math.wiki
Analysis › Continuity › Extreme and intermediate value theorems, proved
Sign in

Extreme and intermediate value theorems, proved

Calculus proved, by halving intervals, that a continuous function on a closed interval has a highest and a lowest value and skips no value between. Why do these theorems hold on every compact set and every interval, and what does a continuous function do to such a set as a whole?

Where you are 5 in Continuity

Why this exists

The extreme value theorem and The intermediate value theorem proved the two existence theorems of calculus on [a, b] by halving intervals. Analysis asks which sets carry them, and what a continuous function does to a whole set. This lesson proves the extreme value theorem on every compact set of Open, closed and compact sets in R, from Subsequences and Bolzano-Weierstrass, and the intermediate value theorem from the least-upper-bound property. Both become one statement: a continuous function carries compact sets to compact sets and intervals to intervals.

Definition

Throughout, S \subseteq \mathbb{R} and f : S \to \mathbb{R}. The image of A \subseteq S is f(A) = \{f(x) : x \in A\}.

Bounded. f is bounded if there is B with \lvert f(x) \rvert \le B for every x \in S.

Absolute extremes. f attains an absolute maximum at c \in S if f(x) \le f(c) for every x \in S, and an absolute minimum at d \in S if f(d) \le f(x) for every x \in S. For bounded f on nonempty S, this says exactly that \sup f(S) or \inf f(S) belongs to f(S).

Interval. A set I \subseteq \mathbb{R} is an interval if whenever x, y \in I and x < z < y, also z \in I. The empty set and single points count; Theorem 4 shows the rest are the sets of Interval and set-builder notation.

Intuition

A compact set keeps the limits of its sequences and lets none run off. Take points where f climbs toward \sup f(K): a subsequence settles at a point of K, and continuity carries the climb there, so the supremum is a value. An interval has no gaps, and a continuous function cannot jump, so its values have no gaps either. Figure 1 shows the first idea.

x y −3 −2 −1 3 4 16 −9
f(K)
Figure 1Example 1. On the compact set K = [-3, -1] \cup [3, 4] the function x^3 - 12x attains its maximum 16 twice and its minimum -9 once. The image f(K) = [-9, 16], thick on the vertical axis, happens to be an interval although K is not.

Results

Theorem 1 (compact images). If K is compact and f : K \to \mathbb{R} is continuous, then f(K) is compact.

Theorem 2 (extreme value theorem). If K is nonempty and compact and f : K \to \mathbb{R} is continuous, then f is bounded and attains an absolute maximum and an absolute minimum on K. In particular this holds on every [a, b].

Theorem 3 (Bolzano’s intermediate value theorem). Let f : [a, b] \to \mathbb{R} be continuous, and let f(a) < y < f(b) or f(a) > y > f(b). Then f(c) = y for some c \in (a, b).

Theorem 4 (intervals). (a) A set is an interval if and only if it is empty, a single point, or one of (\alpha, \beta), [\alpha, \beta), (\alpha, \beta], [\alpha, \beta] with \alpha < \beta, (\alpha, \infty), [\alpha, \infty), (-\infty, \beta), (-\infty, \beta], \mathbb{R}. (b) If I is an interval and f : I \to \mathbb{R} is continuous, then f(I) is an interval.

Theorem 5 (closed bounded intervals). If f : [a, b] \to \mathbb{R} is continuous with minimum value m and maximum value M, then f([a, b]) = [m, M], a single point when m = M.

Theorem 6 (inverses). Let I be an interval and f : I \to \mathbb{R} continuous and injective. Then f is strictly monotone, J = f(I) is an interval, and f^{-1} : J \to I is continuous.

So \displaystyle \sqrt[n]{x} on [0, \infty) and \arcsin on [-1, 1] are continuous.

Why it is true

Continuity enters through the definition and the sequential criterion, Theorem 3 of Continuity, defined and proved.

