16.1 Special Notation16.3 Derivatives and Contiguous Functions

§16.2 Definition and Analytic Properties

Contents

§16.2(i) Generalized Hypergeometric Series

Throughout this chapter it is assumed that none of the bottom parameters b_{1}, b_{2}, \dots, b_{q} is a nonpositive integer, unless stated otherwise. Then formally

16.2.1\mathop{{{}_{{p}}F_{{q}}}\/}\nolimits\!\left({a_{1},\dots,a_{p}\atop b_{1},\dots,b_{q}};z\right)=\sum _{{k=0}}^{\infty}\frac{\left(a_{1}\right)_{{k}}\cdots\left(a_{p}\right)_{{k}}}{\left(b_{1}\right)_{{k}}\cdots\left(b_{q}\right)_{{k}}}\frac{z^{k}}{k!}.

Equivalently, the function is denoted by \mathop{{{}_{{p}}F_{{q}}}\/}\nolimits\!\left({\mathbf{a}\atop\mathbf{b}};z\right) or \mathop{{{}_{{p}}F_{{q}}}\/}\nolimits\!\left(\mathbf{a};\mathbf{b};z\right), and sometimes, for brevity, by \mathop{{{}_{{p}}F_{{q}}}\/}\nolimits\!\left(z\right).

§16.2(ii) Case p\leq q

When p\leq q the series (16.2.1) converges for all finite values of z and defines an entire function.

§16.2(iii) Case p=q+1

Suppose first one or more of the top parameters a_{j} is a nonpositive integer. Then the series (16.2.1) terminates and the generalized hypergeometric function is a polynomial in z.

If none of the a_{j} is a nonpositive integer, then the radius of convergence of the series (16.2.1) is 1, and outside the open disk |z|<1 the generalized hypergeometric function is defined by analytic continuation with respect to z. The branch obtained by introducing a cut from 1 to +\infty on the real axis, that is, the branch in the sector |\mathop{\mathrm{ph}\/}\nolimits\!\left(1-z\right)|\leq\pi, is the principal branch (or principal value) of \mathop{{{}_{{q+1}}F_{{q}}}\/}\nolimits\!\left(\mathbf{a};\mathbf{b};z\right); compare §4.2(i). Elsewhere the generalized hypergeometric function is a multivalued function that is analytic except for possible branch points at z=0,1, and \infty. Unless indicated otherwise it is assumed that in the DLMF generalized hypergeometric functions assume their principal values.

On the circle |z|=1 the series (16.2.1) is absolutely convergent if \realpart{\gamma _{q}}>0, convergent except at z=1 if -1<\realpart{\gamma _{q}}\leq 0, and divergent if \realpart{\gamma _{q}}\leq-1, where

16.2.2\gamma _{q}=(b_{1}+\dots+b_{q})-(a_{1}+\dots+a_{{q+1}}).

§16.2(iv) Case p>q+1

Polynomials

In general the series (16.2.1) diverges for all nonzero values of z. However, when one or more of the top parameters a_{j} is a nonpositive integer the series terminates and the generalized hypergeometric function is a polynomial in z. Note that if -m is the value of the numerically largest a_{j} that is a nonpositive integer, then the identity

16.2.3\mathop{{{}_{{p+1}}F_{{q}}}\/}\nolimits\!\left({-m,\mathbf{a}\atop\mathbf{b}};z\right)=\frac{\left(\mathbf{a}\right)_{{m}}(-z)^{m}}{\left(\mathbf{b}\right)_{{m}}}\mathop{{{}_{{q+1}}F_{{p}}}\/}\nolimits\!\left({-m,1-m-\mathbf{b}\atop 1-m-\mathbf{a}};\frac{(-1)^{{p+q}}}{z}\right)

can be used to interchange p and q.

Note also that any partial sum of the generalized hypergeometric series can be represented as a generalized hypergeometric function via

16.2.4\sum _{{k=0}}^{m}\frac{\left(\mathbf{a}\right)_{{k}}}{\left(\mathbf{b}\right)_{{k}}}\frac{z^{k}}{k!}=\frac{\left(\mathbf{a}\right)_{{m}}z^{m}}{\left(\mathbf{b}\right)_{{m}}m!}\mathop{{{}_{{q+2}}F_{{p}}}\/}\nolimits\!\left({-m,1,1-m-\mathbf{b}\atop 1-m-\mathbf{a}};\frac{(-1)^{{p+q+1}}}{z}\right).

Non-Polynomials

See §16.5 for the definition of \mathop{{{}_{{p}}F_{{q}}}\/}\nolimits\!\left(\mathbf{a};\mathbf{b};z\right) as a contour integral when p>q+1 and none of the a_{k} is a nonpositive integer. (However, except where indicated otherwise in the DLMF we assume that when p>q+1 at least one of the a_{k} is a nonpositive integer.)

§16.2(v) Behavior with Respect to Parameters

Let

16.2.5\mathop{{{}_{{p}}{\mathbf{F}}_{{q}}}\/}\nolimits\!\left(\mathbf{a};\mathbf{b};z\right)=\ifrac{\mathop{{{}_{{p}}F_{{q}}}\/}\nolimits\!\left({a_{1},\dots,a_{p}\atop b_{1},\dots,b_{q}};z\right)}{\left(\mathop{\Gamma\/}\nolimits\!\left(b_{1}\right)\cdots\mathop{\Gamma\/}\nolimits\!\left(b_{q}\right)\right)}=\sum _{{k=0}}^{\infty}\frac{\left(a_{1}\right)_{{k}}\cdots\left(a_{p}\right)_{{k}}}{\mathop{\Gamma\/}\nolimits\!\left(b_{1}+k\right)\cdots\mathop{\Gamma\/}\nolimits\!\left(b_{q}+k\right)}\frac{z^{k}}{k!};

compare (15.2.2) in the case p=2, q=1. When p\leq q+1 and z is fixed and not a branch point, any branch of \mathop{{{}_{{p}}{\mathbf{F}}_{{q}}}\/}\nolimits\!\left(\mathbf{a};\mathbf{b};z\right) is an entire function of each of the parameters a_{1},\dots,a_{p},b_{1},\dots,b_{q}.