x,y,z) are multiplied on the left by the basis elements of the Clifford algebra they stay in the same group. This behaviour is a left ideal over the Clifford algebra as a ring. A left ideal is a subset S⊂Cl(V,Q) where ∀(a∈Cl(V,Q)).∀(b∈S).ab∈S it says in the subset when left multiplied by anything.

We can construct a minimal left ideal from any primitive idempotent p=p2. The choice of idempotent doesn't matter as all minimal left ideals are isomorphic. We will arbitrarily choose

p=12(1+γt)
and
p2=14(1+2γt+γt2)=14(2+2γt)=12(1+γt)=p

Then to generate the minimal left ideal we project all elements a∈CL(V,Q) to ap

Any element x can be written in its 16 dimensional basis:

a=
X
+Xtγt
+Xxγx
+Xyγy
+Xzγz
+Xtxγtγx
+Xtyγtγy
+Xtzγtγz
+Xxyγxγy
+Xxzγxγz
+Xyzγyγz
+Xtxyγtγxγy
+Xtxzγtγxγz
+Xtyzγtγyγz
+Xtyzγxγyγz
+Xtxyzγtγxγyγz

After multiplying by p

ap=a12(1+γt)=12(
X(1+γt)
+Xt(γt+1)
+Xx(γx−γtγx)
+Xy(γy−γtγy)
+Xz(γz−γtγz)
+Xtx(γtγx−γx)
+Xty(γtγy−γy)
+Xtz(γtγz−γz)
+Xxy(γxγy+γtγxγy)
+Xxz(γxγz+γtγxγz)
+Xyz(γyγz+γtγyγz)
+Xtxy(γtγxγy+γxγy)
+Xtxz(γtγxγy+γxγz)
+Xtyz(γtγyγz+γyγz)
+Xxyz(γxγyγz−γtγxγyγz)
+Xtxyz(γtγxγyγz−γxγyγz))

Then grouping them makes an 8 dimensional basis:

ap=12(
(X+Xt)(1+γt)
+(Xx−Xtx)(γx−γtγx)
+(Xy−Xty)(γy−γtγy)
+(Xz−Xtz)(γz−γtγz)
+(Xxy+Xtxy)(γxγy+γtγxγy)
+(Xxz+Xtxz)(γxγz+γtγxγz)
+(Xyz+Xtyz)(γyγz+γtγyγz)
+(Xxyz−Xtxyz)(γxγyγz−γtγxγyγz))

Then if we represent this as a vector ⟨A,B,C,D,E,F,G,H⟩. These elements of a minimal left ideal are the elements of the field ψ(t,x,y,z).

ψ(t,x,y,z)=12(
A(1+γt)
+B(γx−γtγx)
+C(γy−γtγy)
+D(γz−γtγz)
+E(γxγy+γtγxγy)
+F(γxγz+γtγxγz)
+G(γyγz+γtγyγz)
+H(γxγyγz−γtγxγyγz))

Then substituting into the Dirac equation

±iℏE0cψ(t,x,y,z)=γt1c∂ψ(t,x,y,z)∂t+γx∂ψ(t,x,y,z)∂x+γy∂ψ(t,x,y,z)∂y+γz∂ψ(t,x,y,z)∂z

with the subsituted in vectors

±iℏE0c[ABCDEFGH]=γt1c∂∂t[ABCDEFGH]+γx∂∂x[ABCDEFGH]+γy∂∂y[ABCDEFGH]+γz∂∂z[ABCDEFGH]

We can find how the basis elements γt,γx,γy,γz act on elements of the minimal left ideal, or ψ, when left multiplied..

