Is it possible to define differentiable maps between topological spaces without using the idea of manifolds? I mean with using just the topological structure (open sets or neighborhoods).
-
$\begingroup$ You must be more precise in what you are willing to do. The answer to your question is depending on. $\endgroup$– Patrick I-ZCommented Dec 14, 2013 at 14:42
-
3$\begingroup$ You can define anything you like, I suppose. It's especially easy if the object you want to define has no role to play. $\endgroup$– Ryan BudneyCommented Dec 14, 2013 at 14:45
-
$\begingroup$ I just mean the definition. Is there any definition of differentiability of maps between topological space. $\endgroup$– Abdullah AlmariahCommented Dec 14, 2013 at 14:47
-
$\begingroup$ What properties would you like the definition to satisfy ? $\endgroup$– Ryan BudneyCommented Dec 14, 2013 at 14:58
-
1$\begingroup$ After reading the answer by Todd Trimble, I voted for reopening. But the question should be improved. @Almariah: could you provide some context? From where did you get the idea that a surrogate of the notion of differentiable maps could be available for (some classes of?) topological spaces? .... $\endgroup$– QfwfqCommented Dec 14, 2013 at 17:18
1 Answer
While this is not for general topological spaces, you might be interested in a kind of generalization of differential calculus for metric spaces, developed by Burroni and Penon here.
A key notion is that of two functions $f, g: M \to N$ between metric spaces to be tangent to each other at a point $a \in M$: this means $f(a) = g(a)$ and the function $C: M \to \mathbb{R}_+$ defined by
$$C(x) = \frac{d(f(x), g(x))}{d(x, a)} \;\; (x \neq a), \qquad C(a) = 0$$
is continuous at $a$. This is an equivalence relation. Then, given pointed metric spaces $(M, a)$ and $(N, b)$, one defines a jet $(M, a) \to (N, b)$ to be a tangency equivalence class of locally Lipschitz maps $f: M \to N$ such that $f(a) = b$ ("locally Lipschitz" means there is some $k > 0$ such that $f$ is $k$-Lipschitz when restricted to some neighborhood of $a$). If $f: (M, a) \to (N, b)$ is a based function that is tangent to some locally Lipschitz $g: (M, a) \to (N, b)$ at $a$, then they define the tangential (roughly akin to a differential) $T f_a$ to be the corresponding jet, and in that case they say $f$ is "tangentiable" at $a$.
Ultimately Burroni and Penon want to define, given a map $f: M \to N$ between suitable metric spaces, a notion of tangential $tf: M \to {"Jet(M, N)"}$ that plays a role analogous to the differential $df: U \to L(E, E')$ of a differentiable map $f: U \to E'$, where $E, E'$ are normed vector spaces and $U$ is an open set of $E$. They achieve this by considering certain metric spaces equipped with some extra structure, called transmetric spaces. But for this I'll have to refer you to the article.