If x1=(x0+2)/(x0+1), can we show 2 lies between x0 and x1?
Question
If
x0
is any positive rational number and
x1=x0+1x0+2,x2=x1+1x1+2,x3=x2+1x2+2,
prove that
- 2 lies between x0 and x1, and between x2 and x3,
- ∣x12−2∣<∣x02−2∣,
- ∣x22−2∣<91∣x02−2∣.
By taking
x0=57
, or otherwise, show that
2941<2<7099,
and, without using tables, show that each of 2941 and 7099 differ from 2 by less than 5×10−4.
Suggestion
…without using tables, show that each of 2941 and 7099 differ from 2 by less than 5×10−4.
Without knowing 2, we can’t calculate the distance between xn and 2, but what’s the distance between xn and xn+1?
Solution
We can reason as follows:
xn+1<2⟺xn+1xn+2<2⟺(xn+2)<2(xn+1)⟺(1−2)xn<(2−2)⟺xn>2−12−2=22−12−1=2.
Since the same argument applies with < and > reversed, we’ve shown that xn and xn+1 always lie on opposite sides of 2.
Thus x0 and x1, and x2 and x3, always lie on opposite sides of 2.
We can write
∣x12−2∣=(x0+1x0+2)2−2=(x0+1)2(x0+2)2−2(x0+1)2=∣x0+1∣2∣x02+4x0+4−2x02−4x0−2∣=∣x0+1∣2∣2−x02∣=∣x0+1∣2∣x02−2∣<∣x02−2∣.
The final inequality follows since x0+1>1.
The same method as above shows that
∣x22−2∣=∣x1+1∣2∣x12−2∣=∣x0+1∣2∣x1+1∣2∣x02−2∣.
We can write
∣x1+1∣2=(x0+1x0+2+1)2=(x0+1)2(x0+2+x0+1)2=(x0+1)2(2x0+3)2.
Hence,
∣x22−2∣=∣x0+1∣2∣x02−2∣×(2x0+3)2(x0+1)2=(2x0+3)2∣x02−2∣<91∣x02−2∣,
with the inequality holding as x0>0 which implies that (2x0+3)2>9.
By taking
x0=57
, or otherwise, show that
2941<2<7099
From the first part of the question, the iteration goes from one side of
2
to the other, and back again, and so on. Since
x0=1.4<2
, we know that
2<x1=x0+1x0+2=7+57+10=1217
and that
2>x2=x1+1x1+2=17+1217+24=2941
and, finally, that
2<x3=x2+1x2+2=41+2941+58=7099.
…and, without using tables, show that each of 2941 and 7099 differ from 2 by less than 5×10−4.
With
x0=57
, we saw that
x2=2941
and
x3=7099
. We saw from the very first part of the question that
x2
and
x3
lie on opposite sides of
2
, and so
∣x2−2∣<∣x2−x3∣=2941−7099=203041×70−99×29=20301<5×10−4.
The same inequalities above also apply when x2 and x3 are interchanged, which proves the result.
Where did our initial iteration come from? We have
f(x)=x2−2=0⟺x2+x−2=x⟺x(x+1)−2=x⟺x=x+1x+2=g(x).
Thus we have in action here the x=g(x) iteration, or the rearrangement method, where we start with f(x)=0, and move to x=g(x).