I have problem with the divergence of a Sum, I don´t know how criteria use when i has a trigonometrical function because, it´s not monotone, and I can´t prove that this series diverge with the usual methods $$ \sum {\frac{{\cos \left( {\log \left( {\log n} \right)} \right)}} {{\log n}}} $$ it´s different
computing trigonometical
7
$\begingroup$
sequences-and-series
-
0This one is not very friendly, is it. – 2011-08-09
1 Answers
9
Because the denominator varies so slowly and the numerator even more slowly, you can show that the sequence of partial sums $S_k$ can't be Cauchy by adding up a suitable stretch of terms. For instance, with $k=\lceil\exp(\exp(2\pi m))\rceil$ and $l=\lfloor \exp(2\exp(2\pi m))\rfloor$, we have
$$S_l-S_k=\sum_{j=k+1}^l\frac{\cos(\log(\log j))}{\log j}\gt\sum_{j=k+1}^l\frac{\cos(2\pi m+\log2)}{\log l}\gt(l-k)\frac{\cos\log2}{\log l}>\cos\log 2\;,$$
where the last inequality holds for sufficiently large $m$. Since the difference of partial sums doesn't go to zero, the series diverges.
-
0Great answer. +1 – 2011-08-09
-
0Good answer, but i did not understand the first inequality D: only that. – 2011-08-09
-
0@Daniel: The denominators have increased. In the numerator, the arguments of the cosine lie between $2\pi m$ and $2\pi m + \log2$, corresponding to $0$ to $\log2$, and the values of the cosine in that range are $\ge\cos\log2$. – 2011-08-09
-
0Ah ok I got it, very good answer, , i did not saw "l" in the denominator , thanks again :D – 2011-08-09