What a positive derivative says at a point and on an interval, Rolle’s theorem, and why the hypotheses are “continuous on the closed interval, differentiable on the open one”.
So far we have computed derivatives. Now we use them. The derivative is defined by a limit at a single point, yet we want to draw conclusions about a function on a whole interval: that it is increasing, that it is positive, that it has a turning point. §122 provides the bridge, and its key result is Rolle’s theorem: a differentiable function that takes equal values at two points has a horizontal tangent somewhere between them. The mean value theorem, the tests for maxima and minima, and Darboux’s theorem all follow from it.
The idea
Picture a smooth curve that starts and ends at the same height. Unless it is flat, it must go up and come back down (or down and back up), and at the highest (or lowest) point the tangent should be horizontal. That is the whole of Rolle’s theorem. The proof makes each part of the picture precise. A highest point exists because a continuous function on a closed interval attains its bounds (§103). At that point the derivative cannot be positive or negative, because then the curve would rise above the top on one side.
ExploreFind the horizontal tangent
Try this.The curve is sinx on [0,π], with equal end values. Drag the end points (keeping their heights equal is hard, so watch the read-out). Then use the second widget, ∣x∣ on [−1,1].
f(x)=sinx, on [0,3.14]
xy
y = f(x)
chord (drag its ends)
horizontal tangent at c
Rolle: if f is continuous on [a,b], differentiable in (a,b) and f(a)=f(b), then f′(c)=0 for some c in (a,b).
Chord slope b−af(b)−f(a)=3.898×10−17; Rolle looks for f′(c)=0.
One point with the required tangent: c=1.571.
f(x)=∣x∣, on [−1,1]
xy
y = f(x)
chord (drag its ends)
horizontal tangent at c
Rolle: if f is continuous on [a,b], differentiable in (a,b) and f(a)=f(b), then f′(c)=0 for some c in (a,b).
Chord slope b−af(b)−f(a)=0; Rolle looks for f′(c)=0.
Hypothesis fails: f is not differentiable at x = 0 (corner).
No point c in (a, b) works.
What to notice
For sinx the point c=21π appears whenever f(a)=f(b). If you make the end heights unequal, a point with a horizontal tangent may or may not exist: Rolle no longer promises one. For ∣x∣ the end values are equal and the function is continuous, yet no horizontal tangent exists. The lowest point is a corner, where there is no derivative at all.
Precise statements
Hardy uses (a,b) for both kinds of interval and says which he means. We write [a,b] for the closed interval a≤x≤b and (a,b) for the open interval a<x<b. The standing hypotheses of the next sections are:
ϕ is continuous on the closed interval [a,b] (at the ends, continuity from the inside, as in §99);
ϕ′(x) exists at every point of the open interval (a,b).
The mismatch is deliberate. Continuity at the end points is needed, as the quizzes below show. A derivative at the end points is not needed, and would not even make sense if we know nothing about ϕ outside [a,b].
TheoremA positive derivative at a pointHardy: Theorem A, §122
If ϕ′(x0)>0, then ϕ(x)<ϕ(x0) for all x<x0 sufficiently near x0, and ϕ(x)>ϕ(x0) for all x>x0 sufficiently near x0. (Similarly with the inequalities reversed if ϕ′(x0)<0.)
Hardy expresses this by saying that ϕ is increasing at the pointx0. This is weaker than “increasing in an interval containing x0”, as we shall see.
TheoremRolle’s theoremHardy: Theorem B, §122
If ϕ is continuous on [a,b], differentiable in (a,b), and ϕ(a)=ϕ(b), then there is a point ξ with a<ξ<b at which ϕ′(ξ)=0.
If ϕ is continuous on [a,b], differentiable in (a,b), and ϕ′(x)>0 for every x in (a,b), then ϕ is strictly increasing on [a,b]: ϕ(x1)<ϕ(x2) whenever a≤x1<x2≤b.
CorollaryPositive derivative and a non-negative startHardy: Theorem B, Cor. 2
If, in addition, ϕ(a)≥0, then ϕ(x)>0 for a<x≤b.
Proofs
ProofTheorem A
Strategy
A positive limit forces the quotient itself to be positive near the point. A positive quotient means that numerator and denominator have the same sign.
