Logarithmic derivative of Polygamma functions

469 Views Asked by At

While studying Gamma function and related functions I noticed that its logarithmic derivative (the so-called Digamma function) is studied more than its "normal" derivative but on the other hand I can't find anything on the logarithmic derivative of the Digamma function but only its usual derivatives (the so-called Polygamma functions).

So my question is: have already been studied the logarithmic derivatives of the Polygamma functions and they are not interesting (and maybe why they are not?) or none simply ever cared about ?

2

There are 2 best solutions below

0
On BEST ANSWER

The $\Gamma$ function is defined as the analytic continuation (through $\Gamma(z+1)=z\,\Gamma(z)$) of $$ \Gamma(\alpha)=\int_{0}^{+\infty} t^{\alpha-1}e^{-t}\,dt,\qquad\text{Re}(\alpha)>0 $$ so it is a meromorphic function over $\mathbb{C}$ with simple poles at $z=0,-1,-2,\ldots$.

By the Mittag-Leffler theorem or through other ways (for instance, the dominated convergence theorem that leads to the Euler product) we have that $\Gamma(z+1)$ can be represented by an infinite product: $$ \Gamma(z+1) = e^{-\gamma z}\prod_{n\geq 1}\left(1+\frac{z}{n}\right)^{-1}e^{\frac{z}{n}}\tag{1}$$ that is the Weierstrass product. Hence $\log\Gamma $ can be represented by a series, as well as: $$ \psi(z+1) = \frac{d}{dz}\log\Gamma(z+1) = -\gamma+\sum_{n\geq 1}\left(\frac{1}{n}-\frac{1}{n+z}\right).\tag{2}$$ We also have an integral representation: $$ \psi(z+1)=-\gamma+\int_{0}^{1}\frac{1-t^z}{1-t}\,dt,\qquad \text{Re}(z)>-1\tag{3}$$ but the functional equation $\psi(z+1)=\frac{1}{z}+\psi(z)$ does not lead to a practical representation of the digamma function as a product. That is the reason for which we never met the logarithmic derivative of the digamma function: despite it just is $\frac{\psi'}{\psi}$, $\frac{\psi'}{\psi}$ does not satisfy a nice functional equation.

0
On

Your question is founded upon wrong assumptions. The reason for studying the derivative of the natural logarithm of the $\Gamma$ function is because factorials show an explosive growth, which far outweighs even that of the exponential function itself, so, in an era before the advent of modern computers, the natural logarithm of the $\Gamma$ function was studied as a function in its own right, hence the historic definition of the digamma function. But since $\psi(n+1)=H_n-\gamma$, “taming” such a incredibly-slow-increasing function by taking its logarithm makes little or no sense.