Let $X$ be a set.
Let $\mathscr{A}$ and $\mathscr{B}$ be $C^k$-atlases on $X$. Let $\tau_1,\tau_2$ be the canonical topologies on $X$ determined by $\mathscr{A}$ and $\mathscr{B}$ respectively.
I'm curious on the relation between [equivalence of $C^k$-atlases] $\leftrightarrow$ [canonical topologies determined by $C^k$-atlases]
It is true that if $\mathscr{A}$ and $\mathscr{B}$ are equivalent, then $\tau_1=\tau_2$.
What about the converse?
Since topology does not say anything about differentiability, the converse would be false. But what is a countetexample for this?