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 19

Areas

⏟ Δx ℓ r f(ℓ) has areaf(ℓ)Δx ⏟ Δx ℓ r f(ℓ) f(r) has area f(ℓ)Δx+ 12Δx (f(r)-f(ℓ)) = Δx2 ( 2f(ℓ)+ f(r)-f(ℓ) ) = Δx2 ( f(ℓ)+ f(r) ) ⏟ ⏟ Δx Δx ℓ m r f(ℓ) f(m) f(r) has areaΔx3 ( f(ℓ)+4f(m) +f(r) )

Riemann Integral

∫abf(x)dx = limΔx→0 (sum of the areas of the little rectangles) = limΔx→0 Δx ( f(a)+ f(a+Δx)+⋯+ f(b-Δx) ) a b

Trapezoidal integral

∫abf(x)dx = limΔx→0 (sum of the areas of the little trapezoids) = limΔx→0 Δx2 ( f(a)+f(a+Δx) +f(a+Δx)+ f(a+2Δx) +f(a+2Δx) +f(a+3Δx) +⋯+f(b-2Δx) +f(b-Δx) +f(b-Δx)+f(b) ) = limΔx→0 Δx2 ( f(a)+2 f(a+Δx) +2f(a+2Δx) +⋯+2f(b-Δx) +f(b) )

Simpson's Integral

⏟ ⏟ Δx Δx ℓ m r f(ℓ) f(m) f(r) has areaΔx3 ( f(ℓ)+4 (m)+f(r) ) so that ∫abf(x)dx = limΔx→0 (sum of the areas of the little camel humps) = limΔx→0 Δx3 ( f(a)+4f (a+Δx) +f(a+2Δx) +f(a+2Δx) +4f(a+3Δx) +f(a+4Δx) +⋯ +f(b-2Δx) +4f(b-Δx) +f(b) ) = limΔx→0 Δx3 ( f(a)+ 4f(a+Δx)+ 2f(a+2Δx)+ 4f(a+3Δx)+ 2f(a+4Δx)+ ⋯+f(b) ) .

Let N∈ℤ>0 and f:[a,b]→ℝ be a function and M∈ℝ>0.

(a) Assume that f(N+1):[a,b]→ℝ exists and |f(N+1)(c)|<M for c∈[a,b]. Let TaylorErr(N)=f(b)- ( f(a)+ f′(a)(b-a)+ 12!f″(a)(b-a)2 +⋯+1N!f(N) (a)(b-a) ) . Then |TaylorErr(N)|< 1(N+1)! (b-a)NM.
(b) Assume that f″:[a,b]→ℝ exists and |f″(c)|<M for c∈[a,b]. Let Δx=b-aN and TrapErr(N)=∫ab f(x)dx- (b-aN)12 ( f(a)+ 2f(a+Δx) +2f(a+2Δx) +⋯ …+2f(b-Δx) +f(b) ) . Then |TrapErr(N)|< (b-a)312N2 ·M.
(c) Assume that f(4):[a,b]→ℝ exists and |f(4)(c)|<M for c∈[a,b]. Let Δx=b-aN and SimpErr(N)=∫ab f(x)dx- (b-a)3N ( f(a)+4 f(a+Δx)+ 2f(a+2Δx) +4f(a+3Δx) +2f(a+4Δx) +⋯ ⋯+2f(b-2Δx) +4f(b-Δx)+ f(b) ) . Then |SimpErr(N)|< (b-a)5180N4·M.

Find log(2) to within .01. log(1+x)=x- x22+x33- x44+⋯+ xNN+ Error(N). In fact, if f(x)=log(1+x) then f′(x)= 11+x, f″(x)= -1(1+x)2, f(3)(x)= (-1)(-2)(1+x)3, f(4)(x)= (-1)(-2)(-3)(1+x)4, …,f(N+1) (x)= (-1)(-2)⋯(-N) (1+x)N+1 . So TaylorError(N)= (-1)(-2)⋯(-N) (1+c)N+1 ·1(N+1)! ·(2-1e)N+1 withc∈(0,1). So |TayErr(N)|= 1(N+1)(1+c)N+1 <1N+1. So, to get an approximation within 1100=.01 we should let N=99.

So log(2)≈1-12+1/3+⋯+197-198+199, to within .01.

Find log 2 to within .01. log 2=∫121xdx Use a trapezoidal approximation with f(x)=1x, a=1andb=2. How many slices (i.e. what should N be)? 1 2 x y y=1x Then f′(x)=-1x2 and f″(x)=(-1)(-2)𝓍3=2x3.

So |f″(c)|<2 for c∈[1,2].

So |TrapError(N)|< (b-a)312N2 ·2=16·N2. If N=5 then 16·N2=16·25=1150<.01.

So 5 slices will do. log(2) = ∫121xdx ≈ (b-a)512 ( f(a)+2f(a+Δx) +2f(a+2Δx) +2f(a+3Δx)+ 2f(a+4Δx)+ f(b) ) = 110 ( 11+2 11+15+2 11+25+2 11+35+2 11+45+12 ) = 110 ( 1+2·56+2·57+2 ·58+2·59+12 ) = 110+16+17+ 18+19+120.

Notes and References

These are notes from a 2010 course on Real Analysis 620-295. This page comes from 100421Lect19.pdf and was given on 21 April 2010.

page history