Is there a gap in Serre's proof of inverse function theorem?

133 Views Asked by At

On page 73 of 'Lie algebras and Lie groups', Serre proves the inverse function theorem for complete fields. I would like to have some clarification about the following point.


Let $K$ be a complete ultrametric field. Let $f = (f_1,\dots,f_n)$ be an $n$-uple of formal power series in $n$ variables (i.e. $f_i \in K[[X_1, \dots,X_n]]$). Suppose that $f$ is convergent and that $D_0 f$ is invertible.

Serre proves that there exists $g=(g_1,\dots,g_n)$ a formal power series convergent satisfying : $f \circ g(\underline{T}) = \underline{T} = (T_1,\dots,T_n)$ as formal power series (and also satisfying $g \circ f(\underline{T}) = \underline{T} = (T_1,\dots,T_n)$ if I am not mistaken). Serre ends the proof here.

Why does this prove the theorem ?

To conclude, we have to prove that there exist two opens $U,V$ such that $f$ maps U into $V$ and $g$ maps $V$ into $U$. But I can't see why these opens exist. It seems to me that we have to prove $f$ is an open map.