The following is Theorem 2.1 in Ref. 1.
Theorem 2.1. Let (x1, ⋯, xn + 1) be the standard coordinate system on $\mathbb {R}^{n+1}_1$
and (y1, ⋯, yn+1) be another coordinate system on $\mathbb {R}^{n+1}_1$
with n ⩾ 2. Assume that, for any smooth function φ, $\sum \limits _{i=1}^n \frac{\partial ^2 \varphi }{\partial x_i^2} - \frac{\partial ^2 \varphi }{\partial x_{n+1}^2} = 0$
if and only if $\sum \limits _{i=1}^n \frac{\partial ^2 \varphi }{\partial y_i^2} - \frac{\partial ^2 \varphi }{\partial y_{n+1}^2} = 0$
. Then, all rows of Jacobian matrix $\big ( \frac{\partial y_i}{\partial x_j} \big )$
are mutually orthogonal and have the same length. Furthermore, the first n rows are spacelike vectors and the last row is a timelike vector.
Let us denote the operator $\sum \limits _{i=1}^n \frac{\partial ^2 }{\partial x_i^2} - \frac{\partial ^2}{\partial x_{n+1}^2}$
by □ and $\sum \limits _{i=1}^n \frac{\partial ^2 }{\partial y_i^2} - \frac{\partial ^2 }{\partial y_{n+1}^2}$
by □′. Then Theorem 2.1 states that the condition, □φ = 0 ⇔ □′φ = 0 for any C2 function φ, implies that the Jacobian of F is Lorentz matrix multiplied by a positive function, where $F : \mathbb {R}^{n+1}_1 \rightarrow \mathbb {R}^{n+1}_1$
given by (y1, ⋯, yn+1) = F(x1, ⋯, xn+1) is a diffeomorphism. Essentially, this implies that the diffeomorphism F is a conformal transformation if □φ = 0 ⇔ □′φ = 0 for any C2 function φ.
Then, we can use the following known result, called Liouville's theorem.
Theorem 2.2. Every C4 conformal transformation of a region of pseudo-Euclidean space of dimension ⩾3 is a composite of isometries, dilations, and inversions.
Proof. This is Theorem 15.2 in Ref. 2.
It is a well-known fact that any causal automorphisms on $\mathbb {R}^{n+1}_1$
with n ⩾ 2 is C∞ and so we can apply the above theorem to Theorem 2.1. Also, since inversion has a singularity, it cannot be defined on the whole of $\mathbb {R}^{n+1}_1$
and thus any conformal transformation is a composite of isometries and dilations. Therefore, we have the following, which is Theorem 2.3 in Ref. 1.
Theorem 2.3. Let $F : \mathbb {R}^{n+1}_1 \rightarrow \mathbb {R}^{n+1}_1$
given by (y1, ⋯, yn+1) = F(x1, ⋯, xn+1) be a diffeomorphism with n ⩾ 2. Then the necessary and sufficient condition for F to be a causal automorphism on $\mathbb {R}^{n+1}_1$
is that $\frac{\partial y_{n+1}}{\partial x_{n+1}} \ge 0$
, and for any smooth function φ, $\sum \limits _{i=1}^n \frac{\partial ^2 \varphi }{\partial x_i^2} - \frac{\partial ^2 \varphi }{\partial x_{n+1}^2} = 0$
if and only if $\sum \limits _{i=1}^n \frac{\partial ^2 \varphi }{\partial y_i^2} - \frac{\partial ^2 \varphi }{\partial y_{n+1}^2} = 0$
.
Recently, it is shown that causal automorphisms on $\mathbb {R}^2_1$
can also be characterized by invariance of wave equations. Therefore, if Liouvilles's theorem holds for two-dimensional space, causal automorphism on $\mathbb {R}^2_1$
must have the same form as in Theorem 2.3, which is not true. In other words, the characteristic difference between $\mathbb {R}^2_1$
and $\mathbb {R}^n_1$
with n ⩾ 3 comes from the truth of Liouville's theorem.