Digital Library of Mathematical Functions
About the Project
NIST
10 Bessel FunctionsBessel and Hankel Functions

§10.20 Uniform Asymptotic Expansions for Large Order

Contents

§10.20(i) Real Variables

Define \zeta=\zeta(z) to be the solution of the differential equation

that is infinitely differentiable on the interval 0<z<\infty, including z=1. Then

10.20.3\frac{2}{3}(-\zeta)^{{\frac{3}{2}}}=\int_{1}^{z}\frac{\sqrt{t^{2}-1}}{t}dt=%
\sqrt{z^{2}-1}-\mathop{\mathrm{arcsec}\/}\nolimits z,1\leq z<\infty,

all functions taking their principal values, with \zeta=\infty,0,-\infty, corresponding to z=0,1,\infty, respectively.

uniformly for z \in(0,\infty) in all cases, where \mathop{\mathrm{Ai}\/}\nolimits and \mathop{\mathrm{Bi}\/}\nolimits are the Airy functions (§9.2).

In the following formulas for the coefficients A_{k}(\zeta), B_{k}(\zeta), C_{k}(\zeta), and D_{k}(\zeta), u_{k}, v_{k} are the constants defined in §9.7(i), and U_{k}(p), V_{k}(p) are the polynomials in p of degree 3k defined in §10.41(ii).

Interval 0<z<1

10.20.10A_{k}(\zeta)=\sum_{{j=0}}^{{2k}}(\tfrac{3}{2})^{j}v_{j}\zeta^{{-3j/2}}U_{{2k-j%
}}\left((1-z^{2})^{{-\frac{1}{2}}}\right),
10.20.11B_{k}(\zeta)=-\zeta^{{-\frac{1}{2}}}\sum_{{j=0}}^{{2k+1}}(\tfrac{3}{2})^{j}u_{%
j}\zeta^{{-3j/2}}U_{{2k-j+1}}\left((1-z^{2})^{{-\frac{1}{2}}}\right),
10.20.12C_{k}(\zeta)=-\zeta^{{\frac{1}{2}}}\sum_{{j=0}}^{{2k+1}}(\tfrac{3}{2})^{j}v_{j%
}\zeta^{{-3j/2}}V_{{2k-j+1}}\left((1-z^{2})^{{-\frac{1}{2}}}\right),
10.20.13D_{k}(\zeta)=\sum_{{j=0}}^{{2k}}(\tfrac{3}{2})^{j}u_{j}\zeta^{{-3j/2}}V_{{2k-j%
}}\left((1-z^{2})^{{-\frac{1}{2}}}\right).

Interval 1<z<\infty

In formulas (10.20.10)–(10.20.13) replace \zeta^{{\frac{1}{2}}}, \zeta^{{-\frac{1}{2}}}, \zeta^{{-3j/2}}, and (1-z^{2})^{{-\frac{1}{2}}} by -i(-\zeta)^{{\frac{1}{2}}}, i(-\zeta)^{{-\frac{1}{2}}}, i^{{3j}}(-\zeta)^{{-3j/2}}, and i(z^{2}-1)^{{-\frac{1}{2}}}, respectively.

Note: Another way of arranging the above formulas for the coefficients A_{k}(\zeta),B_{k}(\zeta),C_{k}(\zeta), and D_{k}(\zeta) would be by analogy with (12.10.42) and (12.10.46). In this way there is less usage of many-valued functions.

Values at \zeta=0

10.20.14
A_{0}(0)=1,
A_{1}(0)=-\tfrac{1}{225},
A_{2}(0)=\tfrac{1\;51439}{2182\;95000},
A_{3}(0)=-\tfrac{8872\;78009}{250\;49351\;25000},\\
B_{0}(0)=\tfrac{1}{70}2^{{\frac{1}{3}}},
B_{1}(0)=-\tfrac{1213}{10\;23750}2^{{\frac{1}{3}}},
B_{2}(0)=\tfrac{1\;65425\;37833}{3774\;32055\;00000}2^{{\frac{1}{3}}},
B_{3}(0)=-\tfrac{959\;71711\;84603}{25\;47666\;37125\;00000}2^{{\frac{1}{3}}}.

Each of the coefficients A_{k}(\zeta), B_{k}(\zeta), C_{k}(\zeta), and D_{k}(\zeta), k=0,1,2,\ldots, is real and infinitely differentiable on the interval -\infty<\zeta<\infty. For (10.20.14) and further information on the coefficients see Temme (1997).

For numerical tables of \zeta=\zeta(z), (4\zeta/(1-z^{2}))^{{\frac{1}{4}}} and A_{k}(\zeta), B_{k}(\zeta), C_{k}(\zeta), and D_{k}(\zeta) see Olver (1962, pp. 28–42).

§10.20(ii) Complex Variables

The function \zeta=\zeta(z) given by (10.20.2) and (10.20.3) can be continued analytically to the z-plane cut along the negative real axis. Corresponding points of the mapping are shown in Figures 10.20.1 and 10.20.2.

The equations of the curved boundaries D_{1}E_{1} and D_{2}E_{2} in the \zeta-plane are given parametrically by

10.20.15\zeta=(\tfrac{3}{2})^{{\frac{2}{3}}}(\tau\mp i\pi)^{{\frac{2}{3}}},0\leq\tau<\infty,

respectively.

The curves BP_{1}E_{1} and BP_{2}E_{2} in the z-plane are the inverse maps of the line segments

10.20.16\zeta=e^{{\mp i\pi/3}}\tau,0\leq\tau\leq(\tfrac{3}{2}\pi)^{{\frac{2}{3}}},

respectively. They are given parametrically by

where \tau_{0}=1.19968\ldots is the positive root of the equation \tau=\mathop{\coth\/}\nolimits\tau. The points P_{1},P_{2} where these curves intersect the imaginary axis are \pm ic, where

10.20.18c=(\tau_{0}^{2}-1)^{{\frac{1}{2}}}=0.66274\ldots.

The eye-shaped closed domain in the uncut z-plane that is bounded by BP_{1}E_{1} and BP_{2}E_{2} is denoted by \mathbf{K}; see Figure 10.20.3.

See accompanying text
Figure 10.20.1: z-plane. P_{1} and P_{2} are the points \pm ic. c=0.66274\ldots. Magnify
See accompanying text
Figure 10.20.2: \zeta-plane. E_{1} and E_{2} are the points e^{{\mp\pi i/3}}(3\pi/2)^{{2/3}}. Magnify
See accompanying text
Figure 10.20.3: z-plane. Domain \mathbf{K} (unshaded). c=0.66274\dots. Magnify

As \nu\to\infty through positive real values the expansions (10.20.4)–(10.20.9) apply uniformly for |\mathop{\mathrm{ph}\/}\nolimits z|\leq\pi-\delta, the coefficients A_{k}(\zeta), B_{k}(\zeta), C_{k}(\zeta), and D_{k}(\zeta), being the analytic continuations of the functions defined in §10.20(i) when \zeta is real.

For proofs of the above results and for error bounds and extensions of the regions of validity see Olver (1997b, pp. 419–425). For extensions to complex \nu see Olver (1954). For resurgence properties of the coefficients (§2.7(ii)) see Howls and Olde Daalhuis (1999). For further results see Dunster (2001a), Wang and Wong (2002), and Paris (2004).

§10.20(iii) Double Asymptotic Properties

For asymptotic properties of the expansions (10.20.4)–(10.20.6) with respect to large values of z see §10.41(v).