Is there a closed form for the series
\[\sum_{n\ge1}\frac{(-1)^{n!}}{n}\,?\]

- anonymous

Is there a closed form for the series
\[\sum_{n\ge1}\frac{(-1)^{n!}}{n}\,?\]

- schrodinger

See more answers at brainly.com

At vero eos et accusamus et iusto odio dignissimos ducimus qui blanditiis praesentium voluptatum deleniti atque corrupti quos dolores et quas molestias excepturi sint occaecati cupiditate non provident, similique sunt in culpa qui officia deserunt mollitia animi, id est laborum et dolorum fuga.
Et harum quidem rerum facilis est et expedita distinctio. Nam libero tempore, cum soluta nobis est eligendi optio cumque nihil impedit quo minus id quod maxime placeat facere possimus, omnis voluptas assumenda est, omnis dolor repellendus.
Itaque earum rerum hic tenetur a sapiente delectus, ut aut reiciendis voluptatibus maiores alias consequatur aut perferendis doloribus asperiores repellat.

Get this expert

answer on brainly

SEE EXPERT ANSWER

Get your **free** account and access **expert** answers to this

and **thousands** of other questions

- anonymous

Heck, does it even converge?

- Empty

I think it diverges because after n=2 we lose the interesting aspect of \( (-1)^{n!} = 1\) So it simplifies to basically the harmonic series.

- anonymous

Oh, haha didn't notice that \(n!\) would be even. Never mind...

Looking for something else?

Not the answer you are looking for? Search for more explanations.

## More answers

- dan815

you're forgiven

- anonymous

Thanks dan :P
Well I suppose a more general question might be in order. Something like, what conditions on \(b_n\) are sufficient such that
\[\sum_{n\ge1}\frac{(-1)^{b_n}}{n}\]
converges?

- Empty

Yessssssssssssss

- dan815

how about
\[\sum_{n\ge1}\frac{(-1)^{-1^{n}}}{n}\]

- dan815

lmao that is actually such a useless series now that i think about it

- Empty

You know what this is really quite fascinating @SithsAndGiggles I think we already know if it alternates it converges. But what about if we have:
\(b_{3n} = 1\)
\(b_{3n+1} = 0\)
\(b_{3n+2}=0\)
we would have the series:
\[1 + \frac{1}{2} - \frac{1}{3} + \frac{1}{4} + \frac{1}{5} - \frac{1}{6} + \cdots \]
Does this converge? :O

- anonymous

So two positive, one negative, is that the general pattern?

- Empty

Yeah, exactly.

- dan815

i want to say it converges

- Empty

Or I could try to generalize the problem like you two have done and try to make the most complicated looking but really actually trivial version of the harmonic series.

- Empty

:P

- Empty

Ok dan if it converges then can we say it will converge for any value? Let's say we only make one value negative every 1,000 numbers?

- dan815

i think it will still

- dan815

until infinity

- dan815

it feels like its right on the edge

- Empty

The difference between 1 and .999... ? :P

- Empty

That reminds me, the p-series test as they called it was that if p > 1 this series converges:
\[ \sum_{n=1}^\infty \frac{1}{n^p}\]
So you may be on to something here.

- dan815

ya thats what i was thinking about

- dan815

hmm how about rewriting it like this

- dan815

maybe it might diverge actually

- dan815

|dw:1439858847534:dw|

- Empty

This reminds me of something I was looking at earlier, the number of numbers relatively prime to a power of a prime number is given by the totient function,
\[\phi(p^n) = p^n * (1-\frac{1}{p}) \] So we can set up a density of relatively prime numbers always:
\[\rho(p)=\frac{\phi(p^n)}{p^n} = 1-\frac{1}{p} \]
So if there are infinite numbers, they will be relatively prime to any finite number:
\[\rho(\infty) = 1\]
Ok I don't think that actually relates but now I've typed it it's too late but to try to bring this back on track a bit more:
----
What patterns other than these very trivial ones of periodically flopping signs will result in convergence? What if instead we do:
\(b_n = 1 \) if \(p_n\) is prime
\(b_n=0\) otherwise.

- anonymous

I'm okay with accepting that all math is connected in some way, no matter how obscurely :)

- dan815

|dw:1439858901052:dw|

- anonymous

This link gives an interesting approach with asymptotic notation: http://math.stackexchange.com/a/1400659/170231

- anonymous

Not the same series, but in the same vein of this "family".

- dan815

|dw:1439859177697:dw|

- Empty

I don't know if we can rearrange infinite sums like this or not dan, for instance this convergent sum can be rearranged into two divergent sums, so idk
\[\sum \frac{(-1)^n}{n} = \sum \frac{1}{2n} - \sum \frac{1}{2n-1}\]

- dan815

right but thats fine, just because they both diverge doesnt mean it wont converge right

- anonymous

Right, \(\infty-\infty\) isn't a determinate form.

- Empty

|dw:1439859395039:dw|

- dan815

|dw:1439859500069:dw|

- dan815

does this converge or diverge

- anonymous

For \(n\ge1\), you get \(n^{1/n}\le n\), so it looks divergent.

- dan815

ya apparently that thing is actually nearly one order more than a harmonic series

- dan815

|dw:1439859913671:dw|

- dan815

did u know that?

- dan815

for all positive numbers more than 1

- anonymous

@Empty I don't think the series you gave (+, +, -, +, +, -, ...) converges. If \(H_n\) denotes the \(n\)th partial sum \(\sum\limits_{k\ge1}^n\frac{1}{k}\), then
\[\begin{matrix}
H_n&=&1&+&\dfrac{1}{2}&+&\dfrac{1}{3}&+&\cdots\\
\dfrac{2}{3}H_n&=&2\bigg(\dfrac{1}{3}&+&\dfrac{1}{6}&+&\dfrac{1}{9}&+&\cdots\bigg)
\end{matrix}\]
Subtracting gives your series,
\[\frac{1}{3}H_n=1+\frac{1}{2}-\frac{1}{3}+\cdots\]
and \(H_n\) diverges as \(n\to\infty\).

- Empty

Here's an interesting and related series:
\[\sum_{n=1}^\infty \frac{1}{4n} + \frac{1}{4n+1} - \frac{1}{4n+2} - \frac{1}{4n+3}\]

Looking for something else?

Not the answer you are looking for? Search for more explanations.