Can l'Hopital's Rule Be Generalized for Differentiable Maps Between Manifolds?

  • Thread starter Thread starter Mandelbroth
  • Start date Start date
  • Tags Tags
    L'hopital's rule
Click For Summary
The discussion explores the possibility of generalizing l'Hôpital's Rule for differentiable maps between manifolds, noting that the division of two differentiable functions is not generally defined. The author suggests that a potential approach could involve the limits of the norms of the functions and their gradients. They propose a formulation involving the limit of the ratio of the norms of two functions and their gradients at a point. The conversation emphasizes the need for a rigorous framework to validate these ideas, as traditional calculus concepts may not directly apply in this context. The exploration remains open-ended, inviting further insights into the generalization of this mathematical principle.
Mandelbroth
Messages
610
Reaction score
24
I had a wild thought.

Out of curiosity, is anyone aware of a kind of generalization for l'Hopital's Rule from analysis for differentiable maps between differentiable manifolds? I'm having trouble formulating if I could do it or not, because (as far as I know), if I have ##f,g:M\to N##, with ##f## and ##g## differentiable and ##M## and ##N## differentiable, ##f(x)/g(x)## is not, in general, defined.

Again, I don't know if it can be generalized. Ideas are certainly welcome, since I'll probably be stuck thinking about it until I prove something does work or doesn't work.
 
Physics news on Phys.org
Maybe there's something like:

\lim_{x\rightarrow 0} \frac{||f(x)||}{||g(x)||} = \frac{||\nabla f(0)||}{||\nabla g(0)||}

But it's just a guess, I didn't confirm this.
Anyway, you can't define division between two functions unless they are scalars, obviously.

Edit: Well, you can define some sort of division but it wouldn't be like the case we know from calculus.
 
It seems to me that the next identity should be fulfilled.

If ##f(x_1,\cdots , x_m)=(f_1,\cdots, f_n)## and ##g(x_1,\cdots,x_m)=(g_1,\cdots,g_n)##, then:

##||f||/||g|| = \sqrt{f^2_1+\cdots+f^2_n}/\sqrt{g^2_1+\cdots + g^2_n}##

Now it seems to me to be plausible that:
##\lim_{x\to 0} ||f||/||g|| = \frac{\sqrt{(grad \ f_1)^2+\cdots + (grad \ f_n )^2}}{\sqrt{(grad \ g_1)^2+\cdots + (grad \ g_n )^2}}(at \ x=0)##
 

Similar threads

  • · Replies 20 ·
Replies
20
Views
5K
  • · Replies 6 ·
Replies
6
Views
2K
  • · Replies 9 ·
Replies
9
Views
2K
  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 0 ·
Replies
0
Views
2K
  • · Replies 10 ·
Replies
10
Views
3K
  • · Replies 12 ·
Replies
12
Views
4K
  • · Replies 12 ·
Replies
12
Views
3K
  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 9 ·
Replies
9
Views
3K