Equivalence between two versions of Open Mapping Theorem

functional-analysisgeneral-topology

I have seen two versions of the open mapping theorem. I am trying to understand why they are equivalent.

From wikipedia:

If $X$ and $Y$ are Banach spaces and $A : X \rightarrow Y$ is a surjective continuous linear operator, then $A$ is an open map.

From Royden (paraphrased):

Let $X$ and $Y$ are Banach spaces and $T : X \rightarrow Y$ is a continuous linear operator. $T(X)$ is closed as a subspace of $Y$ iff $T$ is an open map.

How are these equivalent?

EDIT:

I've included the releveant portion in Royden. Indeed, he discusses the image as having inherited the subspace topology from $Y$ — I missed this before the discussion in the comments, thanks!

enter image description here

Best Answer

It is clear that the second result implies the first result.

If it is true the first result, we consider a map $T:X\to Y$ such that $T(X)$ is closed. Each closed subspace of a Banach Space $Y$ is also a Banach Space so $T(X)$ is a Banach Space and $T:X \to T(X)$ is a surjective continuos linear operator between Banach Spaces; so, by first result, $T:X\to T(X)$ will be an open map. In any case, it is not true that $T: X\to Y$ will be an open map because in general $T(X)$ it is not open in $Y$, infact if it is open, then $T(X)$ will be a nonempty open and closed subspace of the connected space $Y$, so $T(X)=Y$. An example can be $T:\mathbb{R}\to \mathbb{R}^2$ such that

$T(x):=(x,0)$ . The map is continuos and Linear while the two spaces are Banach, so $T:\mathbb{R}\to \mathbb{R}\times \{0\}$ is open while

$T:\mathbb{R}\to \mathbb{R}^2$ is not open because $T(\mathbb{R})= \mathbb{R}\times \{0\}$ that it is not open in $\mathbb{R}^2$.

Related Question