The definition of a Lie group is that its multiplication and also the map that sends $g$ to $g^{-1}$ are differentiable maps. It is known that the inversion condition is redundant, using the Implicit Function Theorem.
Q1) I would like to ask if there are any books / notes describing it in the simplest possible way. (Reference request) If anyone can put their proof here it will be good too.
I have read this, and unfortunately I do not really understand it, though admittedly it looks short and neat. Is there a simpler way to prove that inversion is redundant, or is that the simplest proof already?
This is a question in Brocker and tom Dieck, which unfortunately I believe cannot be solved using the material covered in the book alone.
Thanks!
Note: I have read Redundance of the Smoothness of the Inversion Map in the Definition of a Lie Group., to no avail as well.
Q2) Also, what version of Implicit Function Theorem is being used in the proof? After some research, I am pretty sure it is not this version in Wikipedia (https://en.wikipedia.org/wiki/Implicit_function_theorem), but not sure which is the most appropriate version to use.
Let $$F:G\times G\to G\times G\quad F(g,h)=(g,gh).$$ Then, $F$ is a smooth map since multiplication is assumed to be smooth.
Proof: This is an easy exercise.
The map $L_g:G\to G$ has a smooth inverse $L_{g^{-1}}$, so it is a diffeomorphism. Thus, $(dL_g)_h$ is an isomorphism and hence $(1)$ is surjective. Since the domain and range have the same dimension, $(1)$ is an isomorphism. This shows that $F$ is a local diffeomorphism. But $F$ is also bijective, so we have shown:
In particular, its inverse $$F^{-1}:G\times G\to G\times G,\quad F^{-1}(g,h)=(g,g^{-1}h)$$ is smooth, and hence the following composition is smooth: $$g\mapsto (g,e)\overset{F^{-1}}{\mapsto} (g,g^{-1})\mapsto g^{-1}.$$