The quotient is eventually positive
{ϕ(x0+h)−ϕ(x0)}/h→ϕ′(x0)>0. Take ε=ϕ′(x0) in the definition of the limit: there is δ>0 such that for 0<∣h∣<δ the quotient lies within ϕ′(x0) of ϕ′(x0), and so is positive.
Read off the signs
For 0<h<δ the numerator has the sign of h, so ϕ(x0+h)>ϕ(x0). For −δ<h<0 it is negative, so ϕ(x0+h)<ϕ(x0).
∎
ProofRolle’s theorem
Strategy
Reduce to ϕ(a)=ϕ(b)=0. If ϕ is identically zero, every point works. Otherwise ϕ is somewhere positive (or somewhere negative). Take a point where the greatest (or least) value is attained, which exists by continuity on a closed interval. It is inside the interval, and Theorem A shows that the derivative there can be neither positive nor negative.
Reduce to end values 0
If ϕ(a)=ϕ(b)=k, consider ϕ(x)−k instead: it has the same derivative, and end values 0.
The trivial case
If ϕ(x)=0 throughout [a,b] then ϕ′(x)=0 at every point of (a,b), and there is nothing to prove.
Otherwise, a maximum inside
Suppose ϕ is positive somewhere (if it is only ever negative or zero, apply the same argument to −ϕ). Being continuous on the closed interval, ϕ has an upper bound M, and ϕ(ξ)=M for some ξ in [a,b] (§103, Theorem 2). Since M>0=ϕ(a)=ϕ(b), ξ is neither a nor b.
The derivative there is zero
If ϕ′(ξ)>0, Theorem A gives points just to the right of ξ with ϕ(x)>ϕ(ξ)=M; if ϕ′(ξ)<0, points just to the left. Either contradicts the definition of M. Since ϕ′(ξ) exists (because ξ is in the open interval), it must be 0.
∎
Hardy proves Cor. 1 directly from Rolle’s theorem. The argument is a nice example of combining the theorems of Chapter V with Rolle. (A shorter proof follows from the mean value theorem, in §126.)
ProofCor. 1: positive derivative ⇒ increasing
Strategy
Take a<x1<x2<b. Rule out ϕ(x1)=ϕ(x2) and ϕ(x1)>ϕ(x2): each would produce a point where ϕ′=0. Then extend to the end points by continuity.
Equality is impossible
If ϕ(x1)=ϕ(x2), Rolle’s theorem on [x1,x2] gives a point between them with ϕ′=0, contrary to ϕ′>0.
A decrease is impossible
Suppose ϕ(x1)>ϕ(x2). Since ϕ′(x1)>0, Theorem A gives a point x3 slightly to the right of x1 (and before x2) with ϕ(x3)>ϕ(x1)>ϕ(x2). By the intermediate value theorem (§101), ϕ takes the value ϕ(x1) at some x4 between x3 and x2. Then ϕ(x1)=ϕ(x4), and Rolle on [x1,x4] gives a zero of ϕ′: a contradiction again.
So ϕ(x1)<ϕ(x2) inside
We have shown ϕ(x1)<ϕ(x2) whenever a<x1<x2<b.
The end points
If a<x<x′<b then ϕ(x)<ϕ(x′), and letting x→a (continuity at a) gives ϕ(a)≤ϕ(x′). Choosing a point x′′ between a and x′, we get ϕ(a)≤ϕ(x′′)<ϕ(x′): strict. The end b is treated in the same way.
∎
Cor. 2 is immediate: ϕ(x)>ϕ(a)≥0 for a<x≤b.
Now compare Cor. 1, which assumes ϕ′>0at every point of an interval, with Theorem A, which assumes ϕ′(x0)>0at one point. Theorem A gives ϕ(x1)<ϕ(x0)<ϕ(x2) for x1<x0<x2 close to x0, but only when x1 and x2 are on opposite sides of x0. It says nothing about two points on the same side. In §125 Hardy gives an example showing that this gap is real.
ExploreIncreasing at a point, but on no interval
Try this.Here ϕ(x)=αx+x2sin(1/x), with ϕ′(0)=α. Keep α=0.5 and zoom in on 0 as far as the slider allows. Then raise α above 1 and zoom again.
