This chapter organizes what one needs to know about functions before calculus: the small family of reference functions whose graphs must be known by heart, the precise meaning of increasing and decreasing, and how variations behave when functions are combined.
11.1 Functions and curves
Definition 11.1(Function, domain, curve)
A functionf assigns to each number x of a set D⊆R (its domain) exactly one number f(x). Its curve (or graph) is the set of points (x,f(x)) for x∈D.
Example 11.2
The formula f(x)=x−1x+2 requires x+2≥0 (for the square root) and x=1 (nonzero denominator): the domain is [−2,1)∪(1,+∞).
11.2 The reference functions
The following five functions and their curves must become reflexes.
Definition 11.3(Reference functions)
x↦x2 (square),x↦x3 (cube),x↦x,x↦x1,x↦∣x∣.
Their domains are R, R, [0,+∞), R∖{0} and R respectively. The curve of x↦x1 is a hyperbola.
The five reference curves: the parabolax2 and the cube x3 (left); x, ∣x∣ and the hyperbolax1 (right).
11.3 Variations on an interval
Definition 11.4(Increasing, decreasing)
Let f be defined on an intervalI. The functionf is increasing on I if it preserves order:
for all u,v∈I:u<v⟹f(u)≤f(v),
and decreasing if it reverses order (u<v⟹f(u)≥f(v)). With strict inequalities on the right, f is strictly increasing (resp. decreasing). A function that is increasing or decreasing on I is monotonic on I.
Method 11.5(Proving a variation from the definition)
Take two arbitrary numbers u<v in the interval, and determine the sign of the difference f(v)−f(u), usually by factoring it. If the sign is constant on the interval, conclude.
Proposition 11.6(Variations of the reference functions)
1.v2−u2=(v−u)(v+u). On [0,+∞): v−u>0 and v+u>0 (as v>u≥0), so v2>u2. On (−∞,0]: v+u<0, so v2<u2.
2.v3−u3=(v−u)(v2+uv+u2), and the second factor equals (v+2u)2+43u2>0 unless u=v=0. Hence v3>u3 whenever v>u.
3. For 0≤u<v: v−u=v+u(v−u)(v+u)=v+uv−u>0.
4. For 0<u<v: v1−u1=uvu−v<0 since u−v<0 and uv>0; same computation for u<v<0. On the union the claim fails: −1<1 but −11=−1<1=11.
5. On [0,+∞), ∣x∣=x is strictly increasing; on (−∞,0], ∣x∣=−x is strictly decreasing. ∎
Remark 11.7
Point 4 is a classic trap: “decreasing on (−∞,0) and on (0,+∞)” is not the same as “decreasing on R∖{0}”. Variations only make sense on intervals.
Example 11.8(Comparing without computing)
Since x is strictly increasing, 7<8; since x2 is strictly decreasing on (−∞,0], (−3)2>(−2)2. Monotonicity turns comparisons of images into comparisons of arguments.
Transformations in action: shifting x2 up by 1 (red) keeps its variations; taking the square root (orange) keeps them again, as in Example 11.10.
11.5 Even and odd functions
Definition 11.11(Parity)
Let f be defined on a domainD symmetric about 0 (i.e. −x∈D whenever x∈D). The functionf is even if f(−x)=f(x) for all x∈D, and odd if f(−x)=−f(x) for all x∈D.
Proposition 11.12(Geometric meaning)
The curve of an even function is symmetric about the y-axis; the curve of an odd function is symmetric about the origin.
Proof. If f is even and (x,y) is on the curve (so y=f(x)), then f(−x)=f(x)=y: the mirror point (−x,y) is also on the curve. If f is odd, f(−x)=−y: the point (−x,−y), symmetric of (x,y) about the origin, is on the curve. ∎
Example 11.13
x2 and ∣x∣ are even; x3 and x1 are odd; f(x)=x2+x is neither: f(−1)=0 while f(1)=2, so f(−1)=±f(1). Parity halves the work: an even or odd function only needs to be studied on [0,+∞).
A functionf has the following variation behaviour: increasing on [−3,1] from f(−3)=−2 to f(1)=4, then decreasing on [1,5] down to f(5)=0. How many solutions does the equationf(x)=1 have on [−3,5]? And f(x)=5?
Solution
Solution of Exercise 11.4.
Read the variation table. On [−3,1], f climbs from −2 to 4 and crosses the value 1 exactly once; on [1,5] it goes down from 4 to 0 and crosses 1 once more: the equationf(x)=1 has 2 solutions. The maximum of f is f(1)=4<5, so f(x)=5 has no solution.
Exercise 11.5★★
Using Method 11.5 (sign of f(v)−f(u)), prove that f(x)=x2−4x is strictly increasing on [2,+∞).
2.g=u1 with u(x)=(x−2)2+1=f(x)+4. The shift by 4 keeps the variations of f, and u≥1>0. Taking the reciprocal reverses them: g is increasing on (−∞,2] and decreasing on [2,+∞), with maximumg(2)=11=1.
