3
$\begingroup$

Can we show that the following function is integrable

\begin{align} f(t) =\frac{2^{\frac{it+1}{1.5}}}{2^{\frac{it+1}{2}}} \frac{\Gamma \left( \frac{it+1}{1.5} \right) }{\Gamma \left(\frac{ it+1}{2} \right)}, \end{align} where $t \in \mathbb{R}$ and $i=\sqrt{-1}$.

That is can we show that \begin{align} \int_{-\infty}^{\infty} |f(t)| dt<\infty. \end{align}

I was wondering if Stirling's approximation can be used, since this is a complex case?

Note if Stirling's approximation can be used than \begin{align} f(t) \approx \sqrt{\frac{1.5}{2}}\frac{2^{\frac{it+1}{1.5}}}{2^{\frac{it+1}{2}}} \frac{ \left( \frac{it+1}{1.5 e} \right)^{\frac{1+it}{1.5}} }{ \left(\frac{ it+1}{2 e} \right)^{\frac{1+it}{2}}}. \end{align}

Note that \begin{align} |f(t)| &\approx \left| \frac{2^{\frac{it+1}{1.5}}}{2^{\frac{it+1}{2}}}\right| \left| \frac{ \left( \frac{it+1}{1.5 e} \right)^{\frac{1+it}{ 1.5}} }{ \left(\frac{ it+1}{2 e} \right)^{\frac{1+it}{2}}}\right|\\ &=\left| \frac{2^{\frac{it+1}{1.5}}}{2^{\frac{it+1}{2}}}\right| \left| \frac{ \left( it+1\right)^\frac{1+it}{1.5} }{ \left( it+1 \right)^{\frac{1+it}{2}}}\right| \left|\frac{ \left(2 e\right)^{\frac{1+it}{2 }}} {(1.5 e)^{\frac{1+it}{ 1.5}}} \right| \end{align}

Also we have that \begin{align} &\left|\frac{2^{\frac{it+1}{1.5}}}{2^{\frac{it+1}{2}}}\right|= 2^{\frac{2}{3}-\frac{1}{2}} \\ & \left|\frac{ \left(2 e\right)^{\frac{1+it}{2 }}} {(1.5 e)^{\frac{1+it}{ 1.5}}} \right| =\frac{ \left(2 e\right)^{\frac{1}{2 }}} {(1.5 e)^{\frac{1}{ 1.5}}} \end{align}

So, in the end we if everthing is corect we have to show that \begin{align} g(t)=\left| \frac{ \left( it+1\right)^\frac{1+it}{1.5} }{ \left( it+1 \right)^{\frac{1+it}{2}}}\right| \end{align} is integrable, but I am not sure how to show if the above equation is integrable or not?

Any ideas would be greatly appreciated. Thank you.

Edit The integrability of $g(t)$ has been shown in one of the answers.

My question now is the following: Since we have that \begin{align} |f(t)| =|g(t)| +e \end{align} and $g(t)$ is integrable does this mean that $f(t)$ is inegrable? Can we prove that the error term $e$ is also integrable?

$\endgroup$
  • $\begingroup$ @polfosol Yes, $i$ here is the imaginary number. I will add this. $\endgroup$ – Boby Jan 5 '17 at 16:25
  • $\begingroup$ Why the title is different than the question? $\endgroup$ – polfosol Jan 5 '17 at 16:31
  • $\begingroup$ @polfosol Yes, you right. Corrected that too. $\endgroup$ – Boby Jan 5 '17 at 16:41
  • 1
    $\begingroup$ "Experimentally" (i.e. by plotting the function), it appears that $\ln |f(t)| \approx -|t|/4$ asymptotically; if this is actually true, it should be integrable. No idea how to prove it, though. $\endgroup$ – Michael Seifert Jan 5 '17 at 22:17
  • $\begingroup$ @MichaelSeifert Thanks. I also think it should be integrable. $\endgroup$ – Boby Jan 5 '17 at 22:50
1
$\begingroup$

We have $$\left| \frac{ \left( it+1\right)^\frac{1+it}{1.5} }{ \left( it+1 \right)^{\frac{1+it}{2}}}\right| = |(1+it)^{\frac{1+it}{6}}|= \left|\exp\left(\left(\frac{1+it}{6}\right)\ln(1+it)\right)\right|.$$ Remember that $|e^z|=e^{\Re(z)}$ and $$\Re\left(\left(\frac{1+it}{6}\right)\ln(1+it)\right)=\frac{\ln(\sqrt{1+t^2})}{6}-\frac{t\arg(1+it)}{6}.$$ (Here we have used the definition of the complex logarithm $\ln(z) = \ln|z|+i\arg(z)$.) Hence $$g(t) = (1+t^2)^{1/12} e^{-\arg(1+it)t/6}.$$ If $t>0$ is big enough, we can assume that $\arg(1+it)>\pi/4$, hence $$g(t) < (1+t^2)^{1/12} e^{-\frac{\pi t}{24}}$$ is clearly integrable at $+\infty.$ You can procede in a same way for $-\infty.$

$\endgroup$
  • $\begingroup$ Can you give more explanation on how you deal with $e^{-\frac{t}{6} {\rm arg}(1+it)}$? $\endgroup$ – Boby Jan 6 '17 at 22:01
  • $\begingroup$ I made a small mistake, now it should be clear. $\endgroup$ – C. Dubussy Jan 6 '17 at 22:18
  • $\begingroup$ Question: Does strings approximation for gamma function hold for complex domain? $\endgroup$ – Boby Jan 6 '17 at 22:48
  • $\begingroup$ Yes, if $\Re(z)>0$, you can see here for ex. en.wikipedia.org/wiki/Stirling's_approximation $\endgroup$ – C. Dubussy Jan 6 '17 at 23:43
  • $\begingroup$ So, you think integrability is not effect by Striling's approximation? In a sense that the error also integrable? $\endgroup$ – Boby Jan 6 '17 at 23:47

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.