I vaguely remember hearing that all global extrema are local extrema long ago. However, now that I look carefully at the definition, it appears that if the domain of a function is closed, then a value can be a global minimum without being a local minimum.

For example, imagine a $f(x)=x^2, 0\leq x \leq 1$. Then, 0 and 1 are a global minimum and maximum, respectively, but since $f(x)$ is not equal to $0$ or $1$ on for $x \in I$ where $I$ is an open interval and contained within the domain of the function, $0$ and $1$ cannot be local extrema.

Is my logic correct or am I missing something? If I'm wrong, what am I missing?

(Sorry for the somewhat "yes" or "no" nature of the question. I just can't find the answer elsewhere.)


  • 2
    $\begingroup$ You are missing the fact that to be a local extremum the required inequality has to hold in a relative neighborhood of the point in the domain of the function, that is, in the intersection of an open interval with the domain of the function. $\endgroup$ – Mariano Suárez-Álvarez Jan 4 '12 at 23:30

The interval $(0.9,1]$ is indeed an open subset of the space $[0,1]$, and the maximum at $1$ is a local maximum.

What you say you "vaguely remember" is correct.

  • $\begingroup$ One sec. Let's say $f(x)=x^2, \forall x \in \mathbb{R}$. Then isn't the interval $(0.9,1]$ still an open subset of $(-\infty,\infty)$ such that 1 is a local maximum? $\endgroup$ – Deets McGeets Jan 5 '12 at 3:31
  • $\begingroup$ $(0.9,1]$ is not an open subset of $\mathbb{R}$, but it is an open subset of $[0,1]$. $\endgroup$ – Michael Hardy Jan 5 '12 at 19:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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