In special relativity, when working with coordinates $(x^0, x^1, x^2, x^3)$ , the matrix corresponding to a Lorentz boost in the $x^2$ direction is $$ \begin{pmatrix} \gamma & -\gamma \beta & 0 & 0 \\ 0 & 1 & 0 & 0 \\ - \gamma \beta & 0 & \gamma & 0 \\ 0 & 0 & 0 & 1 \end{pmatrix}. $$ Similarly, the one corresponding to the $x^3$ direction is $$ \begin{pmatrix} \gamma & -\gamma \beta & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 1 & 0 \\ - \gamma \beta & 0 & 0 & \gamma \end{pmatrix}. $$ I now want to compute the matrix corresponding to a boost in the $x^2 + x^3$ direction. I looked around on the internet and it seems that the product of two non-parellel boosts is not a boost. So my idea was this: I first boost in the $x^2$ direction (the $y$-axis), then I rotate my coordinate axes so that the $x^3$ axis (i.e. $z$-axis) gets mapped to the $x^2$-axis by rotating clockwise about the $x^1$-axis, that is, by applying the matrix $$ R = \begin{pmatrix} 1 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & \cos \theta & - \sin \theta \\ 0 & 0 & \sin \theta & \cos \theta \end{pmatrix} $$ with $\theta = \pi/2$. Then I apply a boost again in the $x^2$ direction. Finally I rotate back by applying $R^{-1}$. So that would give me $$ R^{-1} \Lambda R \Lambda. $$

Can someone tell me if this is the correct reasoning, and that this is the way one finds out a general boost?

  • 2
    $\begingroup$ I say this because I don't believe rotation and boost transformations commute. $\endgroup$ – anon01 Oct 7 '16 at 20:22
  • $\begingroup$ I don't understand what you mean. What is $x'$? If I rotate first, and then boost, I still have to rotate back? Isn't that the same what I did? $\endgroup$ – Kamil Oct 7 '16 at 20:24
  • $\begingroup$ However, wikipedia also has an expression for a lorentz boost in an arbitrary direction... $\endgroup$ – anon01 Oct 7 '16 at 20:29
  • $\begingroup$ @ConfusinglyCuriousTheThird indeed, the commutator of a boost with a rotation is another boost ($\left[J_{m},K_{n}\right] = i \varepsilon_{mnl} K_{l}$). $\endgroup$ – gradStudent Oct 7 '16 at 21:20

The Wikipedia General Lorentz Boosts article, from which the equations below are taken, stresses at least 3 times that, due to the non-commutation of boosts and rotations, that calculating what I think you are doing, is a non-trivial exercise, possibly involving computations as detailed below.

I don't really consider this an answer, I just wanted to draw your attention to potential problems in calculation. My apologies if it's familiar to you already.

The general boost matrix is

$${\displaystyle B({\boldsymbol {\zeta }})=e^{-{\boldsymbol {\zeta }}\cdot \mathbf {K} }}$$

where $K = (K_x, K_y, K_z)$ is a vector of matrices with components

$${\displaystyle K_{x}={\begin{bmatrix}0&1&0&0\\1&0&0&0\\0&0&0&0\\0&0&0&0\\\end{bmatrix}}\,,\quad K_{y}={\begin{bmatrix}0&0&1&0\\0&0&0&0\\1&0&0&0\\0&0&0&0\end{bmatrix}}\,,\quad K_{z}={\begin{bmatrix}0&0&0&1\\0&0&0&0\\0&0&0&0\\1&0&0&0\end{bmatrix}}}$$

and expanding the exponential in a Taylor series reveals a hyperbolic analogue of Rodrigues' rotation formula,

$${\displaystyle B({\boldsymbol {\zeta }})=I-\sinh \zeta (\mathbf {n} \cdot \mathbf {K} )+(\cosh \zeta -1)(\mathbf {n} \cdot \mathbf {K} )^{2}}$$

which reproduces the boost matrix after converting the rapidity back to relative velocity. Similarly the general rotation matrix is

$${\displaystyle R({\boldsymbol {\theta }})=e^{{\boldsymbol {\theta }}\cdot \mathbf {J} }}$$

where $J = (J_x, J_y, J_z)$ and

$${\displaystyle J_{x}={\begin{bmatrix}0&0&0&0\\0&0&0&0\\0&0&0&-1\\0&0&1&0\\\end{bmatrix}}\,,\quad J_{y}={\begin{bmatrix}0&0&0&0\\0&0&0&1\\0&0&0&0\\0&-1&0&0\end{bmatrix}}\,,\quad J_{z}={\begin{bmatrix}0&0&0&0\\0&0&-1&0\\0&1&0&0\\0&0&0&0\end{bmatrix}}}$$

and Taylor expanding the exponential reveals the actual Rodrigues' rotation formula

$${\displaystyle R({\boldsymbol {\theta }})=I+\sin \theta (\mathbf {e} \cdot \mathbf {J} )+(1-\cos \theta )(\mathbf {e} \cdot \mathbf {J} )^{2}\,.}$$

The axis-angle vector θ and rapidity vector ζ are altogether six continuous variables which make up the group parameters (in this particular representation), and J and K are the corresponding six generators of the group. Physically, the generators of the Lorentz group correspond to important symmetries in spacetime: J are the rotation generators which correspond to angular momentum, and K are the boost generators which correspond to the motion of the system in spacetime.

The derivative of any smooth curve $A(t)$ with $A(0) = I$ in the group depending on some group parameter t with respect to that group parameter, evaluated at t = 0, serves as a definition of a corresponding group generator $X$, and this reflects an infinitesimal transformation away from the identity. The smooth curve can always be taken as an exponential as the exponential will always map $X$ smoothly back into the group via t → exp(tX) for all t; this curve will yield $X$ again when differentiated at t = 0.

Three of the commutation relations of the Lorentz generators are

$${\displaystyle {\begin{aligned}\lbrack J_{x},J_{y}\rbrack &=J_{z}\\\lbrack K_{x},K_{y}\rbrack &=-J_{z}\\\lbrack J_{x},K_{y}\rbrack &=K_{z}\end{aligned}}}$$

where the bracket $[A, B] = AB − BA$ is a binary operation known as the commutator, and the other relations can be found by taking cyclic permutations of $x, y, z$ components (i.e. change x to y, y to z, and z to x, repeat).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.