1
$\begingroup$

Let $f_n, f:[a, b] \rightarrow \mathbb{R} $ for $n=1,2,3,...$ and suppose $\lim_{n\rightarrow \infty} f_n(t) =f(t) $, ie for all $t \in [a,b], \epsilon>0$ there exists $N(t, \epsilon) \in \mathbb{N} $ such that for $n>N, |f_n(t) - f(t) |<\epsilon. $

Consider the set $A = \lbrace N(t, \epsilon) :t \in [a, b] \rbrace$. Then for all epsilon greater than zero why wouldn't it be true for $ n>\sup A$ (which is an integer), that $|f_n(t) - f(t) |< \epsilon$, for all $t\in[a, b] $? (ie the sequence is uniformly convergent)

$\endgroup$
  • 1
    $\begingroup$ What if $\sup A = \infty$? $\endgroup$ – Sangchul Lee Jun 13 '17 at 7:08
  • 1
    $\begingroup$ how could that be? By pointwise convergence there is an $N_t$ for each $t$ in the interval so wouldnt the sup (max?) be one of these $N_t's$? $\endgroup$ – Rlanls Jun 13 '17 at 7:13
  • $\begingroup$ @Arthur, No. Improving pointwise convergence to uniform convergence occurs only in rare occasions. For instance, Dini's theorem tells that if increasing sequence of continuous functions on a compact set converges pointwise to a continuous function, then the convergence is indeed uniform. $\endgroup$ – Sangchul Lee Jun 13 '17 at 7:14
  • $\begingroup$ You are taking supremum of $N(t, \epsilon)$ over uncountably many $t$'s. There is no reason to expect that this is finite. For instance, if $f_n, f : [0, 1] \to \mathbb{R}$ are such that $f_n(t) = t^n$ and $f$ is the pointwise limit, then for $t \in (0, 1)$ and $\epsilon \in (0, 1)$ we have the lower bound $$N(t, \epsilon) \geq \left\lceil \frac{\log\epsilon}{\log t} \right\rceil - 1$$ and hence $\sup A = \infty$. $\endgroup$ – Sangchul Lee Jun 13 '17 at 7:16
  • $\begingroup$ ok so the uncountability is the issue. what assumptions would allows the uncountable set of $N(t,\epsilon)$s to have a supremum? $\endgroup$ – Rlanls Jun 13 '17 at 11:43
4
$\begingroup$

Consider the sequence $f_n(t)=t^n$ on $[0,1]$ then $f(t)=0$ for $t\in [0,1)$ and $f(1)=1$. Then $$\sup_{t\in[0,1]}|f_n(t)-f(t)|=\sup_{t\in[0,1)}t^n=1$$ and the convergence is not uniform in $[0,1]$.

What is $N(t,\epsilon)$ in this case? What is $\sup \{ N(t, \epsilon) :t \in [a, b] \}$?

$\endgroup$

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.