Theorem 1. Proof. Let \{y_n\} be a sequence in f(K), with y_n = f(x_n) and x_n \in K. As K is compact, a subsequence x_{n_k} converges to some x \in K. By the sequential criterion f(x_{n_k}) \to f(x), so \{y_{n_k}\} converges to f(x) \in f(K). \square

Theorem 2. Proof. By Theorem 1, f(K) is compact and nonempty. So it is bounded (Theorem 7 of Open, closed and compact sets in R), and \sup f(K) and \inf f(K) belong to it (Theorem 8 there): they equal f(c) and f(d) for some c, d \in K. Finally [a, b] is closed and bounded, hence compact. \square

Theorem 3. Proof. Suppose f(a) < y < f(b) and let g = f - y, continuous with g(a) < 0 < g(b). The set A = \{x \in [a, b] : g(x) < 0\} contains a and is bounded above by b, so c = \sup A exists (Least upper bounds and completeness) and a \le c \le b.

If g(c) < 0, then c \ne b. Continuity at c with \varepsilon = -g(c) gives \delta > 0 with g(x) < g(c) + \varepsilon = 0 for every x \in [a, b] with \lvert x - c \rvert < \delta. Then \displaystyle \min\left\{c + \frac{\delta}{2}, b\right\} lies in A and exceeds \sup A, which is impossible.

If g(c) > 0, then c \ne a. With \varepsilon = g(c) there is \delta > 0 with g(x) > 0 for every x \in [a, b] with \lvert x - c \rvert < \delta. So no point of A lies in (c - \delta, c], and none lies above c: c - \delta is an upper bound of A below \sup A, which is impossible.

Hence g(c) = 0, so f(c) = y, and c \ne a, b since g(a) \ne 0 \ne g(b). If f(a) > y > f(b), apply this to -f and -y. \square

Theorem 4. Proof. (a) Each listed set has the property, since a z between two of its points satisfies the same inequalities. Conversely, let I be an interval with at least two points. Let \alpha = \inf I, or -\infty if I is not bounded below, and \beta = \sup I, or \infty if I is not bounded above. If \alpha < z < \beta, then z is not a lower bound of I, so some x \in I has x < z; likewise some y \in I has y > z; so z \in I. Thus I is (\alpha, \beta) with possibly one or both finite ends added.

(b) Let f(x_1) < z < f(x_2) with x_1, x_2 \in I. The closed interval with ends x_1 and x_2 lies in I, and Theorem 3 on it gives c with f(c) = z. \square

Theorem 5. Proof. By Theorem 2, m and M exist and are values, and f([a, b]) \subseteq [m, M]. By Theorem 4(b), f([a, b]) is an interval containing m and M, so it contains [m, M]. \square

Theorem 6. Proof. Monotone. If not, then, as f is injective, there are a < b in I with f(a) < f(b) and u < v in I with f(u) > f(v). For t \in [0, 1] let x_t = (1 - t)a + tu and y_t = (1 - t)b + tv. Each lies in I, between a and u or between b and v, and x_t < y_t since a < b and u < v. The function h(t) = f(y_t) - f(x_t) is continuous on [0, 1] (Theorems 4 to 6 of Continuity, defined and proved) with h(0) > 0 > h(1), so Theorem 3 gives t with f(x_t) = f(y_t), against injectivity.

Inverse. J is an interval by Theorem 4(b). Let f be increasing, so g = f^{-1} is strictly increasing on J. Let d \in J, c = g(d) and \varepsilon > 0. If I has a point p > c, then \displaystyle c' = \min\left\{p, c + \frac{\varepsilon}{2}\right\} lies in I with c < c' < c + \varepsilon; put d' = f(c') > d. If I has a point below c, choose c'' \in (c - \varepsilon, c) in I and d'' = f(c'') < d the same way. Let \delta be the least of d' - d and d - d'' among those that exist, or 1. If x \in J and \lvert x - d \rvert < \delta, then g(x) < c' if c' exists and g(x) \le c if not, so g(x) < c + \varepsilon; likewise g(x) > c - \varepsilon. If f is decreasing, apply this to -f, whose inverse is y \mapsto g(-y). \square

Worked examples

Example 1

Extremes on a set with a gap

Let f(x) = x^3 - 12x. (a) On K = [-3, -1] \cup [3, 4], explain why f attains a maximum and a minimum, find them, and find f(K). (b) On E = [-3, -1] \cup (3, 4], which extremes are attained?

