Let (V,g)(V,g) be a finite-dimensional real or complex with basis eae_a, and let c:Cl⁡(V,g)→End⁡(Δ)c:\operatorname{Cl}(V,g)\to\operatorname{End}(\Delta) be a written in a basis of Δ\Delta. The gamma matrices are

γa=c(ea).\gamma_a=c(e_a).

Using the geometric Clifford convention v2=−g(v,v)1v^2=-g(v,v)1, they satisfy

γaγb+γbγa=−2gabI.\gamma_a\gamma_b+\gamma_b\gamma_a=-2g_{ab}I.

Thus gamma matrices are the basis-dependent matrices of , not additional invariant geometric objects.

Raised indices

Writing gabg^{ab} for the inverse matrix, set γa=gabγb\gamma^a=g^{ab}\gamma_b. Contracting gamma matrices with the components of a vector or covector is the .

Minkowski-signature convention

For the Lorentzian knowls in this collection, take

η=diag⁡(−1,+1,+1,+1).\eta=\operatorname{diag}(-1,+1,+1,+1).

Under the geometric convention used here,

γ02=+I,γj2=−I(j=1,2,3).\gamma_0^2=+I, \qquad \gamma_j^2=-I\quad(j=1,2,3).

Many physics texts instead use the opposite metric signature ηphys=diag⁡(+1,−1,−1,−1)\eta_{\mathrm{phys}}=\operatorname{diag}(+1,-1,-1,-1) together with

{γμ,γν}=+2ηphysμνI.\{\gamma^\mu,\gamma^\nu\}=+2\eta_{\mathrm{phys}}^{\mu\nu}I.

With ηphys=−η\eta_{\mathrm{phys}}=-\eta, this has the same numerical right-hand side as the convention in the core. If one changes only the Clifford sign while holding the metric fixed, multiplying all complex gamma matrices by ii passes between the two relations. Metric signature and Clifford sign must therefore be recorded separately.

Worked example: the Dirac representation

Work on R1,3\mathbb R^{1,3} with coordinates x0=t,x1,x2,x3x^0=t,x^1,x^2,x^3, metric η=diag⁡(−1,1,1,1)\eta=\operatorname{diag}(-1,1,1,1), and complex spinor module Δ=C4\Delta=\mathbb C^4. Write InI_n for the n×nn\times n identity matrix and i2=−1i^2=-1. The Pauli matrices used in this construction are

σ1=(0110),σ2=(0−ii0),σ3=(100−1).\sigma^1=\begin{pmatrix}0&1\\1&0\end{pmatrix},\qquad \sigma^2=\begin{pmatrix}0&-i\\i&0\end{pmatrix},\qquad \sigma^3=\begin{pmatrix}1&0\\0&-1\end{pmatrix}.

Define the raised-index matrices by 2×22\times2 blocks:

γ0=(I200−I2),γj=(0σj−σj0)(j=1,2,3).\gamma^0=\begin{pmatrix}I_2&0\\0&-I_2\end{pmatrix},\qquad \gamma^j=\begin{pmatrix}0&\sigma^j\\-\sigma^j&0\end{pmatrix} \quad(j=1,2,3).

Thus all four matrices are explicitly

γ0=(1000010000−10000−1),γ1=(000100100−100−1000),\gamma^0=\begin{pmatrix} 1&0&0&0\\0&1&0&0\\0&0&-1&0\\0&0&0&-1 \end{pmatrix},\qquad \gamma^1=\begin{pmatrix} 0&0&0&1\\0&0&1&0\\0&-1&0&0\\-1&0&0&0 \end{pmatrix},
γ2=(000−i00i00i00−i000),γ3=(0010000−1−10000100).\gamma^2=\begin{pmatrix} 0&0&0&-i\\0&0&i&0\\0&i&0&0\\-i&0&0&0 \end{pmatrix},\qquad \gamma^3=\begin{pmatrix} 0&0&1&0\\0&0&0&-1\\-1&0&0&0\\0&1&0&0 \end{pmatrix}.

