Why Can't the Gradient Vector Be Undefined at a Minimum?

Click For Summary
The discussion centers on the properties of the gradient vector at local extrema of functions of two variables. It questions why the gradient cannot be undefined at a minimum and explores the implications of undefined partial derivatives at critical points. Participants note that while a gradient can be undefined, determining the nature of a critical point analytically requires careful consideration, especially when second-order tests fail. The conversation highlights that some functions may lack local extrema or behave unpredictably, necessitating alternative methods to identify optima when traditional derivative techniques fail. Overall, the complexities of analyzing extrema in multivariable calculus are emphasized.
quasar987
Science Advisor
Homework Helper
Gold Member
Messages
4,796
Reaction score
32
I have few questions about extrema of fonctions of two variables. It is written in my textbook: "At a local maximum, the gradient vector must be nul or undefined. A similar reasoning shows that the gradient must be nul at a local minimum." Actually there was no preceeding reasoning to this statement so I don't understand.

- Why couldn't the gradient vector be undefined at the min?

- If one of the partial derivative is undefined at a certain point, does it automatically means the point is a max? If no, how do you tell analytically?

- If (a,b) is a critical point because the gradient at (a,b) is 0 and if the test of the second order partial derivative fails (i.e. =0). How can I conclude analytically to the nature of the critial point?
 
Last edited:
Physics news on Phys.org
The author has been sloppy!
- Why couldn't the gradient vector be undefined at the min?
It certainly can be!
.
 
Ok, then I will rewrite question #2:

- If one of the partial derivatives is undefined at a certain point, how do I conclude analytically to the nature of the critical point?

- Could it be that a derivative in a direction other than the x or y-axis is undefined while it is defined in the direction of the x and y axis? In this case wouldn't the method of analysis of the partial derivatives fail to detect the critical point?
 
Remember that the definition of optima don't involve derivatives at all -- when differential techniques fail, you often have to resort to the definitions to get your answers.
 
Hurkyl said:
Remember that the definition of optima don't involve derivatives at all -- when differential techniques fail, you often have to resort to the definitions to get your answers.
And, I would think, some functions might be perverse enough to refuse yielding up where its extrema are, despite our best efforts..
 
Even worse, there are functions that don't even have local extrema!

example: (here, p and q are relatively prime)


f(x) = 0 if x is irrational
f(p/q) = 1 - 1/q if q is even
f(p/q) = -1 + 1/q if q is odd
 
Last edited:

Similar threads

  • · Replies 9 ·
Replies
9
Views
3K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 8 ·
Replies
8
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 4 ·
Replies
4
Views
2K
Replies
2
Views
4K
  • · Replies 9 ·
Replies
9
Views
3K