ϕ(x)=0.5x+x2sinx1,ϕ′(0)=0.5
xφ
xφ′
y = φ(x)
tangent at 0, slope α
y = φ′(x)
where φ′ < 0 (φ decreasing)
Theorem A holds: since ϕ′(0)>0, the graph is below 0 just to the left of the origin and above it just to the right. But ϕ′(x)=α+2xsin(1/x)−cos(1/x) equals α−1=−0.5 at every x=1/(2kπ). The plot resolves 64 shaded intervals where φ decreases, 60 of them within 0.06 of 0. Zoom in as far as you like: they never stop, so φ is increasing on no interval around 0.
What to notice
With α=0.5, the graph crosses the origin going upward, as Theorem A promises. But the shaded intervals where ϕ′<0 (and ϕ decreases) never stop accumulating at 0, because ϕ′(x)=α−1<0 at every x=1/(2kπ). So no interval around 0, however small, is one on which ϕ increases. With α>1, ϕ′ stays positive near 0, and Cor. 1 applies.
Worked examples and exercises
Exercise●○○Hardy: Examples XLVI, 1
Verify Rolle’s theorem for ϕ(x)=(x−a)m(x−b)n on [a,b], where m, n are positive integers.
Hint
Factor ϕ′(x) as (x−a)m−1(x−b)n−1 times a linear polynomial.
Solution
ϕ(a)=ϕ(b)=0, and ϕ′(x)=(x−a)m−1(x−b)n−1{m(x−b)+n(x−a)}. The linear factor vanishes at ξ=(mb+na)/(m+n), a weighted average of a and b with positive weights, so a<ξ<b. Theorem B is verified. Note that ξ divides [a,b] in the ratio m:n, closer to the root of lower multiplicity.
Exercise●●○Hardy: Examples XLVI, 2
Show that x−sinx is increasing on the whole line, and that tanx−x increases on (−21π,21π). For which a is ax−sinx monotonic on the whole line?
Hint
The derivatives are 1−cosx and tan2x: positive except at isolated points. Apply Cor. 1 between consecutive zeros and join the pieces.
Solution
(x−sinx)′=1−cosx, which is positive except at the points 2kπ. On each interval [2kπ,2(k+1)π] the derivative is positive in the interior, so by Cor. 1 the function is strictly increasing on the closed interval. Consecutive intervals share their end points, so the function is strictly increasing on the whole line.
(tanx−x)′=sec2x−1=tan2x, positive on (−21π,21π) except at 0. Apply Cor. 1 on [c,0] and [0,d] for any −21π<c<0<d<21π, and join.
(ax−sinx)′=a−cosx. If a≥1 this is ≥0, vanishing at most at isolated points, so the function increases. If a≤−1 it decreases. If −1<a<1 the derivative takes both signs (it is a−1<0 at x=0 and a+1>0 at x=π), and by Theorem A the function rises near some points and falls near others, so it is not monotonic.
Exercise●●○Hardy: Examples XLVI, 3
Show that x/sinx increases steadily on (0,21π).
Hint
The derivative is (sinx−xcosx)/sin2x. Divide the numerator by cosx>0, and use the previous exercise.
Solution
(sinxx)′=sin2xsinx−xcosx=sin2xcosx(tanx−x). On (0,21π), cosx>0, and tanx−x>0 because tanx−x is strictly increasing and equals 0 at x=0 (Cor. 2). So the derivative is positive and, by Cor. 1, x/sinx is strictly increasing. Since x/sinx→1 as x→0, it increases from 1 to 21π as x goes from 0 to 21π.
Exercise●●○Hardy: Examples XLVI, 5
Show that, for x>0: sinx<x, then cosx>1−21x2, then sinx>x−61x3.
Hint
Each function in the chain vanishes at 0, and its derivative is (plus or minus) the previous function. Use Cor. 2 repeatedly.
Solution
Let f1(x)=x−sinx. Then f1(0)=0 and f1′(x)=1−cosx≥0, vanishing only at isolated points; by the previous exercise f1 is strictly increasing, so f1(x)>0 for x>0.
Let f2(x)=cosx−1+21x2. Then f2(0)=0 and f2′(x)=x−sinx=f1(x)>0 for x>0. By Cor. 2 on [0,b], for every b>0, f2(x)>0 for x>0.
