Following some toughts on Approximation of $\Big[\Gamma(1+x)\Big]^{-1}$ for $0 \leq x \leq 1$ (for the art for art's sake). .

I have the inequality :


Let $0<x<1$ then we have :

$$\Gamma(x^x+1)< \Big(\Gamma(x+1)\Big)^{x^x}$$

The interesting part is near the value $x=1$.

First I rewrite the inequality as :

$$\ln\left(\Gamma(x^x+1)\right)< x^x\ln\left(\Gamma(x+1)\right)$$

Then I introduce the function :

$$f(x)=\ln\left(\Gamma(x^x+1)\right)-x^x\ln\left(\Gamma(x+1)\right)$$ and now I want to differentiate but it gives nothing good .I have tried also simple refinements without success.

My question :

How to prove the claim ?

Thanks in advance !


Not a final answer but too long for a comment.

Hoping that you do not mind, I shall define $$f(x)=x^x\log\left(\Gamma(x+1)\right)-\log\left(\Gamma(x^x+1)\right)$$

Close to $x=1$, we have $$f(x)=\left(\frac{\gamma }{2}+\frac{\pi ^2}{12}-1\right) (1-x)^3+$$ $$\frac{1}{12} (1-x)^4 \left(8-2 \gamma -\pi ^2-4 \psi ^{(2)}(2)\right)+O\left((1-x)^5\right)$$ which matches the function quite decently for $0.8 \leq x \leq 1.0$.

Close to $x=0$, we have $$f(x)=x ((\gamma -1) \log (x)-\gamma )+$$ $$\frac{1}{12} x^2 \left(\left(6 \gamma -\pi ^2\right) \log ^2(x)-12 \gamma \log (x)+\pi ^2\right)+O\left(x^3\right)$$which matches the function quite decently for $0.0 \leq x \leq 0.1$.

Using the first term only, this predicts a maximum value at $$x_*=e^{\frac{1}{\gamma -1}}\approx 0.0939237$$ and $f(x_*)=0.0330039$ while a rigorous optimization would give a maximum value of $0.0330818$ at $x=0.103099$.


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.