In four dimensional space time an arbitrary element $\Lambda\in\text{SO}(3,1)^{\uparrow}$ of the Lorentz group may be decomposed in the following way: $$\Lambda=\tilde{R}\Lambda_x(\psi)R$$ where $\tilde{R},R$ are spatial rotations and $\Lambda_x(\psi)$ is a Lorentz boost in the x-t plane.

Is the analogous decomposition in three dimensional space-time $$\Lambda=\Lambda_x(\psi)R\text{ ?}$$ I dont think this can be true since this would imply that any 3D Lorentz transformation could be described by 2 continuous parameters but the Lie group $\text{SO}(2,1)^{\uparrow}$ has dimension 3. On the other hand, I have almost proved that the above identity holds from the polar decomposition of real 2x2 matrices $N=OP$ where $N\in\text{ SL}(2,\mathbb{R})$ is the double cover of $\text{SO}(2,1)^{\uparrow}$, $O$ is an orthogonal 2x2 matrix and $P$ is a positive definite symmetric 2x2 matrix.

The piece I am missing in this proof is that a Lorentz transformation is symmetric iff it is a pure Lorentz boost. I have read that this is true, but I am unsure of how to prove it.


After reflecting on ZeroTheHero's answer, I realise that what I was proving using the polar decomposition theorem was the more general decomposition $$\Lambda=\text{R}(\alpha)\Lambda_{\hat{n}}(\psi)$$ given in his answer. So there is no contradiction here, this has three continuous parameters just as its Lie group does.


1 Answer 1


The factorization you want is $$ \Lambda=R_z(\alpha)\cdot \Lambda_x(\psi)\cdot R_z(\gamma) $$ where $R_z(\alpha)$ is a rotation in the $xy$ plane by an angle $\alpha$, and $\Lambda_x$is a boost along $x$.

At the group level, the elements are products of rotations $R_z$ in the $xy$ plane and boosts $\Lambda_{\hat n}$ in the direction $\hat n$ in the $xy$ plane, so general element is trivially a product of the form $R_z(\tilde\alpha)\Lambda_{\hat n}(\psi)$.

If $\gamma$ is the rotation that takes $\hat x$ to $\hat n$, then $$ \Lambda_{\hat n}(\psi)=R_{z}^{-1}(\gamma)\Lambda_x(\psi)R_z(\gamma) $$ so that a general element is now factorized as $$ R_z(\tilde\alpha)R_z^{-1}(\gamma)\Lambda_x(\psi)R_z(\gamma)\, . $$ The first two factors are $\in SO(2)$ so can be combined into a single $R_z$ rotation to finally get $$ R_z(\alpha)\Lambda_x(\psi)R_z(\gamma)\, . $$

  • $\begingroup$ Thank you. I have often seen physical arguments which justify such a decomposition in place of a mathematical proof. How would one go about rigorously proving this decomposition? $\endgroup$ Jul 2, 2017 at 9:51
  • $\begingroup$ @NormalsNotFar Hopefully the bit I added helps. $\endgroup$ Jul 2, 2017 at 10:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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