2

I want to calculate 12.1!!!!!! , Just for curiosity. (One of my friend texted the term to me for some complex reason.)

I searched for multifactorial in terms of gamma function or equation, and found nothing if factorial goes more than double. I know the answer is somewhere between 12*6 and 13*7*1, but can I be more accurate than linearly interpolating those values?

(By the way, I spend an hour to learn thatt 12.1!!≠(12.1!)!. I wonder how huge the answer will be if it was 6 repetitive factorial, but not for now.)

B.So
  • 21
  • How do you define the multifactorial for non-integers ? – Peter May 20 '16 at 12:09
  • Since the notation is ambiguous, I think you should first ask your friend what he means, if three consecutive applications of the double factorial (its alternative form) or six consecutive applications of the factorial (i.e. $\Gamma(x+1)$). Or something else entirely. By the way, $\log_{10}((12!)!)\approx 10^{9.6}$, so you might want to question your friend's "complicated reasons". –  May 20 '16 at 12:18
  • 3
    You should read more carefully: According to your Wikipedia page https://en.wikipedia.org/wiki/Factorial#Alternative_extension_of_the_multifactorial it would be with $k=6, z=12.1$ $$k^{(z-1)/k}\frac{\Gamma(1+z/k)}{\Gamma(1+1/k)}\approx 60.24157$$ – gammatester May 20 '16 at 12:20
  • 2
    @gammatester That's not true. The formula you used is only true for integer values. Mathworld (http://mathworld.wolfram.com/DoubleFactorial.html) gives the true double factorial formula for all complex numbers: $z!!=2^{(1+2z-\cos(\pi z))/4}\pi^{(\cos(\pi z)-1)/4}\Gamma(1+\frac{1}{2}z)$ which follows the relation $z!=z!!(z-1)!!$ for all complex numbers – Jacob May 20 '16 at 12:43

1 Answers1

2

Some years ago I calculated the equation of the multifactorial:

$$z!_{(\alpha)}=\alpha^{\frac{z}{\alpha}}\Gamma\left(1+\frac{z}{\alpha}\right)\prod_{j=1}^{\alpha-1}\left(\frac{\alpha^{\frac{\alpha-j}{\alpha}}}{\Gamma\left(\frac{j}{\alpha}\right)}\right)^{C_{\alpha}(z-j)}$$ Where $$C_{\alpha}(z)=\frac{1}{\alpha}\left(1+2\sum_{k=1}^{\left\lfloor\frac{\alpha-1}{2}\right\rfloor}\cos\left(\frac{2k\pi}{\alpha}z\right)+\text{mod}(\alpha-1,2)\cos(\pi z)\right)$$

Alternative representation of $C_{\alpha}(z)$
Let $U_n(z)$ the Chebyshev polynomial of the second kind

  • $\displaystyle C_{\alpha}\left(x\right)=\frac{1}{\alpha}U_{\alpha-1}\left(\cos\left(\frac{\pi x}{\alpha}\right)\right)\cos\left(\frac{\pi x}{\alpha}\right)^{\text{mod}\left(\alpha-1,2\right)}$
  • $\displaystyle C_{\alpha}(x)=\frac{1}{\alpha}\sum_{k=1}^{\alpha}\cos\left(\theta_{k}\ x\right)\qquad\text{ where }\theta_{k}=\arccos\left(\cos\left(\frac{2k\pi}{\alpha}\right)\right)$

This function respects these relationship:

$\displaystyle\bullet\; z!=\prod_{i=0}^{\alpha-1}(z-i)!_{(\alpha)}$
$\displaystyle\bullet\; z!_{(\alpha)}=z(z-\alpha)!_{(\alpha)}$

Also corresponds to the known cases in which $\alpha=1$ and $\alpha=2$:


For $\color{red}{\alpha=1}$ the function is:

$$z!_{(1)}=1^{\frac{z}{1}}\Gamma\left(1+\frac{z}{1}\right)\prod_{j=1}^{0}\left(\frac{1^{\frac{1-j}{\alpha}}}{\Gamma\left(\frac{j}{1}\right)}\right)^{C_{1}(z-j)}$$ $$\color{blue}{z!_{(1)}=\Gamma(1+z)}$$

