Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

A potentially naïve question about finding the minimum of a function:

If I have a scalar function of a scalar variable $y=f(x)$, which is continuously differentiable, I can find minima but finding the values of $x$ for which $f'(x)=0$.

On the other hand, if I have a scalar function of a vector variable, $y=f(\mathbf{x})$, can I do the same? Can I always find the value of $\mathbf{x}$ where $f'(\mathbf{x})=0$?

I am aware I can take partial derivatives for each value of $\mathbf{x}$, i.e.

$\nabla y = \left[\frac{\partial y}{\partial x_1}, \frac{\partial y}{\partial x_2}, \dots \frac{\partial y}{\partial x_N} \right]$

But how do I go from this vector of derivatives to a final value for $\mathcal{x}$? Do I just set each derivative to zero?

share|improve this question
    
Yes, that is a necessary condition for extremum for $C^1$ functions on $\mathbb R^n$: $\frac{\partial y}{\partial x_i}(\hat{\mathbf x}) = 0$ for all $i=1,\dots,n$ whenever $\hat{\mathbf x}$ is an extremum argument of the function $y$. –  Ilya Aug 1 '12 at 16:17
add comment

1 Answer

The theorem goes as follows: Let $U\subseteq \mathbb{R}^n$ be an open subset, and let $x_0\in U$. Let $f:U\to \mathbb{R}$ be any function such that $x_0$ is a local extremum of $f$. If all partial derivatives of $f$ exist at $x_0$ then $\nabla f(x_0) = 0$. In particular, if $f$ is differentiable at $x_0$ then $\nabla f(x_0) = 0$.

This means that if you want to find extrema of a differentiable multivariable function defined on an open set you may search for points at which all the partial derivatives vanish simultaneously. Such points are called critical points. Any local extremum of $f$ must be a critical point, however not every critical point is a local extremum. Classifying the critical points requires some more work.

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.