Solution

(a) K is a union of two closed intervals, so closed (Theorem 5 of Open, closed and compact sets in R), and bounded, so compact; a polynomial is continuous, and Theorem 2 applies. Since f'(x) = 3(x - 2)(x + 2), by the sign of the derivative (The mean value theorem) f increases on [-3, -2], decreases on [-2, -1] and increases on [3, 4]. With f(-3) = 9, f(-2) = 16, f(-1) = 11, f(3) = -9 and f(4) = 16, Theorem 5 on each piece gives f([-3, -1]) = [9, 16] and f([3, 4]) = [-9, 16]. So f(K) = [-9, 16]: the maximum 16 is attained at -2 and at 4, the minimum -9 at 3 (Figure 1).

(b) E is not closed: \displaystyle 3 + \frac{1}{n} \in E tends to 3 \notin E. On (3, 4], f is increasing, so its values there exceed -9, while \displaystyle f\left(3 + \frac{1}{n}\right) \to -9. By Theorem 4(b), f((3, 4]) = (-9, 16], and f(E) = (-9, 16]. The infimum is not attained, so there is no minimum, while the maximum 16 still is.

Example 2

The nearest point of a compact set

Let K be nonempty and compact and p \in \mathbb{R}. Prove that some q \in K is nearest to p: \lvert p - q \rvert \le \lvert p - x \rvert for every x \in K. Then find the nearest points of \displaystyle K = \{0\} \cup \left\{\frac{1}{n} : n \ge 1\right\} to p = 0.3 and to \displaystyle p = \frac{3}{4}, and show that \displaystyle E = \left\{\frac{1}{n} : n \ge 1\right\} has no point nearest to \displaystyle -\frac{1}{2}.

Solution

Proof. The function x \mapsto \lvert x - p \rvert is continuous, as a polynomial composed with the absolute value (Theorems 5 and 6 of Continuity, defined and proved). By Theorem 2 it attains a minimum on K at some q. \square

K is compact: its only limit point is 0, which it contains, and it lies in [0, 1]. For p = 0.3, between \displaystyle \frac{1}{4} and \displaystyle \frac{1}{3}, the distances are \displaystyle \left\lvert 0.3 - \frac{1}{3} \right\rvert = \frac{1}{30} and \displaystyle \left\lvert 0.3 - \frac{1}{4} \right\rvert = \frac{1}{20}, and every other point is farther than one of these. So \displaystyle q = \frac{1}{3}. For \displaystyle p = \frac{3}{4}, the points \displaystyle \frac{1}{2} and 1 are both at distance \displaystyle \frac{1}{4}, with no point of K between: the theorem gives existence, not uniqueness.

For E, the distances to \displaystyle -\frac{1}{2} are \displaystyle \frac{1}{2} + \frac{1}{n}, with infimum \displaystyle \frac{1}{2} never reached. E misses its limit point 0, so it is not compact.

Example 3

Which zero the proofs find

Let g(x) = 8x^3 - 6x + 1 on [-1, 1] and A = \{x \in [-1, 1] : g(x) < 0\}. (a) Find the zero \sup A that the proof of Theorem 3 produces. (b) Run four steps of bisection from [-1, 1] and say which zero it closes in on.

