2
$\begingroup$

$$\displaystyle \operatorname{lim}_{x\to \infty}(\operatorname{log}x)^{\frac{1}{x}}$$

$t=\frac1x \Rightarrow \displaystyle \operatorname{lim}_{t\to 0^+}(\operatorname{log}\frac1t)^t\Rightarrow(-1)^t(\operatorname{log}t)^t$

In my opininion the limit doesnot exist.Is it Correct?

$\endgroup$
  • 1
    $\begingroup$ Hint: $$(\log x)^{1/x}=\exp (\frac{\log\log x}{x})$$ $\endgroup$ – BAI May 14 '18 at 7:50
  • $\begingroup$ it exists and is $1$ . btw , i used a graph to verify this. I'm not sure how to approach this in a algebraic way. $\endgroup$ – The Integrator May 14 '18 at 7:50
1
$\begingroup$

Note: $(-\ln 0.5)^{0.5}=((-1)\cdot(-\ln 2))^{0.5}\ne (-1)^{0.5}\cdot (-\ln 2)^{0.5}$, because the last is undefined.

If you want to make a change, let: $\ln x=t \Rightarrow x=e^t$. Then: $$\lim_{x\to\infty} (\ln x)^{1/x}=\lim_{t\to\infty} t^{1/e^t}=\lim_{t\to\infty} \exp\left(\frac{\ln t}{e^t}\right)=\exp\left(\lim_{t\to\infty} \frac{\ln t}{e^t}\right)=\exp(0)=1.$$

$\endgroup$
  • $\begingroup$ Thank you understood. $\endgroup$ – DRPR May 14 '18 at 11:46
1
$\begingroup$

Rewrite the functions as $e^\frac{\log \log x }{x}$, since $\exp$ is a continuous functions, you can interchange $\exp$ and limit operator and get $1$ as the limit.

$\endgroup$
1
$\begingroup$

Consider $$y=\sqrt[x]{\log (x)}\implies \log(y)=\frac 1x \log(\log(x))$$ $x$ goes much faster than $\log(x)$ and still faster than $\log(\log(x))$.

So, $\log(y)\to 0$ and $y=e^{\log(y)}\to 1$

$\endgroup$
1
$\begingroup$

Note that by standard limit $t^t\to 1$ as $t\to 0^+$ we have that

$$(\log x)^{\frac{1}{x}}=x^{\frac1x} \left[\left(\frac{\log x}{x}\right)^{\frac{\log x}{x}}\right]^\frac1{\log x}\to 1$$

indeed

  • $x^{\frac1x}$ with $y=\frac 1x \to 0^+ \implies\frac1{y^y}\to 1$
  • $\left(\frac{\log x}{x}\right)^{\frac{\log x}{x}}$ with $t=\frac{\log x}{x}\to 0$ form above then $\left(\frac{\log x}{x}\right)^{\frac{\log x}{x}}\to 1$

  • $\frac1{\log x}\to 0$

$\endgroup$
1
$\begingroup$

You can't do $$ (-\log t)^t=(-1)^t(\log t)^t $$ because general exponentiation is only defined for positive base and here $-1$ and $\log t$ are negative, as $t\to0^+$.

When confronted with $f(x)^{g(x)}$ it's most of the times convenient to compute the limit of $\log(f(x)^{g(x)}=g(x)\log f(x)$ and then use the properties of $\exp$. In this case $$ \lim_{x\to\infty}\log((\log x)^{1/x})= \lim_{x\to\infty}\frac{\log\log x}{x} $$ There are several ways to finish this up. You can observe that $\log\log x<\log x$ and $$ \lim_{x\to\infty}\frac{\log x}{x}=\lim_{t\to\infty}\frac{t}{e^t}=0 $$ Or you can directly apply l'Hôpital: $$ \lim_{x\to\infty}\frac{\log\log x}{x} = \lim_{x\to\infty}\frac{1}{x\log x}=0 $$ Thus $$ \lim_{x\to\infty}(\log x)^{1/x}=e^0=1 $$

$\endgroup$
  • $\begingroup$ Your answer looks correct. But $(-3)^3=(-1)^3(3)^3.$Then why cant i do the same with$(-log\, t)^t$? $\endgroup$ – DRPR May 14 '18 at 9:24
  • $\begingroup$ @DRPR And what's $(-1)^{\sqrt{2}}$ or, more generally, $x^y$ where $x<0$ and $y$ is irrational? By “general exponentiation” I mean “with any real exponent”. $\endgroup$ – egreg May 14 '18 at 9:26

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.