To check the relations directly, matrix multiplication gives (σj)2=I2(\sigma^j)^2=I_2 and

σ1σ2=iσ3=−σ2σ1,σ2σ3=iσ1=−σ3σ2,σ3σ1=iσ2=−σ1σ3.\sigma^1\sigma^2=i\sigma^3=-\sigma^2\sigma^1,\qquad \sigma^2\sigma^3=i\sigma^1=-\sigma^3\sigma^2,\qquad \sigma^3\sigma^1=i\sigma^2=-\sigma^1\sigma^3.

Consequently σjσk+σkσj=2δjkI2\sigma^j\sigma^k+\sigma^k\sigma^j=2\delta_{jk}I_2, where δjk\delta_{jk} is 11 when j=kj=k and 00 otherwise. Block multiplication now gives

(γ0)2=I4,γ0γj=(0σjσj0)=−γjγ0,(\gamma^0)^2=I_4,\qquad \gamma^0\gamma^j=\begin{pmatrix}0&\sigma^j\\\sigma^j&0\end{pmatrix} =-\gamma^j\gamma^0,
γjγk=−(σjσk00σjσk).\gamma^j\gamma^k= -\begin{pmatrix}\sigma^j\sigma^k&0\\0&\sigma^j\sigma^k\end{pmatrix}.

These identities prove all the Clifford anticommutation relations:

{γμ,γν}:=γμγν+γνγμ=−2ημνI4(μ,ν=0,1,2,3).\{\gamma^\mu,\gamma^\nu\} :=\gamma^\mu\gamma^\nu+\gamma^\nu\gamma^\mu =-2\eta^{\mu\nu}I_4\quad(\mu,\nu=0,1,2,3).

In particular, the spatial squares are −I4-I_4. A concrete mixed product is

γ1γ2=−i(σ300σ3)=diag⁡(−i,i,−i,i),γ2γ1=−γ1γ2.\gamma^1\gamma^2 =-i\begin{pmatrix}\sigma^3&0\\0&\sigma^3\end{pmatrix} =\operatorname{diag}(-i,i,-i,i),\qquad \gamma^2\gamma^1=-\gamma^1\gamma^2.

To connect this example with the geometric definition, take eμ=∂/∂xμe_\mu=\partial/\partial x^\mu and set c(eμ)=γμ=ημνγνc(e_\mu)=\gamma_\mu=\eta_{\mu\nu}\gamma^\nu, summing over ν\nu. Thus γ0=−γ0\gamma_0=-\gamma^0 and γj=γj\gamma_j=\gamma^j. Using the metric identification of covectors with vectors, c(dxμ)=c((dxμ)♯)=γμc(dx^\mu)=c((dx^\mu)^\sharp)=\gamma^\mu; the verified relations therefore give a representation of the geometric Clifford algebra. With ηphys=−η\eta_{\mathrm{phys}}=-\eta, the same numerical raised-index matrices satisfy {γμ,γν}=2ηphysμνI4\{\gamma^\mu,\gamma^\nu\}=2\eta_{\mathrm{phys}}^{\mu\nu}I_4. No factor of ii is needed when changing both metric signature and Clifford sign in this way. Lowered indices must, however, use the metric of the chosen convention. The corresponding differential equation is expanded in the knowl.

Change of basis and spin covariance

Changing the basis of the spinor module conjugates every gamma matrix simultaneously. A spin transformation ss satisfies

ρ(s)c(v)ρ(s)−1=c(svs−1),\rho(s)c(v)\rho(s)^{-1}=c(svs^{-1}),

so the matrices transform compatibly with the orthogonal action on VV. Different-looking gamma-matrix realizations can therefore describe equivalent Clifford representations.

Related constructions

In even dimension, the normalized product of all gamma matrices represents the . The normalization is convention-dependent, whereas the resulting grading of the complex spin representation is invariant.

References
  1. H. Blaine Lawson Jr. and Marie-Louise Michelsohn, Spin Geometry, Princeton University Press, 1989. DOI record. Relevant: Chapter I.
  2. Daniel S. Freed, Five Lectures on Supersymmetry, American Mathematical Society, 1999. Relevant: Lectures 1–2.
  3. Michael E. Peskin and Daniel V. Schroeder, An Introduction to Quantum Field Theory, Addison–Wesley, 1995. Relevant: Sections 3.1–3.4 for the common physics sign convention.