Solution

(a) g(-1) = -1 < 0 < 3 = g(1). For x = \cos\theta with 0 \le \theta \le \pi, the identity \cos 3\theta = 4\cos^3\theta - 3\cos\theta gives g(\cos\theta) = 2\cos 3\theta + 1, which vanishes when \displaystyle \cos 3\theta = -\frac{1}{2}, at \displaystyle \theta = \frac{2\pi}{9}, \frac{4\pi}{9}, \frac{8\pi}{9}. A cubic has at most three zeros, so they are

\begin{aligned} r_1 &= \cos\tfrac{8\pi}{9} \approx -0.9397 \\ r_2 &= \cos\tfrac{4\pi}{9} \approx 0.1736 \\ r_3 &= \cos\tfrac{2\pi}{9} \approx 0.7660 \end{aligned}

The zeros are simple, so g changes sign at each, and A = [-1, r_1) \cup (r_2, r_3). Its supremum is r_3 \approx 0.7660 (Figure 2).

x y
r_1
r_2
r_3 = \sup A
−1 1 3
Figure 2Example 3. The set A = \{x \in [-1, 1] : g(x) < 0\}, thick, with open ends at the zeros. The proof of Theorem 3 finds r_3 = \sup A; bisection from [-1, 1] finds r_1.

(b) Bisection keeps a half on which g changes sign. g(0) = 1 > 0 keeps [-1, 0]; \displaystyle g\left(-\frac{1}{2}\right) = 3 keeps \displaystyle \left[-1, -\frac{1}{2}\right]; \displaystyle g\left(-\frac{3}{4}\right) = \frac{17}{8} keeps \displaystyle \left[-1, -\frac{3}{4}\right]; \displaystyle g\left(-\frac{7}{8}\right) = \frac{57}{64} keeps \displaystyle \left[-1, -\frac{7}{8}\right]. These contain only the zero r_1, so bisection closes in on r_1 \approx -0.9397. Two proofs of one theorem pick different solutions, and neither finds r_2.

Example 4

Images of intervals

Let f(x) = \dfrac{x}{1 + x^2}. Find f(I) for (a) I = [-1, 3], (b) I = (0, \infty), (c) I = (-\infty, -2].

Solution

f is a quotient of polynomials with denominator at least 1, so continuous, and f'(x) = \dfrac{1 - x^2}{(1 + x^2)^2}: f increases on [-1, 1] and decreases on (-\infty, -1] and on [1, \infty). By Theorem 4(b) each image is an interval, settled by its infimum and supremum and whether they are attained.

(a) By Theorem 5, the image is [m, M] with \displaystyle m = f(-1) = -\frac{1}{2} and \displaystyle M = f(1) = \frac{1}{2}, as \displaystyle f(3) = \frac{3}{10} lies between: \displaystyle \left[-\frac{1}{2}, \frac{1}{2}\right].

(b) The maximum is \displaystyle f(1) = \frac{1}{2}. Every value is positive and \displaystyle f(n) = \frac{n}{1 + n^2} \to 0, so the image is \displaystyle \left(0, \frac{1}{2}\right].

(c) f decreases on the ray, so the minimum is \displaystyle f(-2) = -\frac{2}{5}; every value is negative and f(-n) \to 0. The image is \displaystyle \left[-\frac{2}{5}, 0\right).

Example 5

Elena’s conservatory

Elena manages the palm house of a city conservatory, and Omar logs its air temperature T(t), in degrees Fahrenheit, t hours after midnight. (a) Prove that if T is continuous on [0, 24] with T(24) = T(0), then T(t) = T(t + 12) for some t \in [0, 12]. (b) Omar’s model for one day is

T(t) = 64 - 9\cos\tfrac{\pi t}{12} + 4\sin\tfrac{\pi t}{6}

Check T(24) = T(0), and find every t \in [0, 12] with T(t) = T(t + 12) and the common temperature. (c) Show that the model reaches 70 degrees between 6 and 12 hours.

