5 Spacetime geometry

We introduce the Minkowski metric and the Lorentz group, and show that the algebra is neatly represented by \(2\times 2\) complex matrices.

Key takeaways:

  1. The Pauli matrices form a basis for \(2\times 2\) complex Hermitian matrices (proposition 5.2).
  2. Representation of \((3+1)\)-spacetime using Hermitian-matrixs (definition 5.2).
  3. In this representation, determinant (of the \(2\times 2\) Hermitian matrix) correspond to the Minkowski norm in \(\mathbb R^4\) (proposition 5.3).
  4. Conjugation in the representation corresponds to a linear transformation in spacetime (theorem 5.1).
  5. We are interested in the Lorentz group (definition 5.4) which preserve the Minkowski norm. Such (linear) Lorentz transforms are represented the conjugate action of \(\mathrm{SL}(2, \mathbb C)\), or \(2\times 2\) complex matrices with determinant 1 (proposition 5.4).
    • Rotations are represented by unitary conjugation \(SU(2)\) (theorem 5.2).
    • Lorentz boosts are represented by Hermitian conjugation \(\mathcal H_2\cap \mathrm{SL}(2)\) (theorem 5.3).

In this section, we work with \(2\times 2\) complex matrices. Some terminology for reference:

  1. \(\mathrm{GL}(2, \mathbb C)\) denotes the set of \(2\times 2\) complex invertible matrices.
  2. \(\mathrm{SL}(2, \mathbb C)\subsetneq \mathrm{SL}(2, \mathbb C)\) denotes the set of \(2\times 2\) matrices with \(\det A=1\).
  3. \(U(2)\) denotes the set of \(2\times 2\) unitary matrices satisfying \(AA^\dagger= A^\dagger A = \mathbf 1\). The intersection \(U(2)\cap \mathrm{SL}(2, \mathbb C)\) is denoted \(SU(2)\), or the special unitary group.
  4. \(\mathcal H_2\) denotes the set of \(2\times 2\) Hermitian matrices satisfying \(A^\dagger= A\), where \(A^\dagger\) is the conjugate transpose.

Matrix representation of spacetime

In this section, we define a representation of spacetime via a bijection between \(\mathbb R^4\) and \(\mathcal H_2\) (definition 5.2) and introduce the Minkowski metric 5.3. the space \(\mathbb R^4\cong \mathbb R\times \mathbb R^3\) will be understood as \(4\)-dimensional spacetime.

Both unitary matrices and Hermitian matrices are normal thus subject to the spectral theorem, so one may think of them as a diagonal matrix of eigenvalues in some orthonormal basis. In particular:

  1. Hermitian matrices have real eigenvalues.
  2. Unitary matrices have complex eigenvalues with unit norm of the form \(e^{i\theta}\).

An immediate corollary of the observation above is

Proposition 5.1 Every unitary \(U=\exp(iH)\) is the complex exponential of some Hermitian matrix \(H\).

Proceeding, recall the Pauli matrices \[ \mathbf 1 = \begin{bmatrix} 1 \\ & 1\end{bmatrix}, \quad \sigma_x = \begin{bmatrix} & 1 \\ 1 \end{bmatrix}, \quad \sigma_y = \begin{bmatrix} & -i \\ i \end{bmatrix}, \quad \sigma_z = \begin{bmatrix} 1 \\ & -1\end{bmatrix} \] The space \(\mathcal H_2\) of \(2\times 2\) Hermitian matrices is a vector space, and it comes equipped with the following inner product:

Definition 5.1 (Hilbert-Schmidt inner product) The Hilbert-Schmidt inner product defined on the space of linear operators over a finite-dimensional Hilbert space of dimension \(d\) is
\[ \langle A, B\rangle= \dfrac 1 d \mathrm{Tr}(A^\dagger B) \] In a basis representation, this corresponds to the flattened vector inner product.

The Pauli matrices are special because they form an orthonormal basis for \(\mathcal H_2\):

Proposition 5.2 The Pauli matrices are involutary \(\sigma_i^2 = \mathbf 1\) and form an orthonormal basis for \(\mathcal H_2\) under the Hilbert-Schmidt inner product.

Proof: Direct computation, one can verify.

Definition 5.2 (representation of spacetime) We identify points in \(\mathbb R^4\) \[ x=(x_t, x_1, x_2, x_3)\in \mathbb R^4 \]
with \(2\times 2\) Hermitian matrices \[ \hat x = x_t\mathbf 1 + x_1\sigma_x + x_2\sigma_y + x_3\sigma_3 = \begin{bmatrix} x_t + x_3 & x_1 - i x_2 \\ x_1 + ix_2 & x_t - x_3 \end{bmatrix}\in \mathcal H_2 \] The representation is invertible by Fourier decomposition onto the Pauli basis using 5.1: \[ x_t = \langle\hat x, \mathbf 1\rangle= \dfrac 1 2 \mathrm{tr}(\hat x\mathbf 1), \quad x_{i\in \{1, 2, 3\}} = \langle\hat x, \sigma_i\rangle= \dfrac 1 2 \mathrm{tr}(\hat x\sigma_i) \]

Definition 5.3 (Minkowski metric) The Minkowski metric over \(\mathbb R^4\) is the bilinear map \(\langle\, \cdot\, , \, \cdot\, \rangle_M:\mathbb R^4\times \mathbb R^4\to \mathbb R\). \[ \langle x, y\rangle_M = x_ty_t - x_1y_1 - x_2y_2 - x_3y_3 \] We denote the Minkowski norm it induces by \(\|\, \cdot\|^2_M:\mathbb R^4\to \mathbb R\), defined by \(\|x\|_M^2 = \langle x, x\rangle_M\).

