OU blog

Personal Blogs

Zafar Bakhromov

Confusion while studying Vector Calculus from Hubbard & Hubbard

Visible to anyone in the world
Edited by Zafar Bakhromov, Monday 22 September 2025 at 16:02

On page 93 of John & Barbara Hubbard's Vector Calculus, Linear Algebra and Differential Forms, I noticed something unusual. The authors state that functional limits behave well under composition (Theorem 1.5.22).

More formally, let cap u subset of double-struck cap r super n comma cap v subset of double-struck cap r super m be subsets, and bold f colon cap u right arrow cap v and bold g colon cap v right arrow double-struck cap r super k be mappings, so that bold g ring operator bold f is defined cap u . If bold x sub bold zero is a point in cap u and 

bold y sub bold zero colon equals lim over bold x right arrow bold x sub bold zero of bold f of bold x and bold z sub bold zero colon equals lim over bold x right arrow bold x sub bold zero of bold g of bold y

both exist, then lim over bold x right arrow bold x sub bold zero of bold g ring operator bold f times left parenthesis bold x right parenthesis exists, and 

lim over bold x right arrow bold x sub bold zero of bold g ring operator bold f times left parenthesis bold x right parenthesis equals bold z sub bold zero

At first, this seemed false, because in order for it to be true the function bold g needs to be continuous. But no such assumption is made here. I was scratching my head the whole afternoon trying to figure this out, and after a long time of googling, I found the following result in wikipedia:

sketch.png

It seems like whether bold g of bold y not equals bold g of bold y sub bold zero whenever bold y is near bold y sub bold zero is important. So I looked back on the definition of function limits in the book, and I found the key fact that was confusing me: the authors did not require absolute value of bold x minus bold x sub bold zero greater than zero whenever bold x is near bold x sub bold zero ! Appearently this definition of functional limits is common in France, while the usual definition is common in the United States.

As an example, consider the functions f comma g colon double-struck cap r right arrow double-struck cap r

f of x colon equals case statement case 1 x times sine of one divided by x if x not equals zero case 2 zero if x equals zero and g of y colon equals case statement case 1 one if y not equals zero case 2 zero if y equals zero

Then, we get the following table of facts:

  French Definition US Definition
lim over x right arrow zero of f of x exists? Yes Yes
lim over y right arrow zero of g of y exists? No Yes
lim over x right arrow zero of g of f of x exists? No No
Is Theorem 1.5.22 True? Yes (vacuously: the hypotheses aren't met) No

It is straightforward to show that lim over y right arrow zero of g of y does not exist under the French definition, because for any c we can always find an y element of left parenthesis negative delta comma delta right parenthesis s.t. absolute value of g of y minus c greater than or equals one (Here, c is either equal to 0 or 1). Hence, the theorem is true.

This little detour reminded me that even definitions we take for granted can vary by culture, and those small differences can have big consequences. For me, it was a reminder that part of learning mathematics is also learning to pay very close attention to the precise conventions being used.

Permalink Add your comment
Share post