Academic Integrity: tutoring, explanations, and feedback — we don’t complete graded work or submit on a student’s behalf.

The proof of the Intermediate Value Theorem does not give a practical method for

ID: 2969256 • Letter: T

Question

The proof of the Intermediate Value Theorem does not give a practical method for finding the point c. This exercise discusses a technique known as the bisection method for approximating the number c. Let f be a continous function defined on the interval [a,b] and let v be any number between f(a) and f(b).

A) Explain why it is sufficient to consider the case in which f(a) and f(b) have opposite signs and v=0.

B) Let m be the midpoint of [a,b]. Unless f(m)=0, show that either f(a)f(m)<0 or f(m)f(b)<0. Thus, the length of the interval in which the point c lies has been shortened by a factor of 2.

C) By continuing the process started in part (B), show that there exists a nested sequence {[a,b]} of closed intervals such that f(an)f(bn)<0 for all n and

     (b_{n+1}-a_{n+1}=(b_{n}-a_{n})/2)     

for all n. This is where the term bisection comes into play; each interval is bisected and one half of it is used to form the next interval.

D) Let c be the unique point in all of the intervals [      (a_{n}.b_{n})    ] (see the Nested Intervals Theorem). Prove that f(c)=0.

E) Find a formula for      (b_{n}-a_{n})    in terms of a,b and n.

Explanation / Answer

(a) if f(a)*f(b) < 0 => f(a) > 0 & f(b) < 0 OR f(a) < 0 & f(b) > 0

without loss of generality let f(a) <0 and f(b) > 0 => f(a) < 0 < f(b) so by choosing our middle point v=0 and applying the Intermediate Mean Value Theorem, we have that=> there exists a c such that f(c) = v = 0


(b) Let f(a) < 0 < f(b); choose m as midpoint of [a,b]

now either f(m) > 0 or <0 (assuming f(m) != 0)

case 1: f(m) > 0 =>

but already we have f(a)<0 => f(a)*f(m) < 0

case 2: f(m) < 0 =>

but already we have f(b) > 0 => f(m)*f(b) < 0

in either case we have shorted the length from [a,b] to either [a,m] or [m,b] which is half the original length


(c) lets design the sequence in this way:

start by choose m = (a+b)/2 => either f(a)*f(m) < 0 or (fm)*f(b) < 0

if f(a)*f(m) < 0 => a1 =a ; b1=m

else if f(m)*f(b) < 0 => a1 = m; b1 = b

^ from part (b) we have b1 - a1 = (b-a)/2

now consider the new interval as [a1,b1]

next choose m2 = (a1+b1)/2 => repeat the same steps as above and get a2,b2

in the same way we have b2-a2 = (b1-a1)/2 ; since [a1,b1] was our new interval and using part (b)

continue this for n steps we have =>

b(n+1) - a(n+1) = (b(n) - a(n))/2


(d) length([a1,b1]) = (b-a)/2

length([a2,b2]) = (b-a)/4 .. and so on

=> length([a(n),b(n)]) = (b-a)/(2^n)

choose n very large,

by continuity, choose epsilon -> 0 => f(an) - f(bn) -> 0 , then delta = (b-a)/(2^n) => f(an) = f(bn)

by f(an)f(bn) <0 =>f(an) = f(bn) = 0 = f(c); since c lies in [an,bn]


(e) as derived in part (d) =>

(bn - an) = (b-a)/(2^n)