7
$\begingroup$

Dear Math Stackexchange,

I'm a physics researcher working on a problem in quantum statistical physics. I've encountered the following function which I do not recognize (neither does Mathematica):

$$R(z; a,b) =\sum_{k=0}^{\infty} \frac{a^k}{1+z ,円 b^k},$$

where $a$ and $b$ are real parameters, $b>1$ and 0ドル<a<b$.

In practice I need values of this function for $z$ on the real positive semi-axis, but I would like to understand it better first. As far as I see, $R(z; a,b)$ is a meromorphic function with poles within the unit circle and an essential singularity at $z=0$.

Can it be related to some known special function? Are there more efficient ways to compute it rather than to sum the defining series directly? I'd appreciate any advice or hint. Physical context for this problem can be found on Physics SE.

asked Aug 20, 2011 at 13:10
$\endgroup$

3 Answers 3

7
$\begingroup$

At least when $|z|>1,ドル $$ R(z;a,b)=\sum_{k=0}^{+\infty}(-1)^k\frac{b^{k+1}}{b^{k+1}-a}\left(\frac1z\right)^{k+1}. $$ Thus, $R(,円\cdot,円;a,b)$ may be expressed as a $q$-hypergeometric function ${}_2\phi_1$. In the special case when $a=1,ドル $R(z;1,b)=-L_1(-b/z,1/b)$ where $L_1(,円\cdot,円 ;q)$ is the $q$-logarithm.

answered Aug 20, 2011 at 13:40
$\endgroup$
4
  • $\begingroup$ Thanks! This seems to be the way to go (numerically) outside the unit circle, although approaching the $|z|=1$ the convergence naturally slows down. $\endgroup$ Commented Aug 20, 2011 at 13:56
  • $\begingroup$ Yes. With the exception that if $z$ is real and 1ドル\le z\le b$ (the case $z=1$ included) then for every $a$ the series is alternated. Hence $R(z)$ is between two successive partial sums. Don't know if this helps. $\endgroup$ Commented Aug 20, 2011 at 14:07
  • $\begingroup$ For searching purposes: ${}_p \phi_q$ is sometimes referred to as a basic hypergeometric series... $\endgroup$ Commented Aug 21, 2011 at 16:14
  • $\begingroup$ Still struggling with $|z|<1$ though. It is easy to prove that $R(z; a,b) = \frac{1}{1-a} - R(1/z; a,1/b)$ but I have difficulty to link it to a convergent $q$-series. $\endgroup$ Commented Aug 22, 2011 at 13:07
1
$\begingroup$

Completing Didier's solution, the answer for $|z|>1$ in terms of $q$-hypergeometric function is

$$R(z; a,b) = \frac{1}{a-1} \left ( {}_2\phi_1\left [ \begin{matrix} 1/b , ,円 a \\ a/b \end{matrix} ; 1/b, -1/z \right ] -1 \right ) .$$

answered Aug 21, 2011 at 15:46
$\endgroup$
0
$\begingroup$

After getting used a bit to $q$-hypergeometric function , I found a way to re-arrange the series for $|a|<1$ and $z>0$:

If $b>1,ドル $$R(z; a,b) =\frac{1}{1-a}- \sum_{k=0}^{\infty} \frac{a^k}{1+b^{-k} z^{-1}}=\frac{1}{1-a}- {}_2\phi_1\left [ \begin{matrix} -1/z , ,円 1/b \\ -1/(b z) \end{matrix} ; 1/b, a \right ] \frac{z}{1+z} .$$

If $b<1,ドル $$R(z; a,b) = {}_2\phi_1\left [ \begin{matrix} -z , ,円 b \\ -b ,円 z \end{matrix} ; b, a \right ] \frac{1}{1+z} .$$

This covers all the relevant cases for the original physics problem.

answered Aug 22, 2011 at 20:15
$\endgroup$
1
  • $\begingroup$ I haven't taken a good look at basic hypergeometric series, but the usual hypergeometric series allows for transformations like these, which I presume are just the limiting cases as $q\to 1^-$. $\endgroup$ Commented Aug 23, 2011 at 3:49

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.