Partial derivative being $0$ implies non-dependence on that coordinate for $f: U \rightarrow \mathbb{R}$ and $U \in \mathbb{R}^n$ open and convex

203 Views Asked by At

Let $U \subset \mathbb{R}^n$ be open and convex and $f : U \rightarrow \mathbb{R}$ differentiable such that $\partial_1f(x)=0$ for all $x \in U.$ Show that the value of $f(x)$ for $x = (x_1,...,x_n) \in U$ does not depend on $x_1.$

$\text{Proof:}$

Suppose for a contradiction that $f$ depends on the first coordinate. Let $x,y \in U$ be such that $x_i=y_i$ except for $i=1.$ Since $U$ is open and convex we can use the Mean Value Theorem. Hence there is $c= t_0x+(1-t_0)y,$ for some $t_0 \in [0,1]$ such that \begin{align*} f(x)-f(y) &=\nabla f(c)\cdot\langle x-y \rangle\\ &=\langle 0,\partial_2f(c),...,\partial_nf(c) \rangle \cdot \langle x_1-y_1,0,...,0 \rangle \\ &=0. \end{align*}

Therefore, we have that $f(x)=f(y),$ which is a contraction, since $f$ is not constant. Thus, $f$ does not depend on the first coordinate.

Is this the correct approach to this problem? Any feedback is much appreciated. Thank you in advance.

1

There are 1 best solutions below

0
On BEST ANSWER

Your proof is fine, but it is needless to phrase it as a proof by contradiction. The same computation that you did shows that $g(t) = f(t,x_2,\dots,x_n)$ is a constant function, which is all you want. As a matter of style, direct proofs are preferred to proofs by contradiction.