Solution to 8.29

(a) -> Suppose  [Graphics:Images/Assignment_8_gr_1.gif] is bounded by [Graphics:Images/Assignment_8_gr_2.gif]. Then so is  [Graphics:Images/Assignment_8_gr_3.gif] Since  [Graphics:Images/Assignment_8_gr_4.gif] is a bounded and increasing, it follows that it converges to some [Graphics:Images/Assignment_8_gr_5.gif].

We show that the double sequence s(p,q) also converges to [Graphics:Images/Assignment_8_gr_6.gif]. Take Epsilon>0. Then there exists;N>0 such that &ForAll p>N, we have [Graphics:Images/Assignment_8_gr_7.gif], i.e. [Graphics:Images/Assignment_8_gr_8.gif]. Take that N. Suppose p,q>N. Let us assume (no loss of generality) that p<=q. Then, by assumption, [Graphics:Images/Assignment_8_gr_9.gif]. But since [Graphics:Images/Assignment_8_gr_10.gif], it follows that [Graphics:Images/Assignment_8_gr_11.gif].

<- If the double sequence [Graphics:Images/Assignment_8_gr_12.gif] converges to [Graphics:Images/Assignment_8_gr_13.gif] (i.e. if the double series [Graphics:Images/Assignment_8_gr_14.gif] converges to [Graphics:Images/Assignment_8_gr_15.gif]), then surely each partial sum is bounded by [Graphics:Images/Assignment_8_gr_16.gif] (since all terms are positive).

(b) Suppose [Graphics:Images/Assignment_8_gr_17.gif] converges to [Graphics:Images/Assignment_8_gr_18.gif]. Want to show that [Graphics:Images/Assignment_8_gr_19.gif] also converges.

Every partial sum  [Graphics:Images/Assignment_8_gr_20.gif]is in the interval [Graphics:Images/Assignment_8_gr_21.gif]: this is true since the absolute value of such partial sum is less than the sum od the absolute values of the terms, which in turn is less than [Graphics:Images/Assignment_8_gr_22.gif].

 

If the set of all numbers [Graphics:Images/Assignment_8_gr_27.gif] is finite, then the set of all numbers [Graphics:Images/Assignment_8_gr_28.gif] is also finite, and so the same is true for the set of numbers [Graphics:Images/Assignment_8_gr_29.gif]. In that case all of these terms except finitely many should be equal to 0, for else [Graphics:Images/Assignment_8_gr_30.gif] would not converge. The rest of that case is easy.

Suppose now that the set of all numbers [Graphics:Images/Assignment_8_gr_31.gif] is infinite. Then (Bolzano), it must have an accumulation point in the interval [Graphics:Images/Assignment_8_gr_32.gif]. If we show that there are no two such, then we are done. Suppose otherwise: suppose [Graphics:Images/Assignment_8_gr_33.gif] and [Graphics:Images/Assignment_8_gr_34.gif] are two accumulation points of the double sequence {[Graphics:Images/Assignment_8_gr_35.gif]}. Denote their distance by [Graphics:Images/Assignment_8_gr_36.gif]. Then, for any (large) N, there are [Graphics:Images/Assignment_8_gr_37.gif]such that [Graphics:Images/Assignment_8_gr_38.gif] is less than [Graphics:Images/Assignment_8_gr_39.gif] units far from [Graphics:Images/Assignment_8_gr_40.gif] and such that  [Graphics:Images/Assignment_8_gr_41.gif] is less than [Graphics:Images/Assignment_8_gr_42.gif] units far from [Graphics:Images/Assignment_8_gr_43.gif]. So that we can find arbitrarily large [Graphics:Images/Assignment_8_gr_44.gif]such that [Graphics:Images/Assignment_8_gr_45.gif] and  [Graphics:Images/Assignment_8_gr_46.gif] are at least [Graphics:Images/Assignment_8_gr_47.gif] units apart. But the difference between  [Graphics:Images/Assignment_8_gr_48.gif] and  [Graphics:Images/Assignment_8_gr_49.gif] is certainly less than [Graphics:Images/Assignment_8_gr_50.gif]  (this is easy and involves inequalities with absolute values). On the other hand, since [Graphics:Images/Assignment_8_gr_51.gif] converges, [Graphics:Images/Assignment_8_gr_52.gif] could be made as small as we please by choosing  sufficiently large [Graphics:Images/Assignment_8_gr_53.gif]. In portraiture, it could certainly be made to be less that [Graphics:Images/Assignment_8_gr_54.gif]. So we get  [Graphics:Images/Assignment_8_gr_55.gif] for some [Graphics:Images/Assignment_8_gr_56.gif]. So, [Graphics:Images/Assignment_8_gr_57.gif] and [Graphics:Images/Assignment_8_gr_58.gif] must be equal.

(c) Since [Graphics:Images/Assignment_8_gr_59.gif] converges (earlier Calculus) it follows from Theorem 8.44 that the double series in that question also converges.


   January 9, 2001