One purpose of the definition 5.2 is the following

Proposition 5.3 Under the Pauli identification of \(\mathbb R^4\) \[ \det \hat x = \langle x, x\rangle_M \]

Proof: Direct computation: \(\det \hat x = x_t^2 - x_1^2 - x_2^2 - x_3^2\).

The Lorentz group

We wish to investigate linear transformations on \(\mathbb R^4\) which leave the Minkowski metric invariant, since these transformations form the Lorentz group.

Definition 5.4 (Lorentz group) The Lorentz group \(\mathcal L\) is the subgroup of \(\mathrm{GL}(4, \mathbb R)\) (group of invertible linear transformations on \(\mathbb R^4\)) which preserve the Minkowski metric \[ A\in \mathcal L\iff \forall x, y\in \mathbb R^4, \langle x, y\rangle_M = \langle Ax, Ay\rangle \iff \forall x\in \mathbb R^4, \|x\|^2_M = \|Ax\|^2_M \]

Recall that inverse-conjugating any matrix \(A\mapsto BAB^{-1}\) will not change the eigenvalues of \(A\). In particular, inverse-conjugating \(\hat x\in \mathcal H_2\) by any \(2\times 2\) invertible matrix yields another element of \(\mathcal H_2\); since \(\mathcal H_2\) represent points in spacetime, it is natural to ask the \(H\mapsto BHB^{-1}\) induces on \(\mathbb R^4\).

Theorem 5.1 (conjugation-induced linear transform) conjugation in \(\mathcal H_2\) yields a linear transformation in \(\mathbb R^4\): Equivalently, the map \(A:\mathbb R^4\to \mathbb R^4\) which makes the diagram below commute is a linear operator.

In particular, the representation of \(\hat A\) as a matrix in \(\mathbb R^4\) is \[ A_{ij} = (Ae_j)_i = \langle\hat A \sigma_j \hat A^\dagger, \sigma_i \rangle = \dfrac 1 2 \mathrm{tr}\left(\hat A \sigma_j \hat A^\dagger\sigma_i\right) \tag{5.1} \]

Proof: Fixing \(\hat A:\mathcal C^2\to \mathcal C^2\) and let \(x, y\in \mathbb R^4\). Let \(\sigma_i\) be any Pauli basis matrix, we show that \(A\) is linear: \[\begin{align} \left[A(x+\alpha y)\right]_i &= \langle\hat A(\hat x + \alpha \hat y)\hat A^\dagger, \sigma_i\rangle = \dfrac 1 2 \mathrm{tr}\left(\hat A (\hat x + \alpha \hat y)\hat A^\dagger\sigma_i \right) \\ &= \dfrac 1 2 \mathrm{tr}\left(\hat A \hat x \hat A^\dagger\sigma_i \right) + \dfrac 1 2 \alpha \mathrm{tr}\left(\hat A \hat x \hat A^\dagger\sigma_i \right) = (Ax)_i + \alpha(Ay)_i \end{align}\]

We are interested in the Lorentz group, i.e. induced transformations \(A:\mathbb R^4\to \mathbb R^4\) which preserve the Minkowski metric. The relation between Minkowski norm and determinant of the Pauli identification in proposition 5.3 gives a convenient characterization.

Proposition 5.4 Conjugation by \(\hat A:\mathbb C^2\to \mathbb C^2\) induce a Lorentz transformation only if \[ \det \hat A = \pm 1 \] Moreover, \(\hat A\) induces the same transformation as \(-\hat A\), so without of loss of generality we can take \(\det \hat A = 1\).

Proof: Use proposition 5.3: \(\|Ax\|^2_M = \det(\hat A\hat x\hat A^\dagger) = \det \hat x (\det A)^2 = (\det A)^2 \|x\|^2_M\)

Within \(\mathrm{SL}(2, \mathbb C)\), the special unitary group \(\mathrm{SU}(2)\) consisting of \(2\times 2\) unitaries and the unit-determinant Hermitians \(\mathcal H_2\cap \mathrm{SL}(2, \mathbb C)\) are special: every \(A\in \mathrm{SL}(2, \mathbb C)\) may be decomposed into a unique product of \(U\in \mathrm{SU}(2)\) and \(H\in \mathcal H_2\cap \mathrm{SL}(2, \mathbb C)\).

