Let f(x)=-2x+1 if x<0 and f(x)=$x^2$+x if x>0. Give a sequence {$x_n$} in $\mathbb R$\ {0} such that {$x_n$} converges to zero but {f($x_n$)} diverges.

I've tried with the sequence {1/n} but it does not seem to work. I can't think of another sequence that would converge to zero. Can anyone help me out please?


  • $\begingroup$ I am a little confused by what the question means by the function at x<0 and x>0. So the function of the sequence must diverge while the sequence converges. $\endgroup$ – Su003 Nov 17 '14 at 0:36
  • $\begingroup$ If x is negative than the function value is computed by the first expression while if x is positive then the function value is computed by the second expression. For example, f(-1)=3 while f(1)=2 $\endgroup$ – JB King Nov 17 '14 at 0:42
  • $\begingroup$ But x is a term of the sequence {$x_n$} right? $\endgroup$ – Su003 Nov 17 '14 at 0:53
  • $\begingroup$ $x$ is just a variable while $x_n$ would be a subscripted variable. $\endgroup$ – JB King Nov 17 '14 at 0:59

How about $x_n=\frac{(-1)^n}{n}$ for a sequence that converges to zero but alternates in sign which would make the function values be different. Thus, the sequence here is: $-1,\frac{1}{2},-\frac{1}{3},\frac{1}{4},-\frac{1}{5},\cdots$

For odd $n$, the function values will converge to 1. The first few function values here would be $3,\frac{5}{3},\frac{7}{5}$ as the general term will be $\frac{n+2}{n}$ which could also be seen as $1+\frac{2}{n}$

For even $n$, the function values would be $\frac{1}{n^2}+\frac{1}{n}=\frac{n+1}{n^2}$ which would have function values of $\frac{3}{4},\frac{5}{16},\frac{7}{36},...\frac{11}{100}, .... \frac{101}{10,000},...\frac{1001}{1,000,000}$ which will likely converge to 0 from this side as the denominator is growing much faster than the numerator.

Putting these together, the function value sequence would look like this: $3,\frac{3}{4},\frac{5}{3},\frac{5}{16},\frac{7}{5},\frac{7}{36},\frac{9}{7},\frac{9}{64}..., \frac{2k+1}{2k-1},\frac{2k+1}{(2k)^2},\frac{2k+3}{2k+1},\frac{2k+3}{(2k+2)^2}...$ which doesn't converge since there are sub-sequences which converge to different values.

If you want, consider $\varepsilon=\frac{1}{10}$ and try to prove the function value sequence converge,i.e. there exists $L,N$ such that for all $n>N\implies|f(n)-L|<\varepsilon$.

If you believe they converge to 0, consider some big odd $n>N$ where the function value will be greater than 1 which doesn't work.

If you believe they converge to 1, consider some big even $n$ where the function value will be close to zero and more than .1 away from 1 as for $n=1000$ then the function value is $\frac{1001}{1,000,000}=.001001$. Any bigger even $n$ will have even smaller values to consider.

If you have an alternative proof of what is wrong with my above answers, show me the error.

  • $\begingroup$ Does {f($x_n$)} diverge in this case? I tried it but it looks like it converges to 1. $\endgroup$ – Su003 Nov 17 '14 at 0:31
  • 2
    $\begingroup$ @Su003, even terms will converge to $0$, odd terms to $1$. $\endgroup$ – lhf Nov 17 '14 at 0:33
  • $\begingroup$ I understand this. But what is confusing me is that the question says that {f($x_n$)} must diverge. Does such an {$x_n$} work in this case? $\endgroup$ – Su003 Nov 17 '14 at 1:37

The main point here is that $f$ is not continuous at $x=0$. It does have one-sided limits at $x=0$, but they are different. So, you need to find a sequence that converges to $0$ but visits both sides infinitely often. The answer by JB King uses the simplest possible such sequence.


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.