Let f3(x)=sinx−x+61x3. Then f3(0)=0 and f3′(x)=cosx−1+21x2=f2(x)>0 for x>0, so f3(x)>0 for x>0.
Continuing in the same way gives Hardy’s general result: the partial sums of the series for cosx and sinx are alternately above and below the functions. These are the first error bounds for Taylor polynomials, which come in Chapter VII.
Suppose u, v, u′, v′ are continuous on an interval, and uv′−u′v never vanishes there. Show that between any two zeros of u lies a zero of v, and conversely. Check with u=cosx, v=sinx.
Hint
If v had no zero on [α,β], where α and β are consecutive zeros of u, then u/v would satisfy the hypotheses of Rolle’s theorem there.
Solution
Let α<β be zeros of u. First, v(α)=0: otherwise uv′−u′v=0 at α. Similarly v(β)=0. Suppose v has no zero in (α,β). Then v=0 on the closed interval [α,β], so u/v is continuous on [α,β] and differentiable inside, and it vanishes at α and β. By Rolle, (u/v)′=(u′v−uv′)/v2=0 at some point inside, so uv′−u′v=0 there, contrary to hypothesis. Hence v vanishes between α and β. The converse follows by exchanging u and v (the hypothesis is symmetric up to sign).
For u=cosx, v=sinx: uv′−u′v=cos2x+sin2x=1=0, and indeed the zeros 21π+kπ of cos and kπ of sin alternate.
Check yourself
Check yourselfTrue or false?
If ϕ′(x0)>0, then ϕ is increasing on some interval (x0−ε,x0+ε).
Why
False. The example αx+x2sin(1/x) with 0<α<1 has ϕ′(0)=α>0, but ϕ′ is negative at points arbitrarily close to 0. Theorem A only compares ϕ(x) with ϕ(x0), not two values on the same side. The statement becomes true if ϕ′ is continuous at x0, because then ϕ′>0 on a whole interval and Cor. 1 applies.
Check yourselfChoose one
Let ϕ(x)=x for 0≤x<1 and ϕ(1)=0. Then ϕ(0)=ϕ(1) and ϕ′(x)=1 throughout (0,1). Why does Rolle’s theorem not apply?
Why
ϕ is differentiable at every point of the open interval, with ϕ′=1=0, so no ξ exists. What fails is continuity at the end point 1. That is why condition (1) is imposed on the closed interval: without it the maximum of ϕ need not be attained, and the proof collapses.
Check yourselfEnter a number
Rolle’s theorem applies to ϕ(x)=x(x−1)2 on [0,1]. At which ξ in (0,1) is ϕ′(ξ)=0?
Why
ϕ′(x)=(x−1)2+2x(x−1)=(x−1)(3x−1), which vanishes at x=1 (an end point, not allowed) and at x=31. This agrees with the formula (mb+na)/(m+n) with m=1 for the root a=0 and n=2 for the root b=1: (1⋅1+2⋅0)/3=31.
Check yourselfTrue or false?
Rolle’s theorem cannot be applied to 1−x2 on [−1,1], because this function has no derivative at x=±1.
Why
False. Rolle needs differentiability only in the open interval (−1,1), where 1−x2 is differentiable, and continuity on the closed interval, which holds. The conclusion holds with ξ=0, the top of the semicircle. This is exactly why the hypotheses are split between the closed and open intervals.
Hardy’s language
It may seem odd to use the closed interval in one condition and the open one in the other, but we shall find the distinction important.
— G. H. Hardy, §122
Summary
Summary
Hypotheses throughout: continuous on the closed interval [a,b], differentiable on the open interval (a,b).
Theorem A: ϕ′(x0)>0 means ϕ is below ϕ(x0) just to the left and above it just to the right. It does not make ϕ increasing on an interval.
Rolle: equal end values ⇒ ϕ′(ξ)=0 for some interior ξ. Proof: an extreme value is attained inside, and there Theorem A forces ϕ′=0.
Cor. 1: ϕ′>0 on the whole open interval ⇒ ϕ strictly increasing on the closed interval. Cor. 2 turns this into inequalities such as sinx<x.
Each hypothesis matters: ∣x∣ (no derivative at one point) and a jump at an end point are counterexamples.