Pay attention to the fact that $$\prod_{i=1}^{0}f(z):=1\text{ for every function, you also have that}\sum_{i=1}^{0}f(z):=0$$


For $\color{red}{\alpha=2}$ the function is:

$$z!_{(2)}=2^{\frac{z}{2}}\Gamma\left(1+\frac{z}{2}\right)\prod_{j=1}^{2-1}\left(\frac{2^{\frac{2-j}{2}}}{\Gamma\left(\frac{j}{2}\right)}\right)^{C_{2}(z-j)}=2^{\frac{z}{2}}\Gamma\left(1+\frac{z}{2}\right)\left(\frac{2^{\frac{2-1}{2}}}{\Gamma\left(\frac{1}{2}\right)}\right)^{C_{2}(z-1)}$$

$$C_2(z)=\frac{1+\cos(\pi z)}{2}\quad\Rightarrow\quad C_2(z-1)=\frac{1-\cos(\pi z)}{2}\qquad \Gamma\left(\frac{1}{2}\right)=\sqrt{\pi}$$

$$z!_{(2)}=2^{\frac{z}{2}}\Gamma\left(1+\frac{z}{2}\right)\left(\frac{\sqrt{2}}{\sqrt{\pi}}\right)^{\frac{1-\cos(\pi z)}{2}}$$ $$\color{blue}{z!_{(2)}=2^{\frac{z}{2}}\Gamma\left(1+\frac{z}{2}\right)\left(\frac{2}{\pi}\right)^{\frac{1-\cos(\pi z)}{4}}}$$


For $\color{red}{\alpha=3}$ the function is:

$$\color{blue}{z!_{(3)}=\frac{3^{\frac{z}{3}}\Gamma\left(1+\frac{z}{3}\right)}{\Gamma\left(\frac{1}{3}\right)^{\frac{2}{\sqrt{3}}\sin\left(\frac{2\pi z}{3}\right)}}\frac{3^{\frac{1}{2}-\frac{1}{2}\cos\left(\frac{2\pi z}{3}\right)-\frac{\sqrt{3}}{18}\sin\left(\frac{2\pi z}{3}\right)}}{\left(2\pi\right)^{\frac{1}{3}-\frac{2}{3}\cos\left(\frac{2\pi z}{3}-\frac{\pi}{3}\right)}}}$$


For $\color{red}{\alpha=4}$ the function is: $$\color{blue}{z!_{(4)}=2^{\frac{z}{2}}\frac{\Gamma\left(1+\frac{z}{4}\right)}{\Gamma\left(\frac{1}{4}\right)^{\sin\left(\frac{\pi z}{2}\right)}}\frac{\pi^{\frac{\cos(\pi z)-3}{8}+\frac{1}{2}\sin\left(\frac{\pi z}{2}\right)+\frac{1}{4}\cos\left(\frac{\pi z}{2}\right)}}{2^{\frac{\cos(\pi z)-5}{8}-\frac{3}{4}\sin\left(\frac{\pi z}{2}\right)+\frac{1}{2}\cos\left(\frac{\pi z}{2}\right)}}}$$


For $\color{red}{\alpha=5}$ the formula is:

$$\color{blue}{z!_{(5)}=5^{\frac{z+z_{0}(z)}{5}}\cdot\frac{\Gamma\left(1+\frac{z}{5}\right)}{\Gamma\left(\frac{1}{5}\right)^{\frac{\gamma_{1}(z)}{5}}\Gamma\left(\frac{2}{5}\right)^{\frac{\gamma_{2}(z)}{5}}}\cdot\left(2\pi\right)^{\frac{c_{1}(z)}{5}(z)}\phi^{\frac{c_{2}(z)}{5}}}$$

Where:

  • $\phi$ is the golden ratio
  • $z_{0}(z):=\frac{5}{2}-\frac{5}{4}\cos\left(\frac{2\pi z}{5}\right)+\frac{1}{4}\sqrt{5+\frac{2}{\sqrt{5}}}\sin\left(\frac{2\pi z}{5}\right)-\frac{5}{4}\cos\left(\frac{4\pi z}{5}\right)+\frac{1}{4}\sqrt{5-\frac{2}{\sqrt{5}}}\sin\left(\frac{4\pi z}{5}\right)$
  • $\gamma_{1}(z):=\sqrt{2\left(5+\sqrt{5}\right)}\sin\left(\frac{2\pi z}{5}\right)+\sqrt{2\left(5-\sqrt{5}\right)}\sin\left(\frac{4\pi z}{5}\right)$
  • $\gamma_{2}(z):=\sqrt{2\left(5-\sqrt{5}\right)}\sin\left(\frac{2\pi z}{5}\right)-\sqrt{2\left(5+\sqrt{5}\right)}\sin\left(\frac{4\pi z}{5}\right)$
  • $c_{1}(z):=-2+\cos\left(\frac{2\pi z}{5}\right)+\sqrt{5+2\sqrt{5}}\sin\left(\frac{2\pi z}{5}\right)+\cos\left(\frac{4\pi z}{5}\right)-\sqrt{5-2\sqrt{5}}\sin\left(\frac{4\pi z}{5}\right)$
  • $c_{2}(z):=-\frac{\sqrt{5}}{2}\cos\left(\frac{2\pi z}{5}\right)+\frac{\sqrt{5-2\sqrt{5}}}{2}\sin\left(\frac{2\pi z}{5}\right)+\frac{\sqrt{5}}{2}\cos\left(\frac{4\pi z}{5}\right)+\frac{\sqrt{5+2\sqrt{5}}}{2}\sin\left(\frac{4\pi z}{5}\right)$

For $\color{red}{\alpha=6}$ the function is: $$\color{blue}{\displaystyle z!_{(6)}=\frac{\sqrt[12]{2}\sqrt{3}6^{\frac{z}{6}}\Gamma\left(1+\frac{z}{6}\right)}{\left(\frac{\sqrt[12]{3}}{\sqrt{2\pi}}\Gamma\left(\frac{1}{3}\right)\right)^{\sqrt{3}\sin\left(\frac{\pi z}{3}\right)+\frac{1}{\sqrt{3}}\sin\left(\frac{2\pi z}{3}\right)}}\cdot\frac{2^{\frac{\sqrt{3}}{9}\sin\left(\frac{\pi z}{3}\right)}\left(2\pi\right)^{\frac{1}{6}\cos\left(\frac{2\pi z}{3}\right)}}{3^{\frac{1}{4}\left(\cos\left(\frac{\pi z}{3}\right)+\cos\left(\frac{2\pi z}{3}\right)\right)}\pi^{\frac{5}{12}}}\cdot\left(\frac{\pi}{2}\right)^{\left(\frac{\cos\left(\pi z\right)}{12}+\frac{1}{6}\cos\left(\frac{\pi z}{3}\right)\right)}}$$


For $\color{red}{\alpha=7}$ it is not worth writing the explicit formula since the goniometric functions with argument $\frac{k\pi}{7}$ with $k\in\mathbb{Z}\setminus 7\mathbb{Z}$ cannot be expressed in simpler terms. You can use the definition in the last part of the answer.


