Given a convergent series

$$\sum\limits_{k = 1}^\infty a_k < \infty$$

Then claim:

$\exists N$, such that $\exists \epsilon > 0$, $\sum\limits_{k = > N+1}^\infty a_k < \epsilon$

I tried to go by definition

$$\sum\limits_{k = 1}^\infty a_k = L < \infty$$ is convergent if $\forall \epsilon > 0, \forall k \geq N, |L - \sum\limits_{k = 1}^\infty a_k | < \epsilon$

But the claim eliminates the $L$...not sure what to do next

  • 2
    $\begingroup$ Let $N=0$ and $\varepsilon=\left|\sum \limits_{k=1}^\infty\left(a_k\right)\right|+1$. $\endgroup$ – Git Gud Mar 26 '16 at 22:19
  • 1
    $\begingroup$ You probably want to show that for all $\epsilon \gt 0$ there is an $N$ such that $\dots$. $\endgroup$ – André Nicolas Mar 26 '16 at 22:25

You have a little mistake in your argument. What $\displaystyle{\sum_{k=1}^\infty a_k=L<\infty}$ means is that the sequence of partial sums $\{\displaystyle{S_n=\sum_{k=1}^n a_k}:n\in\mathbb{N}\}$ converges to $L$, that is, for all $\epsilon>0$ there exists $N\in \mathbb{N}$ such that if $n\geq N$ we have $|S_n-L|<\epsilon$, that is:

$$\forall \epsilon>0 \left(\exists N\in\mathbb{N}\left(n\geq N \rightarrow |L-S_n|=\left|\sum_{k=1}^\infty a_k - \sum_{k=1}^n a_k\right|<\epsilon\right)\right)$$

In particular, for $n=N$ we have:

$$\sum_{k=1}^\infty a_k-\sum_{k=1}^Na_k=\sum_{k=N+1}^\infty a_k<\epsilon ,$$as desired.

  • $\begingroup$ Good job although crazy amount of brackets for your definition where you learn that $\endgroup$ – Shamisen Expert Mar 27 '16 at 0:02
  • 1
    $\begingroup$ Indeed, much too many brackets, especially in view of the missing "$\forall n\in\mathbb N$". $\endgroup$ – Did Mar 27 '16 at 7:36

Your statement is wrong.

$\sum_{k=1}^\infty a_k=L$ means that the sequence of partial sums $s_n=\sum_1^n a_k$ converges to $L$.

That is, for any $\varepsilon>0$, there exists $N$ such that $|L-s_n|<\varepsilon\forall n>N$

However, $L-s_n=\sum_{1}^\infty a_k-\sum_1^n a_k=\sum_{N+1}^\infty a_k$.

Thus, you reach the needed conclusion.

However, if one needs strict rigour regarding the sum from $N+1$ to $\infty$, then one might note:

$s_n-s_m=\sum_{n+1}^m a_k$ for all $m>n$.

and then take the limit of $m$ as it goes to $\infty$. to see that $L-s_n=\sum_{N+1}^\infty a_k$

  • $\begingroup$ Actually the statement in the question is not wrong, it always hold. $\endgroup$ – Did Mar 26 '16 at 22:30
  • $\begingroup$ Not when he says that he's going by the definition ;) (I meant that his statement that this is the definition is wrong) $\endgroup$ – Hasan Saad Mar 26 '16 at 22:31
  • $\begingroup$ Oh, you mean "I tried to go by definition $\sum\limits_{k = 1}^\infty a_k = L < \infty$ is convergent if $\forall \epsilon > 0$, $\forall k \geq N$, $\left|L - \sum\limits_{k = 1}^\infty a_k \right| < \epsilon$"? Right, this part escaped me. (Dunno whether it is more wrong than absurd or the opposite, though...) $\endgroup$ – Did Mar 26 '16 at 22:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.