At the end of Chapter 2 of his Course in Mathematical Physics, Szekeres discusses the notion of a symmetry group. I present my definition, adapted from his, here:
We say that a transformation $g: X \to X$ (i.e. $g \in$ Transf($X$)) leaves the $Y_i$-valued function $F_i$ invariant if, for all $x \in X$, we have $F_i(x) = F_i(gx)$.We have used the shorthand $gx \equiv g(x)$ since $g \in$ Transf($X$) so that it’s (left) action on $G$ is trivially defined by $g \mapsto g$.
We say that a transformation leaves a space $X$ equipped with a set of functions $\mathcal{F}$ invariant if it leaves each $F_i \in \mathcal{F}$ invariant.
The set of all transformations $g$ which leave $\mathcal{F}$ on $X$ invariant is called the symmetry group of that space. I will use the notation Sym$(X,\mathcal{F})$ for this group.
In general, the "functions on the space" may actually be functions of, e.g., Cartesian products of the underlying set (or subsets of the powerset).
Later, Szekeres defines Galilean space:
Define an event to be a point of $\mathbb{R}^4$ characterized by four coordinates $(x, y, z, t)$. Define Galilean space $\mathbb{G}^4$ to be the space of events with a structure consisting of three functions defined thereon:
- Time intervals $\Delta t := t_2 − t_1$.
- The spatial distance $\Delta s = |\mathbf{r}_2 − \mathbf{r}_1|$ between any pair of simultaneous events (events having the same time coordinate, $t_1 = t_2$).
- Motions of inertial (free) particles, otherwise known as rectilinear motions, $\mathbf{r}(t) = \mathbf{u}t + \mathbf{r}_0$, (2.19) where $\mathbf{u}$ and $\mathbf{r}_0$ are arbitrary constant vectors.
Finally, he adds the following discussion:
Note that only the distance between simultaneous events is relevant. A simple example should make this clear. Consider a train travelling with uniform velocity v between two stations A and B. In the frame of an observer who stays at A the distance between the (non-simultaneous) events E1 = ‘train leaving A’ and E2 = ‘train arriving at B’ is clearly $d = vt$, where t is the time of the journey. However, in the rest frame of the train it hasn’t moved at all and the distance between these two events is zero! Assuming no accelerations at the start and end of the journey, both frames are equally valid Galilean frames of reference.
My two questions are as follows:
(1) I have not studied special relativity or mathematical classical mechanics, but I suppose I am to understand that the Galilean group (symmetries of the space described above) is isomorphic to (or defined as?) the set of valid inertial reference frames?
(2) How am I to understand the reference frame example in the context of the above? I believe Szekeres is pointing out that if we allowed distances to be defined between non-simultaneous events, then we would find that different reference frames would "compute" different distances between events. This would, in turn, contradict the axiom of classical mechanics/Galilean relativity that different inertial frames of reference determine the same distances between points. Is this understanding correct? I have bolded the step which I cannot justify; is this an axiom as stated, does it derive from something I've written above, or is it derivable from something I'm missing?
Edit: I'm adding a "bonus" question (bonus since I don't want to layer on an unrelated question, but it's somewhat related -- feel free to indicate if I should just ask this as a separate question). Szekeres goes on to define the Lorentz group by saying that the space of events has on it a function which encodes the "preservation of the light cone":
The Galilean transformations do not preserve the light cone at the origin $$\Delta x^2 + \Delta y^2 + \Delta z^2 = c^2\Delta t^2.$$
Szekeres then goes on to derive a requirement on the form of the symmetry transformations, deriving the structure of the Poincare group thereform. What I am confused about is whether Minkowski space as so defined is to be understood as the same space as Galilean space above (i.e. the 3 functions defined above are still part of the set of functions $\mathcal{F}$ on the space, and we also add that the function $$\Delta s^2 := \Delta x^2 + \Delta y^2 + \Delta z^2 - c^2\Delta t^2$$ must be invariant under the given symmetry transformation. Or are we removing one of the three functions above?