Real Analysis

Arun Ram
Department of Mathematics and Statistics
University of Melbourne
Parkville, VIC 3010 Australia
aram@unimelb.edu.au

Last update: 13 July 2014

Lecture 18

Derivatives by limits

Let f:[a,b]→ℝ. Let c∈[a,b].

The derivative of f at x=c is f′(c)= limx→c f(x)-f(a)x-c. Alternatively, f′(c)= limΔx→0 f(c+Δx)-f(x) Δx .

Let f:[a,b]→ℝ and g:[a,b]→ℝ and let β,γ∈ℝ. Assume that f′(c) and g′(c) exist. Then

(a) (βf+γg)′(c)= βf′(c)+ γg′(c),
(b) (fg)′(c)= f′(c)g(c)+ f(c)g′(c),
(c) Assume f:[a,b]→ℝ is given by f(x)=x. then f′(c)=1.
(d) If f′(c) exists then f is continuous at x=c.

Define f″(c)= (f′)′(c), f(3)(c)= (f″)′(c)and f(N)(c)= (f(N-1))′ (c).

(Taylor's theorem with Lagrange's remainder). If f:[a,b]→ℝ and N∈ℤ≥0 and f(N):[a,b]→ℝ is continuous and f(N+1):[a,b]→ℝ exists then there exists c∈(a,b) such that f(b)=f(a)+f′ (a)(b-a)+ 12!f″(a) (b-a)2+⋯+ 1N′f(N) (a)(b-a)N+ 1(N+1)! f(N+1)(c) (b-a)N+1.

Remarks:

(1) The last term in f(b)=f(a)+⋯+1(N+1)!f(N+1)(c)(b-a)N+1 is Lagrange's form of the remainder.
(2) The special case N=0 is the Mean Value Theorem. If f:[a,b]→ℝ and f*:[a,b]→ℝ is continuous and f′:[a,b]→ℝ exists then there exists c∈(a,b) such that f(b)=f(a)+f′ (c)(b-a) i.e. f′(c)= f(b)-f(a)b-a a b c x y
(3) The special case N=0 and f(a)=f(b) is Rolle's theorem If f:[a,b]→ℝ is continuous and f′:[a,b]→ℝ exists then there exists c∈(a,b) such that f′(c)=0 a b x f ( a ) = f ( b ) y
(4) The proof of these theorems uses Intermediate value theorem
(a) If f:[a,b]→ℝ is continuous and w is between f(a) and f(b) then there exists c∈(a,b) such that f(c)=w.
(b) If f:[a,b]→ℝ is continuous then there exist m,M∈ℝ such that f([a,b])=[m,M] a b f ( a ) f ( b ) m M
(5) Let f:[0,2π]→ℂ be given by f(x)=cos x+ isin x. Then f(0)=f(2π) but f′(x) is never 0. Why is not a contradiction to Rolle's theorem?
(6) The first N terms in (all but the remainder term) f(b)=f(a)+⋯+ 1N!f(N)(c) (b-a)N+ 1(N+1)!f′ (c)(b-a)N+1 are the Taylor approximation to f at x=a of order N.
(7) f:[a,b]→ℝ is differentiable at x=c if f′(c) exists.

Approximate 281/3 to 5 decimal places.

Let f(x)=(27+x)1/3. Then (27+x)1/3= a0+a1x+a2 x2+a3x3+⋯ So a0 = (27+0)1/3=3, a1 = (ddx(27+x)1/3) |x=0=1/3 (27+x)-23 |x=0=1/3 132=127, a2 = 12! (d2dx2(27+x)1/3) |x=0=121/3 (-23) (27+x)-53 |x=0= 12·(-2)32 135=-137, a3 = 13! (d3dx3(27+x)1/3) |x=0=12·31/3 (-2)3 (-5)3 (27+x)-83 |x=0= 534·138= 5312, 14!f(4) (c)(28-27)4 = 14!1/3 (-2)3 (-5)3 (-8)3 (27+c)-113 = -24·523·35 (27+c)-113= -2·535 1(27+c)113. So (27+x)1/3=3+127x-137x2+5312x3+⋯ and 281/3 = (27+1)1/3 ≈ 3+127-137+ 5312 with error equal to 2·535 1(27+c)113 for somec∈(0,1). So the error is less than 2·535 127113= 2·535·311 =2·5316.

Notes and References

These are notes from a 2010 course on Real Analysis 620-295. This page comes from 100419Lect18.pdf and was given on 19 April 2010.

page history