MATH 409 Lecture 15

From Notes
Jump to navigation Jump to search

« previous | Tuesday, October 22, 2013 | next »

Lecture Slides

Review

Derivative is defined as f′(a)=limh→0f(a+h)−f(a)h or equivalently f′(a)=limx→af(x)−f(a)x−a.

Notations:

  • Lagrange: f′
  • Newton: f˙
  • Leibniz: dfdx
  • Euler: Dxf
  • f(1) (higher order derivatives)

Derivative at a point denoted (f(x))′|x=a.

Differentiability Theorems

  • Sum and Difference Rules
  • Product Rule
  • Quotient Rule

Derivatives of Elementary Functions

Higher-Order Derivatives

Defined inductively:

For n≥2 for any a∈ℝ, the n-th derivative of f at a point a, denoted f(n)(a), is defined by f(n)(a)=(f(n−1))′(a).

Derivative Spaces

Let I be an interval of the real line ℝ. We denote C(I) or C0(I) as the set of all continuous functions on I.

For any n∈ℕ, we denote Cn(I) as the set of all functions that are n times continuously differentiable on I. (that is, the n-th derivative is continuous)

C∞(I) represents the set of all infinitely continuously differentiable on I.

Examples

f(0)=0, and f(x)=xsin⁡1x for x≠0.

Product and Chain rules show that f is differentiable on ℝ∖{0}.

For x≠0.

f′(x)=(xsin⁡1x)=sin⁡1x+x(sin⁡1x)=sin⁡1x−1xcos⁡1x

The function f is continuous at 0, but the derivative is not continuous at 0, so f is not differentiable at 0. Indeed f(h)−f(0)h=sin⁡1h has no limit as h→0.

g(0)=0, and g(x)=x2sin⁡1x for x≠0.

As above, the Product and Chain rules show that f is differentiable on ℝ∖{0}.

For x≠0,

g′(x)=(x⋅xsin⁡1x)=2xsin⁡1x−cos⁡1x

The function is differentiable at 0. Indeed g(h)−g(0)h=hsin⁡1h→0 as h→0.

Note that g is not continuously differentiable on ℝ since g′ is not continuous at 0. Namely, limx→0g′(x) does not exist.


Power Rule

Theorem. (xn)′=nxn−1 for all x∈ℝ and n∈ℕ.

Proof by induction. In the case n=1, we have x′=1=1x0 for all x∈ℝ.

Assume that (xn)′=nxn−1 for some n∈ℕ and all x∈ℝ. Using the product rule, we obtain (xn+1)=(xnx)=(xn)′x+xnx′=nxn−1x+xn=(n+1)xn.

quod erat demonstrandum

Note: The theorem can also be proved using the formula xn−anx−a=xn−1+xn−2a+…+xan−2+an−1.

In the same breath, (x−n)′=−nx−n−1 for all x≠0 and n∈ℕ.

Using reciprocal rule, we obtain (x−n)′=(1xn)=−(xn)′(xn)2=−nxn−1x2n=−nx−n−1


Derivative of Inverse Function

Theorem. Suppose f is an invertible continuous function. If f is differentiable at a point a and f′(a)≠0, then the inverse function is differentiable at the point b=f(a) and

(f−1)′(b)=1f′(a)=1f′(f−1(b))

Proof. Since f is differentiable at a, we know that f is defined on an open interval I=(c,d) containing a. Since f is continuous and invertible, it follows from the Intermediate Value Theorem that f is strictly monotone on I, the image f(I) is an open interval containing b, and the inverse function f−1 is continuous and strictly monotone on f(I).

We have limx→af(x)−f(a)x−a=f′(a). Since f′(a)≠0, it follows that limx→ax−af(x)−f(a)=1f′(a). Since f−1 is continuous and monotone on the interval f(I), we obtain that f−1(y)→a and f−1(y)≠a when y→b and y≠b.

Therefore limy→bf−1(y)−ay−b=limy→bf−1(y)−af(f−1(y))−b=limx→ax−af(x)−f(a)=1f′(a).

quod erat demonstrandum

Remark. In the case f′(a)=0, the inverse function f−1 is not differentiable at f(a).

Indeed, if f−1 is differentiable at b=f(a), the chain rule implies that (f−1∘f)(a)=(f−1)′(b)⋅f′(a). Obviously the LHS is the identity function, so (f−1∘f)′(a)=1≠0, so that f′(a)≠0.

Example

f(x)=arccos⁡x, x∈[−1,1].

The function g(y)=cos⁡y is strictly decreasing on [0,π] and maps this interval onto [−1,1]. By definition, the function f(x)=arccos⁡x is the inverse of the restriction of g to [0,π]. Notice that g′(0)=g′(π)=0 and g′(y)≠0 for y∈(0,π). It follows that the function f is continuous on (−1,1) and not differentiable at 1 and −1. Moreover, for any x∈(−1,1),

f′(x)=1g′(f(x))=−1sin⁡arccos⁡x

Let y=arccos⁡x (hence x=cos⁡y). We have sin2y=cos2y=1 by the pythagorean identity. Besides, sin⁡y>0 since y∈(0,π). Consequently, sin⁡y=1−cos2y=1−x2. Thus f′(x)=−11−x2.

quod erat demonstrandum

Homework hint; use similar method to prove for arcsin⁡x and arctan⁡x, just use same identity (divide by cos2y for arctan⁡x)

Exponential and Logarithmic Functions

Theorem. The sequence xn=(1+1n)n for n∈ℕ is increasing and bounded, hence convergent.

The limit is the number e=2.718281828… (number of letters in each word in "I'm forming a mnemonic to remember a constant in analysis")

Corollary. limx→0(1+x)1x=e.

Not proved here... (Ain't nobody got time for that!)

for any a>0 and a≠1, the exponential function f(x)=ax is strictly monotone and continuous on ℝ. It maps ℝ onto (0,∞). Therefore the inverse function g(y)=logay is strictly monotone and continuous on (0,∞). The natural logarithm logey is also denoted just log⁡y.

Since (1+h)1h→e as h→0, it follows that h−1log⁡(1+h)=log⁡(1+h)1h→log⁡e=1 as h→0. In other words, (log⁡y)|y=1=1

Examples

f(x)=ex for x∈ℝ.

f(x+h)−f(x)h=ex(eh−1)h for all x,h∈ℝ. Therefore for any x∈ℝ, f′(x)=limh→0eh−1h=exf′(0)=ex

f(x)=ax for x∈ℝ, where a>0.

Equivalently, f(x)=elog⁡ax=exlog⁡a. So f′(x)=exlog⁡alog⁡a=axlog⁡a.

f(x)=log⁡x for x∈(0,∞).

Since f is the inverse function g(y)=ey, we obtain f′(x)=1g′(log⁡x)=1elog⁡x=1x for all x>0.

Power Rule: General Case

(xα)=αxα−1 for all x>0 and α∈ℝ.

Proof. Let us fix a number α∈ℝ and consider f(x)=xα for x∈(0,∞). For any x>0, we obtain f(x)=elog⁡xα=alog⁡x, where a=eα. Hence f=h∘g, where g(x)=log⁡x for x>0 and h(y)=ay for y∈ℝ. By the chain rule, f′(x)=αxα−1.

quod erat demonstrandum