Solution

(a) Proof. Let h(t) = T(t + 12) - T(t), continuous on [0, 12] by Theorems 4 and 5 of Continuity, defined and proved. Then h(12) = T(24) - T(12) = T(0) - T(12) = -h(0). If h(0) = 0, take t = 0. Otherwise h(0) and h(12) have opposite signs, and Theorem 3 gives t \in (0, 12) with h(t) = 0. \square

(b) T(0) = 64 - 9 = 55 = T(24). Since \displaystyle \cos\left(\frac{\pi t}{12} + \pi\right) = -\cos\frac{\pi t}{12} and \displaystyle \sin\left(\frac{\pi t}{6} + 2\pi\right) = \sin\frac{\pi t}{6},

h(t) = 18\cos\tfrac{\pi t}{12}

which vanishes on [0, 12] only at t = 6. There T(6) = T(18) = 64 (Figure 3).

t T 6 12 18 24 55 64 73
Figure 3Example 5. The palm-house model T(t) in degrees Fahrenheit, with the vertical axis starting at 50. The readings at 6 and at 18 hours agree.

(c) T(6) = 64 < 70 < 73 = T(12), so Theorem 3 gives a time in (6, 12) at 70 degrees.

Example 6

No function takes every value exactly twice

Prove that no continuous f : \mathbb{R} \to \mathbb{R} takes every real value exactly twice.

Solution

Proof. Suppose f does, and let a < b with f(a) = f(b) = y_0. No x \in (a, b) has f(x) = y_0, so f - y_0 keeps one sign on (a, b): if f(s) < y_0 < f(t) there, Theorem 3 would put a third solution of f(x) = y_0 between s and t. Replacing f by -f, which also takes every value exactly twice, we may assume f > y_0 on (a, b).

By Theorem 2, f attains a maximum M on [a, b] at some c; M > y_0, so c \in (a, b). The value M is taken at exactly one other point d.

If d > b, choose y with y_0 < y < M. Theorem 3 gives solutions of f(x) = y in (a, c), in (c, b) and in (b, d): three of them. The case d < a is the same with (d, a).

If d \in (a, b), say c < d, then f is not constant on [c, d], or M would be taken infinitely often, so f(e) < M for some e \in (c, d). Choose y with \max\{y_0, f(e)\} < y < M. Theorem 3 gives solutions in (a, c), (c, e) and (e, d): three again. The case d < c is symmetric. \square

Boundaries

  • Each hypothesis of Theorem 2 is needed. Example 1(b) drops closedness, Example 4(c) boundedness, Exercise 4 continuity; The extreme value theorem tabulates the same failures.
  • Other sets keep no type. Under a continuous map, open, closed and half-open intervals trade places (Example 4); only compactness and being an interval survive.
  • The converse of Theorem 4(b) fails. Exercise 16 maps [0, 1] one-to-one onto [0, 1] and is continuous at one point only. Derivatives have the intermediate value property without being continuous (The mean value theorem and its consequences).
  • Completeness is essential. On \{x \in \mathbb{Q} : 0 \le x \le 2\}, x^2 - 2 changes sign with no zero, and \displaystyle \frac{1}{x^2 - 2} is continuous and unbounded.
  • Beyond \mathbb{R}. Theorems 1 and 2 hold on compact sets of a metric space (Metric spaces, a first look); Theorem 3 needs a connected domain.

Where this goes next

Uniform continuity reruns the subsequence argument of Theorem 1 to show that a continuous function on a compact set is uniformly continuous. The mean value theorem and its consequences builds Rolle’s theorem on Theorem 2 and proves that derivatives have the intermediate value property. Which functions are integrable uses Theorem 2 on each piece of a partition to integrate continuous functions.