Boosts and rotations

We explore \(\mathbb R^4\) transforms induced by the unitaries and unit-determinant Hermitians, beginning with the unitaries. Recall that every unitary \(U=\exp(iH)\) for some Hermitian \(H\). Since the Paulis form a basis, every unitary \(U\) may be written uniquely as \[ U=\exp\left[i(x_t \mathbf 1 + x_1\sigma_x + x_2\sigma_y + x_3\sigma_z)\right] = \exp\left[ix_t + i(x_1\sigma_x + x_2\sigma_y + x_3\sigma_z)\right] \] Inspecting equation (5.1) shows that the induced transformation is invariant in \(x_t\), and, in fact, an overall scaling of \((x_1, x_2, x_3)\). Without loss of generality we consider unitaries of the form \(U=\exp(i(\hat r\cdot \sigma))\), for \(r\in S^2\), the unit \(2\)-sphere in \(\mathbb R^3\).

Theorem 5.2 (unitaries induce spatial rotations) A clockwise rotation \(A\) in the last three spatial coordinates about \(\hat r\in S^2\) by \(\theta\) is induced via equation (5.1) by \[ U(\hat r, \theta) = \exp\left[-\dfrac i 2 \theta (\hat r\cdot \vec \sigma)\right] \in \mathrm{SU}(2) \]

Proof: All conjugation fixes the time axis since \(\mathbf 1\) commutes with everything so \(U(x_t\mathbf 1)U^\dagger= x_t\mathbf 1\). The only Minkowski-norm preserving isometries of \(\mathbb R^4\) which fix time are the spatial rotations. The direct form follows by direct computation or noting the identification with quaterions. We do one example here, let \[ \hat r = (0, 0, 1), \quad \hat R_z(\theta) = \exp\left(-\dfrac i 2 \theta\sigma_z\right) \implies R_z(\theta)_{ij} = \dfrac 1 2 \mathrm{tr}\left(\hat R_z(\theta)\sigma_j \hat R_z(-\theta)\sigma_i \right) \] If \(j=3\) (corresponding to spatial \(z\)) or \(j=0\) (time), \(\sigma_j\) commutes with \(\hat R_z(\theta)\) so \[ R_z(\theta)_{ij\notin\{0, 3\}}=\delta_{ij} \] This implies that \(R_z(\theta)\) acts trivially on the time and \(z\)-axes. Next, let us consider the subspaces spanned by \(\sigma_1, \sigma_2\). Recall that \([\sigma_3, \sigma_1] = -[\sigma_1, \sigma_3]\) \[\begin{align} R_z(\theta)_{11} &= \dfrac 1 2 \mathrm{tr}\left(e^{-i\theta \sigma_3/2}\sigma_1 e^{i\theta \sigma_3/2}\sigma_1 \right) \\ &= \dfrac 1 2 \mathrm{tr}\left(e^{-i\theta \sigma_3}\sigma_1\sigma_1 \right) = \dfrac 1 2 \mathrm{tr}\left(e^{-i\theta \sigma_3}\right) \\ &= \dfrac{e^{i\theta} + e^{-i\theta}} 2 = \cos\theta \end{align}\] The same applies to \(R_z(\theta)_{22}\) Continuing the calculation and remembering the commutation relation \(\sigma_i\sigma_j = i\epsilon_{ijk}\sigma_k\) \[\begin{align} R_z(\theta)_{12} &= \dfrac 1 2 \mathrm{tr}\left(e^{-i\theta \sigma_3/2}\sigma_1 e^{i\theta \sigma_3/2}\sigma_2 \right) \\ &= \dfrac 1 2 \mathrm{tr}\left(e^{-i\theta \sigma_3}\sigma_1\sigma_2 \right) = \dfrac 1 2 \mathrm{tr}\left(e^{-i\theta \sigma_3}i\sigma_3\right) \\ &= \dfrac{ie^{i\theta} - ie^{-i\theta}} 2 = \dfrac{e^{-i\theta}-e^{i\theta}}{2i} = -\sin\theta \end{align}\] Then \(R_z(\theta)\) restricted to the \(xy\) plane is a spatial rotation: \[ R_z(\theta)\big|_{xy} = \begin{bmatrix} \cos\theta & -\sin\theta \\ \sin\theta & \cos\theta \end{bmatrix} \]

We spend some time here to recall the hyperbolic functions. \[\begin{align} \cosh x &= \sinh'x = \dfrac{e^x + e^{-x}} 2 \\ \sinh x &= \cosh'x = \dfrac{e^x - e^{-x}} 2 \\ \sinh(x+y) &= \sinh x \cosh y + \cosh x \sinh y \\ \cosh(x+y) &= \cosh x \cosh y + \sinh x \sinh y \\ 1 &= \cosh^2 x - \sinh^2 x \end{align}\]

Theorem 5.3 (special Hermitians induce Lorentz boosts) Given \(\hat r\in S^2, \chi\in \mathbb R\), the following Hermitian induces a Lorentz boost in the direction of \(\hat r\in S^2\) by \(\chi\). \[ H(\hat r, \chi) = \exp \left[-\dfrac 1 2 \chi (\hat r\cdot \sigma)\right]\in \mathcal H_2 \]

Proof: explicit calculation. Given \(H(\hat r, \chi)\), compute the matrix elements of the linear transform of \(\mathbb R^4\) it corresponds to using equation (5.1); this turns out to be the Lorentz boost.