Good morning. I have been thinking about the following question for a while without much success, therefore I'm starting to doubt its validity, although I don't have a clear counterexample in mind.
Let $X$ be a Banach space and $F=(F_1,\dotso, F_m):X\to \mathbb R^m$ be a smooth map such that $F(0)=0$. Assume that $F$ is open at zero, i.e. for every $\epsilon>0$ there exists $\delta=\delta(\epsilon)$ such that $B(0,\delta)\subset F(B_X(0,\epsilon))$.
We consider vectors $\lambda\in S^{m-1}$ and the compositions $\Phi_\lambda:X\to \mathbb R$ defined by $\Phi_\lambda(x):= \langle\lambda, F(x)\rangle=\sum_{i=1}^m \lambda_i F_i(x)$.
If $F$ is open at $0$, $\Phi_\lambda$ is open at zero for every $\lambda\in S^{m-1}$.
I'm now interested in the converse statement, specifically I want to pose these two questions:
- Is it true that, under the assumption that $\Phi_\lambda:X\to \mathbb R$ is open at zero for every $\lambda\in S^{m-1}$, then also $F$ is open at zero?
- In case point 1. is false, can we strengthen somewhat the hypotheses so that it becomes true, or is it unavoidably false?
Thanks in advance,
Gil