Exercises

Extremes on compact sets

  1. Let f(x) = x^4 - 8x^2 and \displaystyle K = [-3, -1] \cup \left[\frac{1}{2}, 1\right]. Explain why f attains a maximum and a minimum on K, find them, and find f(K).
  2. Decide whether each function attains a maximum and a minimum on the set given; find them, or the supremum or infimum not attained, and the image. (a) x^2 - 4x on [1, 5); (b) \dfrac{4x}{(x + 1)^2} on [0, \infty); (c) \dfrac{1}{x} on \{x : 1 \le \lvert x \rvert \le 2\}.
  3. Let K = \{0\} \cup [1, 4]. Find the points of K nearest to p = 0.6, to \displaystyle p = \frac{1}{2} and to p = 5, and every p with more than one nearest point in K.
  4. Let \displaystyle h(x) = x - \frac{1}{2} for 0 < x < 1 and h(0) = h(1) = 0. Find \sup h([0, 1]) and \inf h([0, 1]), show that neither is attained, and name the hypothesis of Theorem 2 that fails.

Intermediate values and images

  1. Find f(I): (a) x^2 - 2x on (-1, 2]; (b) \sin x + \cos x on [0, \pi]; (c) \cos x on (0, 2\pi).
  2. Let g(x) = (x - 1)(x - 3)(x - 4) on [0, 5] and A = \{x \in [0, 5] : g(x) < 0\}. Find \sup A. Run three steps of bisection from [0, 5] and say which zero it closes in on. Find \inf\{x \in [0, 5] : g(x) > 0\}.
  3. Omar’s second model of the palm house is \displaystyle T(t) = 68 - 7\cos\frac{\pi t}{12} + 2\sin\frac{\pi t}{4} on [0, 24]. Check that T(24) = T(0), and find every t \in [0, 16] with T(t) = T(t + 8) and the common temperature.
  4. Let f(x) = x^3 + 2x. Find f([-1, 2]) and f^{-1}(3). Following the proof of Theorem 6 with c' = 1.1 and c'' = 0.9, find \delta such that \lvert y - 3 \rvert < \delta gives \lvert f^{-1}(y) - 1 \rvert < 0.1, and show that no larger \delta works.