Exercise 11.8★★
The curve of a functionf passes through the points (0,1), (2,3) and (4,1), and f is increasing on [0,2] and decreasing on [2,4]. Sketch a possible curve, then sketch the curves of f+2, of −f, and of 2f.
Solution
Solution of Exercise 11.8.
Any hill-shaped curve through the three points works. Then: f+2 passes through (0,3), (2,5), (4,3), same variations (vertical shift); −f passes through (0,−1), (2,−3), (4,−1) with reversed variations (the peak becomes a valley); 2f passes through (0,2), (2,6), (4,2), same variations but amplified vertically.
Exercise 11.9★★
Show that the functionf(x)=x+12x+1 can be written f(x)=2−x+11, and deduce its variations on (−1,+∞).
Solution
Solution of Exercise 11.9.
2−x+11=x+12(x+1)−1=x+12x+1=f(x). On (−1,+∞), u(x)=x+1 is increasing and positive, so u1 is decreasing, −u1 is increasing, and adding 2 changes nothing: f is strictly increasing on (−1,+∞).
1. True. If u<v, then (f+g)(v)−(f+g)(u)=(f(v)−f(u))+(g(v)−g(u)) is a sum of two nonnegative numbers.
2. False.f(x)=g(x)=x are increasing on R, but their product x2 is not increasing on R (it decreases on (−∞,0]). (The statement becomes true if both functions are also nonnegative.)
3. False. Same counterexample: f(x)=x is increasing on R but f2(x)=x2 is not.
Exercise 11.11★★★
Let f(x)=x+x1 for x>0.
Show that for 0<u<v, f(v)−f(u)=(v−u)uvuv−1.
Deduce that f is strictly decreasing on (0,1] and strictly increasing on [1,+∞).
Conclude that x+x1≥2 for all x>0, with equality only at x=1.
2. Always v−u>0 and uv>0. If 0<u<v≤1, then uv<1, so uv−1<0 and f(v)<f(u): f is strictly decreasing on (0,1]. If 1≤u<v, then uv>1 and f(v)>f(u): f is strictly increasing on [1,+∞).
3.f decreases before 1 and increases after, so f(x)≥f(1)=2 for all x>0, with equality exactly at x=1. (This inequality reappears in many guises; compare Example 10.3 and the tangent-line inequalities of Chapter 22.)
11.7 Problem: The symmetry factory
Problem 11.1
Weekend problem — every function is an even part plus an odd part: parity rules, the transformation game, and minima won without calculus
Some functions are mirror-symmetric, some are half-turn symmetric, most are neither — yet a small algebraic miracle says that everyfunction is, in exactly one way, the sum of a mirror-symmetric piece and a half-turn-symmetric piece. This problem proves the miracle, plays the transformation game on the reference curves (Proposition 11.9), and closes with the difference-factoring technique of Exercise 11.11, which wins minimization problems a full chapter before the derivative arrives.
State the geometric meaning of each verdict (Proposition 11.12): which symmetry does the curve of an even function have? Of an odd one?
Parity multiplies like signs: prove that the product of two odd functions is even, and complete the whole multiplication table (even×even, even×odd, odd×odd).
Quick consequences: if f is odd and defined at 0, what is f(0)? If f is even with f(3)=7, what is f(−3)? And show that the only function both even and odd is the zero function.
How much of an even function’s curve determines all of it? Give a function that is neither even nor odd, and explain what fails.
Part II — The decomposition theorem. For a functionf defined on a domain symmetric about 0, set
E(x)=2f(x)+f(−x),O(x)=2f(x)−f(−x).
Compute E and O for f(x)=x3+x2+1, and check: Eeven, Oodd, E+O=f.
Prove the general statement: for any f, the functionE is even, O is odd, and f=E+O — every function splits into an even part and an odd part.
Prove the splitting is unique: if f=E1+O1=E2+O2 with Eieven and Oiodd, use question 4 on E1−E2.
Decompose f(x)=1−x1 (for x=±1): put E and O over the common denominator 1−x2 and simplify.
A look ahead, in one sentence each: the even and odd parts of the exponential function will be the twin functions of grade 12 (hyperbolic cosine and sine); and the same instinct — split an object into symmetric pieces and treat each with its own symmetry — powers the university volumes’ analysis of sound and signals. Why is such a split useful, in general?
Part III — The transformation game.
Starting from the curve of x↦x2, describe precisely the curves of (x−3)2, x2+2, (x−3)2+2, −x2, 2x2 (Proposition 11.9).
Reduce f(x)=∣x−2∣+∣x+2∣ to a piecewise formula on the three intervals cut by −2 and 2, and describe its curve (it has a famous flat bottom).
