[Math] Is recognizing ‘peaks’ of a sequence from its monotone sub-sequence possible

real-analysissequences-and-series

In page 45 in Advanced Calculus by P M Fitzpatrick, it is written:

enter image description here

A question raises from the converse of Case 1 and of Case 2, respectively. Is this statement correct: "If a subsequence of a sequence is strictly increasing then the original sequence has finitely many peaks and if a subsequence of a sequence is monotonically decreasing then the original sequence has infinitely many peaks"?

Best Answer

The first statement is false. For $n\in\Bbb N$ let $a_{2n}=0$ and $a_{2n+1}=-\frac1{2n+1}$. Then the sequence

$$\langle a_n:n\in\Bbb N\rangle=\left\langle 0,-1,0,-\frac13,0,-\frac15,\ldots\right\rangle$$

has a peak at each even index, but the subsequence $\langle a_{2n+1}:n\in\Bbb N\rangle$ is strictly increasing.

The second statement is also false: for $n\in\Bbb N$ let $a_n=(-1)^nn$, so that

$$\langle a_n:n\in\Bbb N\rangle=\langle 0,-1,2,-3,4,-5,\ldots\rangle\;;$$

this sequence clearly has a monotonically strictly decreasing subsequence, and it has no peaks at all.