Let $(f_n)_n$ be a sequence of real functions of a single real variable with compact support in $[0,1]$ and of bounded variation all of them. Let the sequence be uniformly convergent to $0$. Is it true that the sequence of their total variations (on $[0,1]$), say $(V_0^1(f_n))_n$, is convergent to $0$?


Definitely not. Consider the sequence $f_n(x)=\frac{1}{n}\cos(2n\pi x)$. This converges uniformly to $0$ on $[0,1]$, but $V_0^1(f_n)>1$ for all $n$.

  • $\begingroup$ I wonder, would it make a difference if $f_n = F_n - F$, where $F_n$ and $F$ are cumulative distribution functions such that $(F_n)_n$ is uniformly convergent to $F$? This is more tightly related to my original problem. $\endgroup$ – Marcos Dec 29 '18 at 6:30
  • $\begingroup$ I can't think of this on the top of my head, but I imagine you'll have better results. I'll come back to this after I've had some sleep. Are there any other hypotheses on the $F_n$ and $F$ which would simplify this? $\endgroup$ – Aweygan Dec 29 '18 at 7:05
  • $\begingroup$ Thank you very much indeed for your effort. I am not sure that this helps, but every $F_n$ is an increasing step function with precisely $n+1$ jumps uniformly spaced on $[0,1]$ at $x_k = k/n$ for $k=0, \dotsc, n$ (but not in general of the same height all of them), and it is bounded (as a cdf) by $0$ from below and by $1$ from above. $\endgroup$ – Marcos Dec 29 '18 at 7:16

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.