I've been asked to construct a proof for the intermediate value theorem in higher dimensions, $\mathbb{R}^n$.


Let $f:\mathbb{R}^n\rightarrow \mathbb{R}$ be a continuous function. Let a and b be points in $\mathbb{R}^n$.

Define the "help" function $g(t)=f(ta+(1-t)b)$


Prove that if d is a number such that $f(a)<d<f(b)$, then there exists a point $c\in\mathbb{R}^n$ such that $f(c)=d$.

Give specific proofs in the cases where f is defined on a rectangle in $\mathbb{R}^2$ and a ball in $\mathbb{R}^3$

Solution (Attempted)

g is a (continuous) function in $\mathbb{R}$, which means we can use the "normal" Intermediate value theorem (proof already given in our book). So for any d between $g(t)$ and $g(t')$ there is a c such that $g(c)=d$.

We note that $g(0)=f(b)$ and $g(1)=f(a)$. So $g[0,1]=[f(b),f(a)]$.

Is this sufficient? I feel that it's enough to conclude that f attains any value d between $f(a)$ and $f(b)$.

Sadly, I am rather clueless on the issue, when f's domain is narrowed to being a rectangle, or a ball.

  • $\begingroup$ What you wrote is enough to me. I am not sure I understand what is being asked by "specific proofs" for the other cases, but can you imagine domains where it does not work ? $\endgroup$
    – nicomezi
    Feb 3, 2020 at 15:11
  • $\begingroup$ Looks fine to me. $g$ is a bijection between $[0,1]$ and $[a,b]$. $\endgroup$
    – copper.hat
    Feb 3, 2020 at 15:11
  • $\begingroup$ @nicomezi: Why not? $\endgroup$
    – copper.hat
    Feb 3, 2020 at 15:15
  • $\begingroup$ Yes, you are correct, I will fix it when I get to my laptop $\endgroup$
    – copper.hat
    Feb 3, 2020 at 15:18
  • $\begingroup$ As @nicomezi pointed out, the bijection is $t \mapsto ta + (1-t)b$. $\endgroup$
    – copper.hat
    Feb 3, 2020 at 16:05

1 Answer 1


Your proof is okay apart from some writing stuff $($like $[f(b),f(a)]$ when we may have $f(a) < f(b))$.

It does not matter that $f$'s domain is a rectangle or a ball, provided they're filled. What matters for this proof is that the segment $ta+(1-t)b$ lie in the domain.

More generally, what matters here is continuity and connectedness. The intermediate value theorem is a consequence of two general facts: continuous functions map connected sets to connected sets, and connected sets of the real line are intervals.

  • $\begingroup$ Well yeah, it did seem very backwards to write $g[0,1]=[f(b),f(a)]$. But I'm not entirely sure how to fix it. Notation wise. Do you just say "assume $f(a)<f(b)$ and then $g[0,1]=[f(a),f(b)$ without loss of generality? $\endgroup$ Feb 3, 2020 at 17:28
  • $\begingroup$ Notice that while $\{g(0),g(1)\} = \{f(a),f(b)\}$, for $x\in(0,1)$ one might even have $g(x)$ outside of $[f(a),f(b)]$ or $[f(b),f(a)]$. Regardless of whether $g(0) = f(a)$ or $g(0) = f(b)$, regardless of whether $f(a)<f(b)$ or $f(a) > f(b)$, and regardless of whether $g([0,1]) = [f(a),f(b)]$ or strictly contains it, the theorem holds. The usual way to say it with words: for each $y$ between $f(a)$ and $f(b)$, there is some $c$ between $0$ and $1$ with $g(c) = y$. $\endgroup$ Feb 3, 2020 at 17:35
  • $\begingroup$ For instance, consider $f(x) = x^2$. With $a = -1$ and $b = 2$ we have $f(a) = 1$ and $f(b) = 4$. But we don't have $f([a,b]) = [1,4]$, we have $f([a,b]) = [0,4]$. The theorem still holds, and for each $y\in(1,4)$ we can find $x\in(-1,2)$ with $f(x) = y$. $\endgroup$ Feb 3, 2020 at 17:37

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.