So I've been reading Mathematical Foundations of Supersymmetry by Carmeli, Caston and Fioresi, and I cannot wrap my head around the proof of lemma 4.2.3 from that book. The first part of the lemma (which is the one I have a problem with) goes as follows:
Let $M$ and $N$ be supermanifolds, and let $\phi := (|\phi|, \phi^*): M \to N$ be a super ringed space morphism. Then for any $U \subset N$ open and $f \in \mathcal{O}_N(U)$, we have $\phi^*(f)(x) = f(|\phi|(x))$.
The second part of this lemma is a simple corollary that says that any super ringed space morphism between supermanifolds is automatically a supermanifold morphism (since the lemma is equivalent to $\phi$ being local).
The authors give a proof as follows:
Suppose that $\phi^*(f)(x) \neq f(|\phi|(x))$. Define $g = f - f(|\phi|(x)) \in \mathcal{O}_N(U)$. Then $\phi^*(g)(x) \neq 0$ by assumption. Furthermore, by construction $g(|\phi|(x)) = 0$, therefore g is not invertible on any neighbourhood of $|\phi|(x)$. However, since $\phi^*$ is a superalgebra morphism and $\varphi^*(g)(x) \neq 0$, we have a contradiction.
I've thought about it for a while now, but I really cannot see where the contradiction comes from. It seems to me that the authors are trying to say that if $g$ is not invertible on any neighbourhood of $|\phi|(x)$, then its image $\phi^*(g)$ cannot be invertible either, but non units can map into units, right?