How to approximate the sum of a convergent, positive series

443 Views Asked by At

$$S=\sum_{n=0}^{\infty}\frac1{e^n(n^2+1)}$$

This series converges because the general term goes to $0$ faster than $\dfrac1{n^2}$

I am asked to approximate the series with an error $R<10^{-3}$.

How can I do this? I know that

$$R=\left\lvert S-S_k\right\rvert=\left\lvert \sum_{n=0}^{\infty}\frac1{e^n(n^2+1)}-\sum_{n=0}^{k}\frac1{e^n(n^2+1)}\right\rvert=\sum_{n=k+1}^{\infty}\frac1{e^n(n^2+1)}$$ and that

$$\int_{k+1}^\infty \frac1{e^x(x^2+1)}\mathrm dx\le\sum_{n=k+1}^{\infty}\frac1{e^n(n^2+1)}\leq\int_{k}^\infty \frac1{e^x(x^2+1)}\mathrm dx$$

My attempt:

$$R=\sum_{n=k+1}^{\infty}\frac1{e^n(n^2+1)}\le\frac1{(k+1)^2+1}\sum_{n=k+1}^{\infty}\frac1{e^n}\underbrace\le_{(1)}\frac1{(k+1)^2+1}\int_k^\infty \frac1{e^x}\mathrm dx= \frac1{(k+1)^2+1}\frac1{e^k}<10^{-3}\quad (*)$$

$(1):$ as the series is decreasing, positive and continuous, I can switch to the integral.

$(*)$ is true for $k=4$.

Is this a right approach?

P.S. The exercise comes from Calculus Problems, 16.18, page 231.

1

There are 1 best solutions below

3
On BEST ANSWER

An easily computable upper bound is

$$R =\sum_{n=k+1}^{\infty}\frac1{e^n(n^2+1)} \lt \frac1{(k+1)^2}\sum_{n=k+1}^{\infty}\frac1{e^n} =\frac1{(k+1)^2e^{k+1}(1-1/e)}$$