Given complex numbers $a=x+iy$, $b=m+in$ and a gamma function $\Gamma(z)$ with $x\gt0$ and $m\gt0$, it is conjectured that the following continued fraction holds

$$\frac{\displaystyle4\Gamma\left(\frac{2a+3}{4}\right)\Gamma\left(\frac{2b+3}{4}\right)}{\displaystyle\Gamma\left(\frac{2a+1}{4}\right)\Gamma\left(\frac{2b+1}{4}\right)}=\cfrac{(2a+1)(2b+1)}{a+b+2+\cfrac{(a-b+1)(b-a+1)} {a+b+4+\cfrac{(2a+3)(2b+3)}{a+b+6+\cfrac{(a-b+3)(b-a+3)}{a+b+8+\ddots}}}}\tag{1a}$$



Q: How do we prove the continued fraction $(1a)$ rigorously?

  • $\begingroup$ Equation (1b) is wrong. The left side should be $1/\pi.$ For example, $1/(3+1/5)=5/16=0.3125$ while $1/\pi\approx 0.3183.$ $\endgroup$ – Somos May 20 at 19:05
  • $\begingroup$ @Somos : I doubt it is wrong. It likely has a very slow rate of convergence. I'll post some of the reasoning that led me to that conjecture shortly $\endgroup$ – Nicco May 21 at 19:22
  • $\begingroup$ I see the problem. A bit of calculation shows that the left and right side of your first equation multiplied is $4$. Thus, when $a=b=1/2$ the left side gamma quotient is $\pi$ and the right side continued fraction is $4/\pi.$ $\endgroup$ – Somos May 21 at 20:25
  • $\begingroup$ @Somos : Yes you're right, a bit of calculation does indicate that the second continued fraction converges to $\frac{1}{\pi}$ $\endgroup$ – Nicco May 21 at 22:42
  • 1
    $\begingroup$ The third continued fraction in the Wikipedia article for Pi is similar to yours. See the image here. $\endgroup$ – Somos May 27 at 14:07

This can be deduced from Gauss' continued fraction $$\frac{_2F_1(a+1,b;c+1;z)}{_2F_1(a,b;c;z)}=\cfrac{c}{c+\cfrac{(a-c)bz}{c+1+\cfrac{(b-c-1)(a+1)z}{c+2+\cfrac{(a-c-1)(b+1)z}{c+3+\cfrac{(b-c-2)(a+2)z}{c+4+\ddots}}}}}$$ which evaluates the continued fraction in $(1a)$ as $$\frac{(2a+1)(2b+1)}{a+b+2}\frac{_2 F_1\left(b+\frac32,\frac{b-a+1}{2};\frac{a+b}{2}+2;-1\right)}{_2 F_1\left(b+\frac12,\frac{b-a+1}{2};\frac{a+b}{2}+1;-1\right)},$$ and Kummer's formula $_2 F_1(a,b;a-b+1;-1)=\dfrac{\Gamma(a/2+1)\Gamma(a-b+1)}{\Gamma(a+1)\Gamma(a/2-b+1)}$.

After all the necessary substitutions and cancellations, this yields exactly the expected result. The proofs are sketched in the linked articles, but you may want to follow further references for a deeper treatment.

| cite | improve this answer | |

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.