9
$\begingroup$

Let $K = \mathrm{SU}(2) = \{ k[\alpha ,\beta] \mid \alpha ,\beta \in \mathbb{C}, |\alpha |^2 + |\beta |^2 = 1 \} $ with $$ k [ \alpha , \beta ] = \begin{pmatrix} \alpha & \beta \\ - \overline{\beta} & \overline{\alpha} \end{pmatrix}$$ Let $V_n$ denote the space of polynomials in one variable of degree at most $n$. We choose the set $\{ z^{l-q} : q\in\mathbb{Z},\ |q|\leq l\}$ as a basis for $V_{2l}$. Then the (irreducible and unitary) representations of $K$ are given by the formula $$ \sigma_l(k [\alpha , \beta]) z^{l-q} = ( \alpha z - \overline{\beta} )^{l-q} ( \beta z + \overline{\alpha})^{l+q} .$$ Let $ \Phi_{p,q}^l $ denote the coefficient of $z^{l-p}$ in the polynomial expansion of $\sigma_l(k [\alpha , \beta]) z^{l-q}$. i.e, $$ \sum_{|p|\leq l} \Phi_{p,q}^l(k[\alpha,\beta]) z^{l-p} = ( \alpha z - \overline{\beta} )^{l-q} ( \beta z + \overline{\alpha})^{l+q} ,\quad for \ |q|\leq l. $$

There is an explicit formula of $$ \int_K \Phi_{p,q}^l(k) \overline{\Phi_{p_1,q_1}^{l_1}(k)} dk$$ in the book Sum Formula for SL2 over Imaginary Quadratic Number Fields by Lokvenec-Guleska.

I have to solve the triple product $$ \int_K \Phi_{p,q}^l(k) \Phi_{p_1,q_1}^{l_1}(k) \Phi_{p_2,q_2}^{l_2}(k) dk.$$

I know how to compute this equation through brute force. However, my advisor doesn’t want me to include the complex calculation process in my paper. Is there a place where I can directly find this formula? Any suggestion would be helpful. Thanks!


In fact I have obtained the following formula: For any $a,b,c,d \in\mathbb{Z}$ and $k = k[\alpha,\beta]$, we have $$\int_K \alpha^{a} \bar{\alpha} ^{b} \beta^{c} \bar{\beta} ^{d} d k = \delta ( a=b, c=d ) \frac{a!c!}{(a+c+1)!},$$ where $\delta$ denotes the Kronecker symbol.

$\endgroup$
6
  • 1
    $\begingroup$ The corresponding post on Mathematics: Triple Product Formula on $K = SU(2)$. On meta, you can find some basic advice about cross-posting - it is recommended to provide at least a link. $\endgroup$ Commented Nov 4, 2023 at 7:30
  • 1
    $\begingroup$ Off the top of my head, I don't know of a specific source, but have you tried looking up the literature on Weingarten calculus, and does it help? e.g. the expository article of Collins-Matsumoto-Novak arxiv.org/abs/2109.14890 $\endgroup$
    – Yemon Choi
    Commented Nov 4, 2023 at 14:44
  • 1
    $\begingroup$ In physicists language, aren’t you trying to compute triple products of Wigner D-matrices ? Or am I missing something ? $\endgroup$ Commented Nov 4, 2023 at 19:14
  • $\begingroup$ @DavidLehavi I haven't heard of that thing ... $\endgroup$ Commented Nov 4, 2023 at 19:30
  • $\begingroup$ @YemonChoi Thank you for your suggestion! However this article seems to be unrelated to my field of expertise, I might need some time to understand it. (╥﹏╥) $\endgroup$ Commented Nov 4, 2023 at 19:33

1 Answer 1

2
$\begingroup$

This is one of those questions like

Determinant of the random matrix $X^2+Y^2$

which are extremely easy to do if one is familiar with graphical/diagrammatic algebra, but would otherwise lead to rather cumbersome calculations.

The rules for computing $SU(2)$ or more general $SU(N)$ normalized Haar integrals were explained in my two answers to MO question

How to constructively/combinatorially prove Schur-Weyl duality?

In the present situation, the triple product integral, should be some sign and harmless ratio of factorials times a product of two Wigner $3jm$ symbols $$ \left(\begin{array}{ccc} l_1 & l_2 & l\\ p_1 & p_2 & -(p_1+p_2) \end{array}\right) \times \left(\begin{array}{ccc} l_1 & l_2 & l\\ q_1 & q_2 & -(q_1+q_2) \end{array}\right) $$

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .