Continuity of partial derivatives in a ball implies differentiability

Join the discussion
Registration is free. Ask a follow-up in this thread, or start your own.
3 replies · 6K views
Mathmos6
Messages
76
Reaction score
0
Hi all, I'm looking at the following problem:
Suppose that f:[itex]\mathbb{R}^2\to\mathbb{R}[/itex] is such that [itex]\frac{\partial{f}}{\partial{x}}[/itex] is continuous in some open ball around (a,b) and [itex]\frac{\partial{f}}{\partial{y}}[/itex] exists at (a,b): show f is differentiable at (a,b).

Now I know that if both partial derivatives are continuous in a ball around (a,b) it is differentiable, but I don't know how to deal with the case where we only have continuity of one partial derivative and existence of the other - could anyone help me with this?

I've thought about it for a good hour or so now and don't seem to be getting anywhere (I know I'm meant to post what I've got so far but I literally have nothing!) so the more help you can give me the better!

Many thanks :)
 
Last edited:
Physics news on Phys.org
Use the definition of differentiable.

A function of two variables, x and y, is "differentiable" at (a, b) if and only if there exist a function [itex]\epsilon(x,y)[/itex] such that [itex]f(a+ x, b+ y)= f(a, b)+ x (\partial f/\partial x(a,b))[/itex][itex]+ y(\partial f/\partial y(a,b))+ \epsilon(x,y)[/itex] and
[tex]\lim_{(x,y)\to (a,b)}\frac{\epsilon(x,y)}{\sqrt{x^2+ y^2}}= 0[/tex]

Since both [itex]\partial f/\partial x(a,b)[/itex] and [itex]\partial f/\partial y(a,b)[/itex] exist, we can define [itex]\epsilon(x,y)[/itex] as [itex]f(a+ x, b+ y)- f(a, b)[/itex][itex]+ x (\partial f/\partial x(a,b))+ y(\partial f/\partial y(a,b))[/itex].

Try to use the fact that [itex]\partial f/\partial x[/itex] is continuous to show that goes to 0 as (x,y) goes to (0, 0).
 
HallsofIvy said:
Use the definition of differentiable.

A function of two variables, x and y, is "differentiable" at (a, b) if and only if there exist a function [itex]\epsilon(x,y)[/itex] such that [itex]f(a+ x, b+ y)= f(a, b)+ x (\partial f/\partial x(a,b))[/itex][itex]+ y(\partial f/\partial y(a,b))+ \epsilon(x,y)[/itex] and
[tex]\lim_{(x,y)\to (a,b)}\frac{\epsilon(x,y)}{\sqrt{x^2+ y^2}}= 0[/tex]

Since both [itex]\partial f/\partial x(a,b)[/itex] and [itex]\partial f/\partial y(a,b)[/itex] exist, we can define [itex]\epsilon(x,y)[/itex] as [itex]f(a+ x, b+ y)- f(a, b)[/itex][itex]+ x (\partial f/\partial x(a,b))+ y(\partial f/\partial y(a,b))[/itex].

Try to use the fact that [itex]\partial f/\partial x[/itex] is continuous to show that goes to 0 as (x,y) goes to (0, 0).

Got it, thanks very much! :)