Which transformations respect evenness? Decide, with proof or counterexample: horizontal shift f(x−3); vertical shift f(x)+2; vertical stretch 2f(x) (all applied to an evenf).
Solve ∣x−2∣+∣x+2∣=6 using the piecewise formula of question 12.
Mirror Exercise 11.11 for g(x)=x+x4 on (0,+∞): show g(v)−g(u)=(v−u)uvuv−4, locate the minimum, and give its value.
Apply: among all rectangles of area 16, which has the least perimeter? (Write P=2(x+x16) and use question 16’s machine.) An old friend concludes.
Determine the variations of h(x)=x2+1 on all of R, combining a composition argument on [0,+∞) with a parity argument for the rest.
The closest point: find the point of the curve y=x nearest to (3,0). (Minimize the squared distance — a quadratic in x — and explain why minimizing the square is legitimate.)
Finale, the factory tour: reference curves are the raw stock; transformations the machines; parity and the decomposition the quality control; difference-factoring the measuring bench. One sentence on each — and name the power tool the next chapter delivers.
Solution
Solution of Problem 11.1.
1.x4−3x2: even ((−x)4−3(−x)2=x4−3x2). x3−x: odd. x2+x: neither (f(−1)=0=f(1)=2 and =−f(1)). ∣x∣: even. x1: odd.
2.Even: the curve is symmetric across the y-axis (mirror). Odd: symmetric about the origin (half-turn).
4.Odd at 0: f(0)=−f(0), so f(0)=0. Even: f(−3)=f(3)=7. Both even and odd: f(x)=f(−x)=−f(x) for all x, so 2f(x)=0: f is the zero function.
5. The right half (x≥0) determines an even function completely: the left half is its mirror. Neither even nor odd: x2+x (question 1) — its two symmetries fail at x=1.
6.E(x)=2(x3+x2+1)+(−x3+x2+1)=x2+1 (even), O(x)=2(x3+x2+1)−(−x3+x2+1)=x3 (odd); their sum is f.
7.E(−x)=2f(−x)+f(x)=E(x): even. O(−x)=2f(−x)−f(x)=−O(x): odd. And E(x)+O(x)=f(x) by direct addition: every function splits.
8. From E1+O1=E2+O2: E1−E2=O2−O1. The left side is even, the right side odd — so this function is both, hence zero (question 4): E1=E2 and O1=O2. One split only.
9.E(x)=21(1−x1+1+x1)=21⋅1−x2(1+x)+(1−x)=1−x21, and O(x)=21⋅1−x2(1+x)−(1−x)=1−x2x. Check: E is even, O is odd, and their sum is 1−x21+x=(1−x)(1+x)1+x=1−x1=f(x).
10. Because each piece obeys a symmetry, half the work (and half the data) suffices for each: solve on one side, mirror to the other — and many operations (integration, series, signal analysis) treat symmetric pieces far more simply than raw functions.
12. For x≥2: (x−2)+(x+2)=2x. For −2≤x≤2: (2−x)+(x+2)=4. For x≤−2: −2x. The curve descends with slope−2, runs flat at height 4 between −2 and 2, then climbs with slope2: a valley with a flat floor.
13. Vertical shift: f(x)+2 stays even (f(−x)+2=f(x)+2). Vertical stretch: 2f(x) stays even. Horizontal shift: (x−3)2 is not even (f(−1)=16=4=f(1)): sliding sideways destroys the mirror.
14. On the flat floor the value is 4=6. For x≥2: 2x=6, x=3. For x≤−2: −2x=6, x=−3. Solutions: ±3.
15.f(x)=a(x−2)2−3 with f(0)=4a−3=1: a=1, so f(x)=(x−2)2−3.
16.g(v)−g(u)=(v−u)+4uvu−v=(v−u)uvuv−4: negative for u<v with uv<4, positive with uv>4: decreasing on (0,2], increasing on [2,+∞); minimum at x=2, of value g(2)=4.
17. Sides x and x16: P(x)=2(x+x16), minimized (question 16’s machine with 4→16, turning point 16=4) at x=4: the 4×4square, perimeter 16 — Dido’s answer, now proved for every real side length.
18. On [0,+∞): x↦x2+1 increases, and x increases, so the composite increases. And h(−x)=x2+1=h(x): h is even, so by mirror symmetry it decreases on (−∞,0]. Minimumh(0)=1.
19. Squared distance to (3,0) of the point (x,x): d2(x)=(x−3)2+x=x2−5x+9, minimal at x=25: closest point (25,25), distance 411=211. Minimizing d2 is legitimate because squaring is strictly increasing on positive numbers: d and d2 bottom out at the same spot.
20. Raw stock: the reference curves, known by heart. Machines: shifts, flips and stretches assembling new graphs from old. Quality control: parity and the even–odd split, which halve every study. Measuring bench: factoringf(v)−f(u), which reads variations and minima directly. The next chapter’s power tool: the derivative, which reads them at a glance.