The next talk is by Krishnan Rajkumar. This talk will be recorded but not available online immediately.
Talk Announcement:
Title: The Binet function and telescoping continued fractions
Speaker: Krishnan Rajkumar (Jawaharlal Nehru University (JNU))
When: Thursday, Feb 24, 2022 - 4:00 PM - 5:00 PM (IST)
Where: Zoom. Please write to the organizers for the link.
Tea or Coffee: Please bring your own.
Abstract:
The Binet function $J(z)$ defined by the equation $\Gamma(z) = \sqrt{2 \pi} z^{z-\frac{1}{2}}e^{-z} e^{J(z)}$ is a well-studied function. The Stirling approximation comes from the property $J(z) \rightarrow 0$ as $z\rightarrow \infty$, $|arg z|<\pi$. In fact, an asymptotic expansion $J(z) \sim z^{-1} \sum_{k=0}^{\infty} c_k z^{-2k}$ holds in this region, with closed form expressions for $c_k$ and explicit integrals for the error term for any finite truncation of this asymptotic series.
In this talk, we will discuss two different classical directions of research. The first is exemplified by the work of Robbins (1955) and Cesaro (1922), and carried forward by several authors, the latest being Popov (2018), where elementary means are used to find rational lower and upper bounds for $J(n)$ which hold for all positive integers $n$. All of these establish inequalities of the form $J(n)-J(n+1) > F(n)-F(n+1)$ for an appropriate rational function $F$ to derive the corresponding lower bounds by telescoping.
The second direction is to use moment theory to derive continued fractions of specified forms for $J(x)$. For instance, a modified S-fraction of the form $\frac{a_1}{x \ +} \frac{a_2}{x \ +}\frac{a_3}{x \ +} \cdots$ can be formally derived from the above asymptotic expansion using a method called the qd-algorithm. The resulting continued fraction can then be shown to converge to $J(x)$ by the asymptotic properties of $c_k$ and powerful results from moment theory. There are no known closed-form expressions for the $a_k$.
We will then outline what we call the method of telescoping continued fractions to extend the elementary methods of the first approach to derive the modified S-fraction for $J(x)$ obtained in the second by a new algorithm. We will describe several results that we can prove and some conjectures that together enhance our understanding of the numbers $a_k$ as well as provide upper and lower bounds for $J(x)$ that improve all known results.
This is joint work with Gaurav Bhatnagar.
Gaurav Bhatnagar (Ashoka), Atul Dixit (IIT, Gandhinagar) and Krishnan Rajkumar (JNU)
%5EB "q \le x/ (\log x)^B")
