pub fn hyp3f2(
a1: f64,
a2: f64,
a3: f64,
b1: f64,
b2: f64,
z: f64,
) -> Option<f64>Expand description
Generalized hypergeometric function $_3F_2$ for real arguments
This function is not available in SciPy.
§Notes
For $\abs z < 1$,
$$ \hyp 3 2 {a_1\enspace a_2\enspace a_3} {b_1\enspace b_2} z = \sum_{n=0}^\infty {\rpow{a_1}{n} \ \rpow{a_2}{n} \ \rpow{a_3}{n} \over \rpow{b_1}{n} \ \rpow{b_2}{n}} {z^n \over n!} , $$
with the rising factorial $\rpow \square n$ (pow_rising). For
$z < -1$, it is the analytic continuation 1 2. A non-positive integer $a_i = -n$ makes it
a polynomial of degree at most $n$, defined for all $z$.
As in hyp2f1, an $a_i$ equal to a $b_j$ cancels it, also if it is a
non-positive integer, so that the result is the $_2F_1$ of the other parameters. Where that
converges at $z = 1$, it is Gauss’s sum 3
$$ \hyp 2 1 {a_1\enspace a_2} b 1 = {\Gamma(b) \ \Gamma(b - a_1 - a_2) \over \Gamma(b - a_1) \ \Gamma(b - a_2)} . $$
It is $+\infty$:
- if a $b_j$ is a non-positive integer $-m$, unless it is a polynomial of degree $\lt m$;
- at $z = 1$ if $b_1 + b_2 \le a_1 + a_2 + a_3$, unless it is a polynomial.
It is NaN for $z > 1$, where it is complex, unless it is a polynomial, and if an argument is
NaN. For an infinite argument, it is the limit as in hyppfq.
§Returns None
If the relative error could exceed about $10^{-13}$:
- at $z = 1$, unless a parameter cancels so that Gauss’s sum applies, and for $1 - 10^{-5} \lesssim z < 1$;
- for $z < -1$ if two $a_i$ differ by an integer: often for $z < -10$, and nearly always for $z \lesssim -10^6$;
- if the terms cancel too much (e.g. large parameters), overflow or underflow;
- for some infinite arguments, as for
hyppfq.
§See also
hyp2f1: Gauss’ hypergeometric function, $\hyp 2 1 {a_1\enspace a_2} b z$hyp1f1: Kummer’s confluent hypergeometric function, $\hyp 1 1 a b z$
NIST Digital Library of Mathematical Functions https://dlmf.nist.gov/16.2 ↩
NIST Digital Library of Mathematical Functions https://dlmf.nist.gov/16.8.E8 ↩
NIST Digital Library of Mathematical Functions https://dlmf.nist.gov/15.4.E20 ↩