I have to show that the recursive sequence given by $$ b_1 = 1, ~ b_{n+1} = b_n + \frac{1}{(3 + (-1)^n)^n} $$ converges.
I can show its convergence by showing its monotone and bounded by 2 but I wondered, if I can find a implicit definition of the sequence without an $(-1)^n$ terms in it. Obviously, $$b_n = b_1 + \sum_{k = 0}^{n-1} \frac{1}{(3 + (-1)^n)^n}$$
I figured out that
$$ b_{2n} = \sum_{k = 0}^{n} \frac{1}{16^k} + \frac{1}{2 \cdot 4^k} = \sum_{k = 0}^{n} \frac{4^k + 2}{2^{4k+1}} $$
which converges by the geometric series test (also WolframAlpha says so), whilst it WolframAlpha doesn't make any statement about the convergence of $\displaystyle \sum_{k = 0}^{n-1} \frac{1}{(3 + (-1)^n)^n}$.
Is there a implicit definition of $(b_n)_{n \in \mathbb{N}}$ without the $(-1)^n$, or is that simply not possible?
The sequence is increasing. It sufficient to show that it is bounded.
We have $$b_{n+1}-b_n=\frac 1{(3+(-1)^n)^n}\leqslant\frac1{2^n}\implies b_n-b_1\leqslant \frac1{2^{n-1}}+\frac1{2^{n-2}}+\ldots+\frac1{2}<1$$