1. I suspect that you started from the Minkowski metric $$ds^2=(cdt)^2-dx^2$$ and you passed it through the Shubert transforms, right?
No, or at least not the full transformation. Eugene claims his transformation factorises as $$C \,=\, \theta \,\circ\, \Lambda \,\circ\, \theta^{-1}$$, where $$\Lambda$$ is a Lorentz transformation, and I took $$\theta$$ to mean
$$
\theta(t,\,x^{i}) \,=\, \bigl( t \,+\, S(x^{i}),\, x^{i} \bigr) \;.
$$
(the definition of $$\theta$$ as it appears in Eugene's essay has some extra indices, including an index on the
t variable. I have no idea what they're supposed to add). I only applied $$\theta$$ to the Minkowski metric. With that in mind, you might want to re-read this part of my earlier post, which should explain why I found this more interesting than applying the entire transformation:
In terms of the decomposition $$\theta \,\circ\, \Lambda \,\circ\, \theta^{-1}$$ of his transformation, $$\theta^{-1}$$ brings you back to the Minkowski metric, the Lorentz transformation $$\Lambda$$ leaves the Minkowski metric invariant, and $$\theta$$ takes you back to the metric above.
so Eugene's transformation $$C$$ leaves invariant the metric expression I posted earlier. Obviously if I'd just applied $$C$$ directly on the Minkowski metric it wouldn't leave its components invariant. It would take the Minkowski metric to some other metric.
A quick examination shows that you missed the fact that $$S_i=S_i(x,t,S_j(x,t)$$.
As Guest pointed out earlier, if you think there's a problem with "differentiating
S with respect to itself" - which I take to mean you think differentiating Eugene's transformation produces terms like $$\frac{\partial S}{\partial S}$$ - then you've got a misconception about the chain rule. Specifically, it sounds like you're not clear on the distinction between differentiating a function with respect to its parameters, and
evaluating the function or differential at a point.
Differentiating something like $$S \bigl( S (x) \bigr)$$ produces a term like
$$\frac{\partial S}{\partial x} \bigl( S(x) \bigr) \;.$$
In words: the partial derivative of
S with respect to its parameter (here called
x, not to be confused with the spatial coordinate also called
x) and
evaluated at
S(
x).
Firstly, you cannot prove that the jacobian is non-degenerate.
The Jacobian matrix of $$\theta$$ is just
$$
J_{\theta} \,=\,
\begin{bmatrix}
1 & \partial_{1} S & \partial_{2} S & \partial_{3} S \\
0 & 1 & 0 & 0 \\ 0 & 0 & 1 & 0 \\ 0 & 0 & 0 & 1
\end{bmatrix} \;.
$$
The determinant of this matrix is just 1. The Jacobian determinant of the full non-linear transformation is
$$\det \bigl[ J_{C} \bigr]
\,=\, \det \bigl[ J_{\theta} \,\circ\, \Lambda \,\circ\, J_{\theta^{-1}} \bigr]
\,=\, \det \bigl[ J_{\theta} \bigr] \, \det \bigl[ \Lambda \bigr] \,
\det \bigl[ J_{\theta^{-1}} \bigr] \,=\, 1 \;,$$
so clearly the Jacobian matrix of the transformation is non-degenerate. This shouldn't be surprising, since non-degeneracy of the Jacobian matrix everywhere is equivalent to invertibility of the transformation, and Eugene explicitly gave the inverse of $$\theta$$ in his essay.
Secondly, in order for the covariance of the metric to be preserved under transformation, the jacobian needs to be a unit matrix[1] , a condition that is clearly not satisfied.
I'm not familiar with the term "unit matrix". If you mean a matrix whose determinant is of absolute value 1, then this is true and Eugene's transformation satisfies this requirement, as shown above.
4. Not if $$\theta^{-1}$$ does not exist since the jacobian cannot be proven non-degenerate.
As stated above, Eugene explicitly gave $$\theta^{-1}$$ in his essay. It's easily seen to be
$$
\theta^{-1}(t,\, x^{i}) \,=\, \bigl( t \,-\, S(x^{i}),\, x^{i} \bigr) \;.
$$
6. Here you made another error. The spacetime being flat, the Riemann tensor is null so the Christoffel symbols are all null. [2]. The covariant derivatives reduce to standard derivatives.
I've already told you this isn't correct. The Christoffel symbols are not a measure of curvature. Intuitively, they measure the pseudo-forces present in non-inertial coordinate systems, eg. centrifugal force or the gravitational "force" present in accelerating frames. It's easy to prove that the Christoffel symbols are non-zero in any coordinate system in which the metric components are inhomogenous. This is the case for the metric I gave in my earlier post, unless the function
S, and therefore Eugene's entire transformation, happens to be linear or affine.