3
$\begingroup$

Applying ratio test, we can prove this series $\displaystyle \sum_{n=1}^{\infty} \sin\left(\frac{1}{2^n}\right)$ converges.

How can we calculate or estimate the sum?

Any help is appreciated, thank you.

$\endgroup$
1
  • 3
    $\begingroup$ You can use the fact that when $x$ is very small, $\sin x = x+o(x)$. So after a few terms, you can effectively (for the sake of approximation, nor exact computation) replace $\sin\frac{1}{2^n}$ by $\frac{1}{2^n}$. $\endgroup$
    – Clement C.
    Commented Jan 22, 2017 at 15:49

3 Answers 3

3
$\begingroup$

Suppose we want to compute the series with an error of at most $\epsilon>0$. We know (why?) that for $x>0$ we have $x-\frac16x^3<\sin x<x$. Hence for the tail of the series after the $N$th summand we have the bounds $$\sum_{n=N+1}^\infty\left(2^{-n}-\frac16\cdot 8^{-n} \right)<\sum_{n=N+1}^\infty \sin(2^{-n})<\sum_{n=N+1}^\infty2^{-n},$$ an using the formula for the geometric series $$2^{-N}-\frac16\cdot 7^{-N}<\sum_{n=N+1}^\infty\left(2^{-n}-\frac16\cdot 8^{-n} \right)<\sum_{n=N+1}^\infty \sin(2^{-n})<2^{-N}.$$ Thus it suffices to pick $N\ge \log_7 \frac{12}\epsilon $ (e.g., with $\epsilon =10^{-9}$, we can use $N=12$) and then use $$ 2^{-N}-\frac1{12}\cdot 7^{-N}+\sum_{n=1}^N\sin2^{-n}$$ as approximation (provided, the computational error for computing the sines is small enough).

$\endgroup$
2
$\begingroup$

$$\sum_{n=1}^{\infty} \sin\left(\frac{1}{2^n}\right)$$

  • For $n\ge1$, $0<\sin\left(\dfrac{1}{2^n}\right)<\dfrac{1}{2^n}\implies$ the given sum is convergent (in fact, absolutely convergent)

  • To have some fun with it, we can write:

$$\sum_{n=1}^{\infty} \sin\left(\frac{1}{2^n}\right)=\sum_{n=1}^{\infty} \sum_{k=0}^{\infty}\frac{(-1)^k}{(2k+1)!2^{n(2k+1)}}$$

$$=\sum_{k=0}^{\infty} \frac{(-1)^k}{(2k+1)!} \sum_{n=1}^{\infty}2^{-n(2k+1)} =\sum_{k=0}^{\infty} \frac{(-1)^k}{(2k+1)!}\frac{1}{2^{2k+1}-1}$$

This isn't an "exact" form for the sum, but is fairly quick to converge, if you're interested in computing the sum.

$\endgroup$
1
  • 2
    $\begingroup$ Interestingly, the last series is alternating hence the true value of the sum is between two successive partial sums. And since the terms go to zero pretty fast, such approximations would be rigorous and quite accurate. $\endgroup$
    – Did
    Commented Jan 22, 2017 at 17:19
1
$\begingroup$

As we know $\dfrac{x}{2}\leqslant\sin x\leqslant x$ for $0\leqslant x\leqslant\dfrac{\pi}{2}$, so

$$\frac12=\sum_{n=1}^{\infty}\frac{1}{2^{n+1}}\leqslant\sum_{n=1}^{\infty}\sin\frac{1}{2^n}\leqslant\sum_{n=1}^{\infty}\frac{1}{2^n}=1$$

$\endgroup$
2
  • 1
    $\begingroup$ It seems unreasonable to use asymptotic estimates of the function to get an idea of the value of the sum. Note that the same function, but modified on $(1/2^{100},\infty)$, would satisfy the same asymptotics but yield a quite different sum, hence more precise arguments are needed here. The first part of the answer is allright. $\endgroup$
    – Did
    Commented Jan 22, 2017 at 16:22
  • $\begingroup$ @Did Thanks. Corrected. $\endgroup$
    – Nosrati
    Commented Jan 22, 2017 at 16:54

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .