Let $X$ be a normed vector space and $M$ a linear subspace of $X$. Using the Hahn-Banach Theorem, I managed to prove that
For each $g\in X^*$ there exists $h\in M^\perp$ such that $||g-h||=d(g,M^\perp).$
Here, $X^*$ denotes the dual space of $X$ and $M^\perp=\{f\in X^*:f(x)=0~\forall x\in M\}$. Now, I'm trying to prove:
If for each functional $f\in M^*$ there exists a unique functional $F\in X^*$ such that $||F||_{X^*}=||f||_{M^*}$, then for each $g\in X^*$ there exists a unique functional $h\in M^\perp$ such that $||g-h||=d(g,M^\perp)$.
What I've tried so far is letting $f=g|_M$, the restriction of $g$ to $M$. Clearly $f\in M^*$, then there exists a unique functional $||F||_{X^*}=||f||_{M^*}$. I defined the function $h=g-F$, and tried to prove that $h\in M^\perp$. But my hypothesis doesn't say $F$ extends $f$, just preserves the norm. I'm not sure how to use the uniqueness part, there might be somehow a way to use that in order to show that $F|_M=f=g|_M$, or maybe I'm in the wrong direction. Any help would be appreciated.
Edit: that cannot be unique since $||F||=||-F||$, if I add $F|_M=f$ maybe everything works. I'm not sure if there is a weaker hypthesis we can add than that one. Just in case, this is an introductory Functional Analysis problem.