For $\color{red}{\alpha=8}$ the function is:

$$\color{blue}{z!_{(8)}=8^{\frac{z+z_0(z)}{8}}\cdot\frac{\Gamma\left(1+\frac{z}{8}\right)}{\Gamma\left(\frac{1}{8}\right)^{\gamma_{1}(z)}\Gamma\left(\frac{1}{4}\right)^{\gamma_{2}(z)}}\pi^{c_{1}(z)}\left(\frac{1}{2}\right)^{c_{2}(z)}\left(\sqrt{2}+1\right)^{c_{3}(z)}}$$

Where

  • $z_{0}(z):=\frac{7}{2}-\cos\left(\frac{\pi z}{4}\right)+\left(1+\sqrt{2}\right)\sin\left(\frac{\pi z}{4}\right)-\cos\left(\frac{\pi z}{2}\right)+\sin\left(\frac{\pi z}{2}\right)-\cos\left(\frac{3\pi z}{4}\right)+\left(\sqrt{2}-1\right)\sin\left(\frac{3\pi z}{4}\right)-\frac{\cos\left(\pi z\right)}{2}$
  • $\gamma_{1}(z):=\frac{\sqrt{2}}{2}\left(\sin\left(\frac{\pi}{4}z\right)+\sin\left(\frac{3\pi}{4}z\right)\right)$
  • $\gamma_{2}(z):=\frac{2-\sqrt{2}}{4}\sin\left(\frac{\pi}{4}z\right)+\frac{1}{2}\sin\left(\frac{\pi z}{2}\right)-\frac{2+\sqrt{2}}{4}\sin\left(\frac{3\pi z}{4}\right)$
  • $c_{1}(z):=-\frac{7}{16}+\frac{1}{8}\cos\left(\frac{\pi z}{4}\right)+\frac{2+\sqrt{2}}{8}\sin\left(\frac{\pi z}{4}\right)+\frac{1}{8}\cos\left(\frac{\pi z}{2}\right)+\frac{1}{4}\sin\left(\frac{\pi z}{2}\right)+\frac{1}{8}\cos\left(\frac{3\pi z}{4}\right)-\frac{2-\sqrt{2}}{8}\sin\left(\frac{3\pi z}{4}\right)+\frac{1}{16}\cos\left(\pi z\right)$
  • $c_{2}(z):=\frac{1}{4}-\frac{2+3\sqrt{2}}{16}\sin\left(\frac{\pi z}{4}\right)-\frac{1}{8}\cos\left(\frac{\pi z}{2}\right)-\frac{3\sqrt{2}}{16}\sin\left(\frac{3\pi z}{4}\right)+\frac{1}{8}\sin\left(\frac{3\pi z}{4}\right)-\frac{1}{8}\cos\left(\pi z\right)$
  • $c_{3}(z)=-\frac{1}{4}\cos\left(\frac{\pi}{4}z+\frac{\pi}{4}\right)+\frac{1}{4}\cos\left(\frac{3\pi z}{4}-\frac{\pi}{4}\right)$

Etc...


So in your case $$12.1!_{(6)}= 72.7394522427...$$

Bearing in mind that $12!_{(6)}:=12\cdot(12-6)=72$ you can see that the values ​​are also quite similar.


Fourier expansion

If you only need an approximate value you can also use its Fourier expansion:

$$x!_{(\alpha)}\approx \alpha^{\frac{x}{\alpha}}\Gamma\left(1+\frac{x}{\alpha}\right)\sum_{j=1}^{\alpha}\frac{\alpha^{\frac{\alpha-j}{\alpha}}}{\Gamma\left(\frac{j}{\alpha}\right)}C_{\alpha}\left(x-j\right)$$

which is a very similar form to

$$z!_{(\alpha)}=\alpha^{\frac{z}{\alpha}}\Gamma\left(1+\frac{z}{\alpha}\right)\prod_{j=1}^{\alpha-1}\left(\frac{\alpha^{\frac{\alpha-j}{\alpha}}}{\Gamma\left(\frac{j}{\alpha}\right)}\right)^{C_{\alpha}(z-j)}$$

With the difference that:

  1. product from 1 to $\alpha-1$ $\mapsto$ sum from 1 to $\alpha$
  2. power $\mapsto$ product

Using this approximation you get $12.1!_{(6)}\approx 72.8179454437$.

