Neumann polynomial

In mathematics, the Neumann polynomials, introduced by Carl Neumann for the special case ${\displaystyle \alpha =0}$, are a sequence of polynomials in ${\displaystyle 1/t}$ used to expand functions in term of Bessel functions.¹

The first few polynomials are

${\displaystyle O_{0}^{(\alpha )}(t)={\frac {1}{t}},}$

${\displaystyle O_{1}^{(\alpha )}(t)=2{\frac {\alpha +1}{t^{2}}},}$

${\displaystyle O_{2}^{(\alpha )}(t)={\frac {2+\alpha }{t}}+4{\frac {(2+\alpha )(1+\alpha )}{t^{3}}},}$

${\displaystyle O_{3}^{(\alpha )}(t)=2{\frac {(1+\alpha )(3+\alpha )}{t^{2}}}+8{\frac {(1+\alpha )(2+\alpha )(3+\alpha )}{t^{4}}},}$

${\displaystyle O_{4}^{(\alpha )}(t)={\frac {(1+\alpha )(4+\alpha )}{2t}}+4{\frac {(1+\alpha )(2+\alpha )(4+\alpha )}{t^{3}}}+16{\frac {(1+\alpha )(2+\alpha )(3+\alpha )(4+\alpha )}{t^{5}}}.}$

A general form for the polynomial is

${\displaystyle O_{n}^{(\alpha )}(t)={\frac {\alpha +n}{2\alpha }}\sum _{k=0}^{\lfloor n/2\rfloor }(-1)^{n-k}{\frac {(n-k)!}{k!}}{-\alpha \choose n-k}\left({\frac {2}{t}}\right)^{n+1-2k},}$

and they have the "generating function"

${\displaystyle {\frac {\left({\frac {z}{2}}\right)^{\alpha }}{\Gamma (\alpha +1)}}{\frac {1}{t-z}}=\sum _{n=0}O_{n}^{(\alpha )}(t)J_{\alpha +n}(z),}$

where J are Bessel functions.

To expand a function f in the form

${\displaystyle f(z)=\left({\frac {2}{z}}\right)^{\alpha }\sum _{n=0}a_{n}J_{\alpha +n}(z)\,}$

for ${\displaystyle |t|<c}$, compute

${\displaystyle a_{n}={\frac {\Gamma (\alpha +1)}{2\pi i}}\oint _{|t|=c'}f(t)O_{n}^{(\alpha )}(t)\,dt,}$

where ${\displaystyle c'<c}$ and c is the distance of the nearest singularity of f(z) from ${\displaystyle z=0}$.

An example is the extension

${\displaystyle \left({\tfrac {1}{2}}z\right)^{s}=\Gamma (s)\cdot \sum _{k=0}(-1)^{k}J_{s+2k}(z)(s+2k){-s \choose k},}$

or the more general Sonine formula²

${\displaystyle e^{i\gamma z}=\Gamma (s)\cdot \sum _{k=0}i^{k}C_{k}^{(s)}(\gamma )(s+k){\frac {J_{s+k}(z)}{\left({\frac {z}{2}}\right)^{s}}}.}$

where ${\displaystyle C_{k}^{(s)}}$ is Gegenbauer's polynomial. Then,

${\displaystyle {\frac {\left({\frac {z}{2}}\right)^{2k}}{(2k-1)!}}J_{s}(z)=\sum _{i=k}(-1)^{i-k}{i+k-1 \choose 2k-1}{i+k+s-1 \choose 2k-1}(s+2i)J_{s+2i}(z),}$

${\displaystyle \sum _{n=0}t^{n}J_{s+n}(z)={\frac {e^{\frac {tz}{2}}}{t^{s}}}\sum _{j=0}{\frac {\left(-{\frac {z}{2t}}\right)^{j}}{j!}}{\frac {\gamma \left(j+s,{\frac {tz}{2}}\right)}{\,\Gamma (j+s)}}=\int _{0}^{\infty }e^{-{\frac {zx^{2}}{2t}}}{\frac {zx}{t}}{\frac {J_{s}(z{\sqrt {1-x^{2}}})}{{\sqrt {1-x^{2}}}^{s}}}\,dx,}$

the confluent hypergeometric function

${\displaystyle M(a,s,z)=\Gamma (s)\sum _{k=0}^{\infty }\left(-{\frac {1}{t}}\right)^{k}L_{k}^{(-a-k)}(t){\frac {J_{s+k-1}\left(2{\sqrt {tz}}\right)}{({\sqrt {tz}})^{s-k-1}}},}$

and in particular

${\displaystyle {\frac {J_{s}(2z)}{z^{s}}}={\frac {4^{s}\Gamma \left(s+{\frac {1}{2}}\right)}{\sqrt {\pi }}}e^{2iz}\sum _{k=0}L_{k}^{(-s-1/2-k)}\left({\frac {it}{4}}\right)(4iz)^{k}{\frac {J_{2s+k}\left(2{\sqrt {tz}}\right)}{{\sqrt {tz}}^{2s+k}}},}$

the index shift formula

${\displaystyle \Gamma (\nu -\mu )J_{\nu }(z)=\Gamma (\mu +1)\sum _{n=0}{\frac {\Gamma (\nu -\mu +n)}{n!\Gamma (\nu +n+1)}}\left({\frac {z}{2}}\right)^{\nu -\mu +n}J_{\mu +n}(z),}$

the Taylor expansion (addition formula)

${\displaystyle {\frac {J_{s}\left({\sqrt {z^{2}-2uz}}\right)}{\left({\sqrt {z^{2}-2uz}}\right)^{\pm s}}}=\sum _{k=0}{\frac {(\pm u)^{k}}{k!}}{\frac {J_{s\pm k}(z)}{z^{\pm s}}},}$

(cf.³) and the expansion of the integral of the Bessel function,

${\displaystyle \int J_{s}(z)dz=2\sum _{k=0}J_{s+2k+1}(z),}$

are of the same type.