TheoremComplete

Intermediate Value Theorem (Topological)

The intermediate value theorem, a cornerstone of real analysis, is fundamentally a statement about connected spaces and continuous maps. The topological version reveals the true nature of this result and extends it far beyond functions on the real line.


Topological Statement

Theorem4.9Generalized Intermediate Value Theorem

Let f:XYf: X \to Y be a continuous map where XX is connected and YY is an ordered topological space (or more generally, Y=RY = \mathbb{R}). If a,bf(X)a, b \in f(X) with a<ba < b, then for every cc with a<c<ba < c < b, we have cf(X)c \in f(X).

In other words, the image f(X)f(X) is an interval (or a connected subset of YY).

Proof

By Theorem 4.3 (or Theorem 2.8), the continuous image f(X)f(X) of the connected space XX is connected. Since f(X)Rf(X) \subseteq \mathbb{R} is connected, it is an interval (by Theorem 4.2). Since a,bf(X)a, b \in f(X) and f(X)f(X) is an interval, every cc between aa and bb belongs to f(X)f(X).


Classical Intermediate Value Theorem

Theorem4.10Classical IVT

Let f:[a,b]Rf: [a, b] \to \mathbb{R} be continuous. If cc is any value between f(a)f(a) and f(b)f(b), then there exists x0[a,b]x_0 \in [a, b] with f(x0)=cf(x_0) = c.

Proof

[a,b][a, b] is connected (it is an interval), so f([a,b])f([a, b]) is a connected subset of R\mathbb{R}, hence an interval. Since f(a)f(a) and f(b)f(b) belong to f([a,b])f([a,b]), every value between them does as well.

ExampleExistence of Roots

Consider f(x)=x5+x1f(x) = x^5 + x - 1 on [0,1][0, 1]. We have f(0)=1<0f(0) = -1 < 0 and f(1)=1>0f(1) = 1 > 0. By the IVT, there exists c(0,1)c \in (0, 1) with f(c)=0f(c) = 0. The IVT guarantees existence but does not provide a formula for cc.

More generally, every polynomial of odd degree has a real root: for large x|x|, the leading term dominates, giving opposite signs at ++\infty and -\infty.


Applications

Theorem4.11Fixed Point Theorem on $[0,1]$

Let f:[0,1][0,1]f: [0, 1] \to [0, 1] be continuous. Then ff has a fixed point: there exists x[0,1]x \in [0, 1] with f(x)=xf(x) = x.

Proof

Define g:[0,1]Rg: [0, 1] \to \mathbb{R} by g(x)=f(x)xg(x) = f(x) - x. Then gg is continuous, g(0)=f(0)0g(0) = f(0) \geq 0, and g(1)=f(1)10g(1) = f(1) - 1 \leq 0. By the IVT, there exists c[0,1]c \in [0, 1] with g(c)=0g(c) = 0, i.e., f(c)=cf(c) = c.

ExampleThe IVT in Higher Dimensions

The IVT generalizes to higher dimensions in various ways:

  1. Borsuk--Ulam theorem (dimension 1): For every continuous f:S1Rf: S^1 \to \mathbb{R}, there exist antipodal points x,xS1x, -x \in S^1 with f(x)=f(x)f(x) = f(-x). This follows from the IVT applied to g(x)=f(x)f(x)g(x) = f(x) - f(-x).

  2. Bolzano's theorem: If f:RnRf: \mathbb{R}^n \to \mathbb{R} is continuous on a connected domain DD and takes both positive and negative values, then ff vanishes somewhere on DD.


Connected Subsets of Ordered Spaces

Theorem4.12Connected Subsets of Linear Continua

Let LL be a linear continuum (a totally ordered set with the least upper bound property in which every pair of points has a point strictly between them). Then every interval in LL is connected, and the connected subsets of LL are exactly the intervals.

Proof

Let [a,b]L[a, b] \subseteq L and suppose [a,b]=AB[a, b] = A \cup B is a separation with aAa \in A. Let c=sup(A[a,b])c = \sup(A \cap [a, b]). Since AA is closed in [a,b][a, b] (complement of the open set BB), cAc \in A. Since AA is open in [a,b][a, b] and cbc \neq b (otherwise bABb \in A \cap B), there exists ϵ\epsilon with (c,c+ϵ)[a,b]A(c, c + \epsilon) \cap [a, b] \subseteq A. But this contradicts c=sup(A[a,b])c = \sup(A \cap [a, b]). So no separation exists.

Conversely, if SLS \subseteq L is not an interval, there exist a,bSa, b \in S and cSc \notin S with a<c<ba < c < b. Then S=(S(,c))(S(c,))S = (S \cap (-\infty, c)) \cup (S \cap (c, \infty)) is a separation.

RemarkThe Role of Completeness

The IVT fundamentally depends on the completeness (least upper bound property) of R\mathbb{R}. It fails for Q\mathbb{Q}: the function f(x)=x22f(x) = x^2 - 2 is continuous on Q\mathbb{Q}, negative at x=1x = 1 and positive at x=2x = 2, but has no rational root.

Completeness ensures that R\mathbb{R} has no "gaps," which is precisely what connectedness formalizes.