Here * means the cartesian product, and P(X) is the set of all subsets of $X$. $E^F$ is the set of all applictions from $F$ to $E$. The question is contained in the title : i wish to find a natural bijection between those sets. Here is my try :
Let $f:F \rightarrow P(E)$ be a random function of $P(E)^F$. We call $F_0=\{x\in F, f(x)\neq \emptyset\}$. Then using the axiom of choice, we can define a function that associates, to every $x\in F$, some $y_x\in f(x)$ and get a pair $(x,y_x)$.
Let's call $X_f=\bigcup_{x\in F_0}\{(x,y_x)\}$ for every function $f$. It is clearly a subset of $E*F$.
My idea would be to consider after that the application $g:P(E)^F\rightarrow P(E*F)$ such that : $\forall f\in P(E)^F$, $g(f)=X_f$.
Now i get stucked because i cannot prove that $g$ is bijective. To do so, i would have to make further suppositions about the choices of the $y_x$ that were made for every function $f$, in order to get both surjectivity and injectivity. My idea to get surjectivity was to consider the size of $F_0$ (in order to control the size of $X_f$ by controlling the elements of $F$ that i could put into a pair that is in $X_f$). However, i feel like this is not the right method. Could anyone give me a hint about this ? I would be delighted if you could ! :)