Proofs

  1. Prove that every continuous f : [a, b] \to [a, b] has a fixed point, a c with f(c) = c.
  2. Let f : [a, b] \to \mathbb{R} be continuous, and suppose that for every x there is y \in [a, b] with \displaystyle \lvert f(y) \rvert \le \frac{1}{2}\lvert f(x) \rvert. Prove that f has a zero.
  3. Let A and B be nonempty compact sets. Prove that there are a \in A and b \in B with \lvert a - b \rvert = \inf\{\lvert x - y \rvert : x \in A,\ y \in B\}.
  4. Let f : \mathbb{R} \to \mathbb{R} be continuous, with f(x_0) > 0 for some x_0, and suppose that for every \varepsilon > 0 there is R with \lvert f(x) \rvert < \varepsilon whenever \lvert x \rvert > R. Prove that f attains an absolute maximum.
  5. Prove that no continuous function maps [a, b] onto (0, 1), and none maps [a, b] onto \mathbb{R}.
  6. Let I be a nonempty interval and f : I \to \mathbb{R} continuous with every value rational. Prove that f is constant.
  7. Let T be continuous on [0, 24] with T(24) = T(0). Prove that T(t) = T(t + 8) for some t \in [0, 16].
  8. Let f(x) = x for rational x \in [0, 1] and f(x) = 1 - x for irrational x \in [0, 1]. Prove that f maps [0, 1] one-to-one onto [0, 1], is continuous at \displaystyle \frac{1}{2} and at no other point, and is not monotone.
Answers
  1. Maximum 9 at -3, minimum -16 at -2, f(K) = [-16, 9]. K is closed and bounded, so Theorem 2 applies. f'(x) = 4x(x - 2)(x + 2), so f decreases on [-3, -2], increases on [-2, -1] and decreases on \displaystyle \left[\frac{1}{2}, 1\right]. With f(-3) = 9, f(-2) = -16, f(-1) = -7, \displaystyle f\left(\frac{1}{2}\right) = -\frac{31}{16} and f(1) = -7, Theorem 5 gives f([-3, -1]) = [-16, 9] and \displaystyle f\left(\left[\frac{1}{2}, 1\right]\right) = \left[-7, -\frac{31}{16}\right], inside it.
  2. (a) Minimum -4 at 2, no maximum, supremum 5, image [-4, 5): x^2 - 4x = (x - 2)^2 - 4 increases on [2, 5) toward 5, and f(1) = -3. (b) Minimum 0 at 0, maximum 1 at 1, image [0, 1]: \displaystyle 1 - \frac{4x}{(x + 1)^2} = \frac{(x - 1)^2}{(x + 1)^2} \ge 0, and the values tend to 0 as x \to \infty. (c) The set [-2, -1] \cup [1, 2] is compact; maximum 1 at 1, minimum -1 at -1, image \displaystyle \left[-1, -\frac{1}{2}\right] \cup \left[\frac{1}{2}, 1\right], not an interval, as the domain is not one.
  3. Nearest to 0.6: 1. Nearest to \displaystyle \frac{1}{2}: both 0 and 1. Nearest to 5: 4. Only \displaystyle p = \frac{1}{2} has two: for p \le 0 the nearest point is 0, for p \in [1, 4] it is p, for p > 4 it is 4, and for 0 < p < 1 the distances p and 1 - p agree only at \displaystyle \frac{1}{2}.
  4. \displaystyle \sup = \frac{1}{2} and \displaystyle \inf = -\frac{1}{2}, neither attained: on (0, 1) the values fill \displaystyle \left(-\frac{1}{2}, \frac{1}{2}\right), and h(0) = h(1) = 0 lies inside. The domain is compact; continuity fails at 0 and 1.
  5. (a) [-1, 3): x^2 - 2x = (x - 1)^2 - 1 has minimum -1 at 1; for -1 < x \le 2, (x - 1)^2 < 4, and the values tend to 3 as x \to -1. (b) \displaystyle [-1, \sqrt{2}]: \displaystyle \sin x + \cos x = \sqrt{2}\sin\left(x + \frac{\pi}{4}\right) with \displaystyle x + \frac{\pi}{4} \in \left[\frac{\pi}{4}, \frac{5\pi}{4}\right], maximum at \displaystyle \frac{\pi}{4}, minimum -1 at \pi, and Theorem 5. (c) [-1, 1): minimum -1 at \pi, \cos x < 1 on (0, 2\pi), \displaystyle \cos\frac{1}{n} \to 1, and Theorem 4(b).
  6. \sup A = 4; bisection closes in on 1; the infimum is 1. g is negative on [0, 1) \cup (3, 4) and positive on (1, 3) \cup (4, 5]. Bisection: \displaystyle g\left(\frac{5}{2}\right) = \frac{9}{8} > 0 keeps \displaystyle \left[0, \frac{5}{2}\right]; \displaystyle g\left(\frac{5}{4}\right) = \frac{77}{64} > 0 keeps \displaystyle \left[0, \frac{5}{4}\right]; \displaystyle g\left(\frac{5}{8}\right) = -\frac{1539}{512} < 0 keeps \displaystyle \left[\frac{5}{8}, \frac{5}{4}\right], which contains only the zero 1.
