1
$\begingroup$

In this question, I went quite a bit over the top, so I now tried to rephrase it in a much simpler way.

Take $a \in \mathbb{R}, s \in \mathbb{C}$ and:

$$\displaystyle C(s,a) := \prod_{n=1}^\infty \left(1- \frac{s}{a + n i} \right) \left(1- \frac{s}{{a - n i}} \right) = \frac{\xi_{c}(0 -a+s)}{\xi_{c}(0-a)}$$

with $\xi_{c}(z) = \frac{\sinh(\pi z)}{z}$.

And then also:

$$\displaystyle R(s,a) := \prod_{n=1}^\infty \left(1- \frac{s}{a + n i} \right) \left(1- \frac{s}{{1-(a + n i)}} \right) = \frac{\xi_{r}(0 -a+s)}{\xi_{r}(0-a)}\frac{\xi_{r}(1 -a-s)}{\xi_{r}(1-a)}$$

with $\xi_{r}(z) = \frac{1}{z\, \Gamma(z \, i)}$.

Questions:

For the infinite product of the conjugated zeros $C(s,a)$, the closed form has only two factors, however the closed form for the product of the reflexive zeros $R(s,a)$ does require four factors.

1) Is there a way to also reduce the closed form for $R(s,a)$ to two $\xi$-factors by altering $\xi_r(z)$?

Additional clarification:

I do believe that it is impossible to reduce these four factors:

$$\displaystyle \frac{\xi_{r}(0 -a+s)}{\xi_{r}(0-a)}\frac{\xi_{r}(1 -a-s)}{\xi_{r}(1-a)}$$

into a division of only two factors.

The most logical extension of $\xi_{r}(z)$ would be to assume: $\xi_{r}(z) = \frac{1}{z\, \Gamma(z \, i)} \, \, \frac{1}{(1-z)\, \Gamma((1-z) \, i)}$

and then to reduce the division to: $\displaystyle \frac{\xi_{r}(0-a+s)}{\xi_{r}(0-a)}$.

However this only correctly induces the original four factors for the nominators only and it gives the wrong results for the two denominators. Any tweak I tried to make to $\xi_r(z)$ fails on this apparently fundamental conflict, hence my conjecture that it is impossible. Keen to learn if this could be proven and/or which branch of Mathematics deals with this type of problems.

2) If not (per my conjecture), is this four factor closed form with $\xi_r(z)$ then dependent on the choice of $n$ or does it simply originate from the reflexive nature of the zeros in the product? To illustrate what I am after with an example; the relations $C(s,a) = C(2a-s,a)$ and $R(s,a) = R(1-s,a)$ are fully independent of the choice of $n$ and the product being finite or infinite, and just stem from the conjugated/reflexive relation between the zeros.

Thanks.

$\endgroup$

1 Answer 1

0
$\begingroup$

Too long for comment Think I made some progress, but not sure how to formalize this further. It seems that also for the individual factors $C(s,a,n)$ and $R(s,a,n)$, the closed forms are structured in exactly the same way as the full infinite product:

$C(s,a,n) := \left(1- \frac{s}{a + n i} \right) \left(1- \frac{s}{{a - n i}} \right) = \frac{\xi_{c n}(0 -a+s,n)}{\xi_{c n}(0-a,n)}$

with $\xi_{c n}(z,n) = z^2 + n^2$.

$R(s,a,n) := \left(1- \frac{s}{a + n i} \right) \left(1- \frac{s}{{1-(a + n i)}} \right) = \frac{\xi_{r n}(0 -a+s,n)}{\xi_{r n}(0-a,n)}\frac{\xi_{r n}(1 -a-s,n)}{\xi_{r n}(1-a,n)}$

with $\xi_{r n}(z,n) = z - n i$.

You can now obviously pick any $n$ you like (as long as $a \pm n i \ne 0, 1- (a+n i) \ne 0$), the structure remains the same. So does this imply that the structure of the closed forms is independent of $n$?

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.