More efficient definition

it is possible to give a more efficient formulation of the multifactorial, which is essentially the one that comes out when all the summations and all the products are carried out (exploiting the symmetry properties of the Gamma function, of the powers and of the goniometric functions):

For brevity, I indicate $\beta:=\left\lfloor\frac{\alpha-1}{2}\right\rfloor$ $$z!_{(\alpha)}=\alpha^{\frac{z+z_0(z)}{\alpha}}\Gamma\left(1+\frac{z}{\alpha}\right)\frac{1}{\pi^{p(z)}}\prod_{j=1}^{\beta}\frac{\sin\left(\frac{j\pi}{\alpha}\right)^{C_{\alpha}(z+j)}}{\Gamma\left(\frac{j}{\alpha}\right)^{\gamma_j(z)}}$$

Where:

  • $\displaystyle C_{\alpha}\left(z\right)=\frac{1}{\alpha}\left(1+2\sum_{k=1}^{\beta}\cos\left(\frac{2k\pi}{\alpha}z\right)+\operatorname{mod}\left(\alpha-1,2\right)\cos\left(\pi z\right)\right)\qquad$ (as defined above)
  • $\displaystyle z_{0}\left(z\right)=\frac{\alpha-1}{2}+\sum_{k=1}^{\beta}\csc\left(\frac{k\pi}{\alpha}\right)\sin\left(\frac{2k\pi}{\alpha}z-\frac{k\pi}{\alpha}\right)-\operatorname{mod}\left(\alpha-1,2\right)\frac{\cos\left(\pi z\right)}{2}$
  • $\displaystyle p\left(z\right)=\frac{\alpha-1}{2\alpha}+\frac{2}{\alpha}\sum_{k=1}^{\beta}\csc\left(\frac{k\pi}{\alpha}\right)\sin\left(\frac{k\pi}{\alpha}\beta\right)\cos\left(\frac{2k\pi}{\alpha}z+\left(\beta+1\right)\frac{k\pi}{\alpha}\right)+\frac{\operatorname{mod}\left(\alpha-1,2\right)}{\alpha}\left(\sum_{k=1}^{\beta}\left(-1\right)^{k}\cos\left(\frac{2k\pi}{\alpha}z\right)-\frac{\cos\left(\pi z\right)}{2}\right)$
  • $\displaystyle \gamma_j\left(z\right)=\frac{4}{\alpha}\sum_{k=1}^{\beta}\sin\left(\frac{2k\pi}{\alpha}j\right)\sin\left(\frac{2k\pi}{\alpha}z\right)$

This formulation is equivalent and immediately returns the "final form". Therefore (except in particular cases where $\Gamma\left(\frac{j}{\alpha}\right)$ or $\sin\left(\frac{j\pi}{\alpha}\right)$ can be written in different terms) there is no need to further develop the formula once $\alpha$ has been set.

Note that in this case half of the products are made and the terms involving the powers of $\pi$ and $\alpha$ are taken out of the production, so there is no longer any need to perform two summations to find the final terms.

  • In the end, I see no differences between what I finally posted and your formula. Note: $\phi=\tau/\sqrt5$ with my $\tau$, $\sqrt{5\pm2\sqrt5}=(\sqrt\tau\pm\sqrt{5/\tau})$. With your integration I have since deleted my answer. – Oscar Lanzi Sep 16 '23 at 08:54
  • @OscarLanzi I honestly don't know, however I updated the answer and put the "final" form for each $\alpha$. So there's no need to do the calculations every time. – Math Attack Sep 16 '23 at 12:58
  • Nice result for $\alpha=8$. We can render the constants as real radicals only if the Euler totient of $\alpha$ is a power of $2$, hence no simplification for $\alpha=7$. Seems strange $\alpha=6$ gives no terms with $\Gamma(1/6)$; is that related 8n some way to $\Gamma(1/3)$? – Oscar Lanzi Sep 16 '23 at 13:42
  • 1
    @OscarLanzi $\Gamma\left(\dfrac{1}{6}\right)=\dfrac{1}{2^{1/3}}\sqrt{\dfrac{3}{\pi}}\Gamma\left(\dfrac{1}{3}\right)^2$ – Math Attack Sep 16 '23 at 14:11