I am currently studying uniform and pointwise convergence and I am stuck at a somewhat basic distinction. Until now our lecturer has been talking about sequences of functions. For example:

$$f_n(x)=\frac{n}{x} \space \space \text{or} \space \space f_n(x)=x^n$$

However, in our problem set I am supposed to test the following series for uniform convergence:


Are sequences of functions and series related? Is the process of showing uniform convergence different for series?


From any series $\sum_{k=0}^{\infty} a_k$, you can construct the sequence of partial sums, $$ S_n = \sum_{k=0}^n a_k, $$ and treat it as you would a sequence. Conversely, given any sequence $(b_n)_{n=0}^{\infty}$, the differences $$ a_0 = b_0, \quad a_k = b_k-b_{k-1} $$ can be made into a series, $$ b_n = \sum_{k=0}^{n} a_k. $$ This is much like differentiation and integration being "inverse operations" (scare quotes required for technical reasons, of course).

  • $\begingroup$ So in your example $S_1$ would be equal to $\sum_{k=0}^{1}a_K$; $S_2=\sum_{k=0}^{2}a_K$? Am I basically looking at the sequence of sums? $\endgroup$ – qmd Jun 3 '15 at 18:39
  • $\begingroup$ Aside: What is a series itself? I think that a sequence is well-defined mathematical object but series is not. $\endgroup$ – user87690 Jun 3 '15 at 18:42
  • $\begingroup$ Why is a series not well-defined? $\endgroup$ – qmd Jun 3 '15 at 18:47
  • $\begingroup$ @Rzeta: So how do you define a series? $\endgroup$ – user87690 Jun 3 '15 at 18:52
  • $\begingroup$ $\sum_0^n a_n=a_0+a_1+a_2+a_3+...+a_n$? $\endgroup$ – qmd Jun 3 '15 at 18:54

You can view a series as a sequence of partial sums. Try to think of the sequence of functions $$f_i(t)=\sum_{n=0}^i \frac{\sin(nt)}{e^n} $$

  • $\begingroup$ How would I prove/disprove uniform convergence in this case. As far as I know I always need some sort of a limit function when I am trying to test uniform convergence but In this case I am not sure how to even get to that limiting function. $\endgroup$ – qmd Jun 3 '15 at 18:53
  • 1
    $\begingroup$ @Rzeta: You don't need the limiting function. There are theorems you can use to prove that a series is convergent. In this case we have $|\sin(nt) / e^n| ≤ 1 / e^n$ and $∑_n 1 / e^n$ converges, so the original series converges uniformly. $\endgroup$ – user87690 Jun 3 '15 at 19:12
  • $\begingroup$ @user87690 Can you name me some of the theorems. Is the Weierstrass-M-test one of them? I just came accross that while researching. Also, is there any way I can apply things like the ratio test or the root test when testing for uniform convergence? $\endgroup$ – qmd Jun 3 '15 at 19:21
  • $\begingroup$ @Rzeta: Yes, Weierstrass-M-test is one of them and that's actually what I've used above. It says that if $|f_n(t)| ≤ a_n$ for every $n, t$ and $∑_n a_n$ converges, then $∑_n f_n(t)$ converges uniformly. You use things like the ratio test and the root test to show that your series $∑_n a_n$ chosen for Weierstrass actually converges. $\endgroup$ – user87690 Jun 3 '15 at 19:29
  • $\begingroup$ That actually helped me a ton. Things are starting to fall into place ;). So basically I just have to be "skilled" at choosing an appropriate convergent $\sum_n a_n$ for the Weierstrass-M-test and I am done? $\endgroup$ – qmd Jun 3 '15 at 19:35

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.