  7. T(0) = 61 = T(24); t = 8, at 71.5 degrees. With \displaystyle u = \frac{\pi t}{12}, the sine term has period 8 and cancels, and \displaystyle T(t + 8) - T(t) = 7\left(\cos u - \cos\left(u + \frac{2\pi}{3}\right)\right), equal to \displaystyle 7\sqrt{3}\sin\left(u + \frac{\pi}{3}\right), which vanishes for t \in [0, 16] only at t = 8. T(8) = 68 + 3.5 = T(16).
  8. f([-1, 2]) = [-3, 12], f^{-1}(3) = 1, \delta = 0.471. f'(x) = 3x^2 + 2 > 0, so f is strictly increasing, with f(-1) = -3 and f(2) = 12. f(1.1) = 3.531 and f(0.9) = 2.529, so \delta = \min\{0.531, 0.471\}. As f^{-1} is increasing, \lvert f^{-1}(y) - 1 \rvert < 0.1 exactly when 2.529 < y < 3.531, so a larger \delta admits some y \le 2.529.
  9. Proof. Let g(x) = f(x) - x, continuous, with g(a) \ge 0 and g(b) \le 0. If either is 0, that end is a fixed point. Otherwise g(a) > 0 > g(b) and Theorem 3 gives c with g(c) = 0. \square
  10. Proof. \lvert f \rvert is continuous (Theorems 5 and 6 of Continuity, defined and proved), so by Theorem 2 it attains a minimum at some c. If f(c) \ne 0, some y has \displaystyle \lvert f(y) \rvert \le \frac{1}{2}\lvert f(c) \rvert < \lvert f(c) \rvert, against minimality. So f(c) = 0. \square
  11. Proof. The distances are bounded below by 0, so their infimum D exists. By the approximation property (Least upper bounds and completeness) choose a_n \in A, b_n \in B with \displaystyle D \le \lvert a_n - b_n \rvert < D + \frac{1}{n}. A subsequence a_{n_k} \to a \in A; a further subsequence of \{b_{n_k}\} converges to some b \in B, while the matching a terms still tend to a. Along it \lvert a_n - b_n \rvert tends to \lvert a - b \rvert, by the difference law and the continuity of \lvert \cdot \rvert, and to D, by the squeeze theorem. So \lvert a - b \rvert = D. \square
  12. Proof. Take \varepsilon = f(x_0) and R with \lvert f(x) \rvert < f(x_0) for \lvert x \rvert > R; then \lvert x_0 \rvert \le R. By Theorem 2, f attains a maximum on [-R, R] at some c, and f(c) \ge f(x_0) > f(x) for every \lvert x \rvert > R. So f(c) is the maximum on \mathbb{R}. \square
  13. Proof. By Theorem 5, the image of [a, b] is [m, M], which has a largest element and is bounded. (0, 1) has no largest element and \mathbb{R} is unbounded. \square
  14. Proof. f(I) is an interval by Theorem 4(b). If it contained y_1 < y_2, it would contain [y_1, y_2], which holds an irrational number (The Archimedean property and density of the rationals). So f(I) is a single point. \square
  15. Proof. Let g(t) = T(t + 8) - T(t), continuous on [0, 16]. The sum g(0) + g(8) + g(16) telescopes to T(24) - T(0) = 0. If some term is 0, done. Otherwise two of them have opposite signs, and Theorem 3 between those two times gives a zero of g. \square
  16. Proof. f maps the rationals of [0, 1] onto themselves and, as x \mapsto 1 - x keeps irrationals irrational, the irrationals onto themselves, one-to-one on each part. So f is a bijection of [0, 1]. At \displaystyle \frac{1}{2}, \displaystyle \left\lvert f(x) - \frac{1}{2} \right\rvert = \left\lvert x - \frac{1}{2} \right\rvert in both cases, so \delta = \varepsilon works. At \displaystyle c \ne \frac{1}{2}, rationals tending to c give values tending to c, irrationals give values tending to 1 - c \ne c, so by the sequential criterion f is discontinuous at c. Not monotone: \displaystyle f(0) < f\left(\frac{1}{2}\right), yet the irrational \displaystyle \frac{\sqrt{2}}{4} < \frac{1}{2} has \displaystyle f\left(\frac{\sqrt{2}}{4}\right) = 1 - \frac{\sqrt{2}}{4} > \frac{1}{2}. So Theorem 6 needs continuity. \square
math.wiki · CC-BY-SA-4.0