many struggle to solve this infinite series

logistic_guy

Full Member
Joined
Apr 17, 2024
Messages
998
Solve.

n=1(1)n1n\displaystyle \sum_{n=1}^{\infty}(-1)^{n}\frac{1}{n}
 
Solve.

n=1(1)n1n\displaystyle \sum_{n=1}^{\infty}(-1)^{n}\frac{1}{n}
n=1(1)n1n=n=1(1)n+11n=ln2\displaystyle \sum_{n=1}^{\infty}(-1)^{n}\frac{1}{n} = -\sum_{n=1}^{\infty}(-1)^{n+1}\frac{1}{n} = -\ln 2

A detailed solution was given here:


Now suppose that you have no idea what a Taylor series is and you still wanna find the value of the sum. Is there a way? Yes if it is an alternating series, you can approximate the sum within what you desire!

Say you wanna guarantee to get the sum of that alternating series within 0.01\displaystyle 0.01.

First step take 1n\displaystyle \frac{1}{n}. Add one to n\displaystyle n, that gives 1n+1\displaystyle \frac{1}{n + 1}.

Now solve for n\displaystyle n.

1n+10.01\displaystyle \frac{1}{n + 1} \leq 0.01

This gives:

n99\displaystyle n \geq 99

This means that you need at least 99\displaystyle 99 terms to get an error no more than 0.01\displaystyle 0.01.

Let us calculate.

n=199(1)n1n=0.698172\displaystyle \sum_{n=1}^{99} (-1)^{n} \frac{1}{n} = -0.698172

The actual sum is:

n=1(1)n1n=0.693147\displaystyle \sum_{n=1}^{\infty} (-1)^{n} \frac{1}{n} = -0.693147

This shows that:

0.6981720.010.6931470.698172+0.01\displaystyle -0.698172 - 0.01 \leq -0.693147 \leq -0.698172 + 0.01

Or

0.7081720.6931470.688172\displaystyle -0.708172 \leq \textcolor{red}{-0.693147} \leq -0.688172

Which guarantees that our sum is within the desired approximation with an absolute error equal to:

0.698172(0.693147)=0.005025<0.01\displaystyle |-0.698172 - (-0.693147)| = 0.005025 < 0.01 (Our result meets the error requirement.)

But without Taylor series, there is no way you will know that the sum converges to exactly ln2\displaystyle -\ln 2.

:sneaky:😏
 
Last edited:
Top