γt(A(1+γt)
B(γx−γtγx)+C(γy−γtγy)+D(γz−γtγz)
E(γxγy+γtγxγy)+F(γxγz+γtγxγz)+G(γyγz+γtγyγz)
H(γxγyγz−γtγxγyγz))
=
A(γt+1)
B(γtγx−γx)+C(γtγy−γy)+D(γtγz−γz)
E(γtγxγy+γxγy)+F(γtγxγz+γxγz)+G(γtγyγz+γyγz)
H(γtγxγyγz−γxγyγz)
=[A−B−C−DEFG−H]

So the matrix representation is

γt↦[100000000−100000000−100000000−100000000100000000100000000100000000−1]

Doing it for γx.

γx(A(1+γt)
B(γx−γtγx)+C(γy−γtγy)+D(γz−γtγz)
E(γxγy+γtγxγy)+F(γxγz+γtγxγz)+G(γyγz+γtγyγz)
H(γxγyγz−γtγxγyγz))
=
A(γx−γtγx)
B(−1−γt)+C(γxγy+γtγxγy)+D(γxγz+γtγxγz)
E(−γy+γtγy)+F(−γz+γtγz)+G(γxγyγz−γtγxγyγz)
H(−γyγz−γtγyγz)
=[−BA−E−FCD−HG]

So the matrix representation is

γx↦[0−1000000100000000000−100000000−10000100000000100000000000−100000010]

Doing it for γy.

γy(A(1+γt)
B(γx−γtγx)+C(γy−γtγy)+D(γz−γtγz)
E(γxγy+γtγxγy)+F(γxγz+γtγxγz)+G(γyγz+γtγyγz)
H(γxγyγz−γtγxγyγz))
=
A(γy−γtγy)
B(−γxγy−γtγxγy)+C(−1−γt)+D(γyγz+γtγyγz)
E(γx−γtγx)+F(−γxγyγz+γtγxγyγz)+G(−γz+γtγz)
H(γxγz+γtγxγz)
=[−CEA−G−BHD−F]

So the matrix representation is

γy↦[00−1000000000100010000000000000−100−1000000000000010001000000000−100]

Doing it for γz.

γz(A(1+γt)
B(γx−γtγx)+C(γy−γtγy)+D(γz−γtγz)
E(γxγy+γtγxγy)+F(γxγz+γtγxγz)+G(γyγz+γtγyγz)
H(γxγyγz−γtγxγyγz))
=
A(γz−γtγz)
B(−γxγz−γtγxγz)+C(−γyγz−γtγyγz)+D(−1−γt)
E(γxγyγz−γtγxγyγz)+F(γx−γtγx)+G(γy−γtγy)
H(−γxγy−γtγxγy)
=[−DFGA−H−B−CE]

So the matrix representation is

γz↦[000−100000000010000000010100000000000000−10−100000000−10000000001000]

Now we have some choices for how i2=−1 can be represented in the Clifford algebra. The value i can of be γxγy,γxγz,γyγz as all 3 of those bivectors square to −1. If we arbitrarily choose i=γyγx, and be careful to consider that i isn't commutative with everything, we can find its action on the spinor.

(A(1+γt)
B(γx−γtγx)+C(γy−γtγy)+D(γz−γtγz)
E(γxγy+γtγxγy)+F(γxγz+γtγxγz)+G(γyγz+γtγyγz)
H(γxγyγz−γtγxγyγz))i
=
A(−γxγy−γtγxγy)
B(γy−γtγy)+C(−γx+γtγx)+D(−γxγyγz+γtγxγyγz)
E(1+γt)+F(γyγz+γtγyγz)+G(−γxγz−γtγxγz)
H(γz+γtγz)

So the mapping is

i[ABCDEFGH]=[E−CBH−A−GF−D]

And if we change the vector representation in ℝ8 to ℂ4 the mapping still holds, and left multiplication with i acts as the complex numbers do.

i[E+iAF+iGH+iDB+iC]=[−A+iE−C+iB−D−iH−G+iF]

the corresponding matricies map to

γt↦[1000010000−10000−1],γx↦[000−i00−i00−i00−i000],γy↦[000−100100−1001000],γz↦[00−10000−110000100]

and these just so happen to be a basis for the Dirac matricies, showing their derivation.