@cindex Mathieu functions

The routines described in this section compute the angular and radial
Mathieu functions, and their characteristic values.  Mathieu
functions are the solutions of the following two differential
equations:
@tex
\beforedisplay
$$
\eqalign{
{{d^2 y}\over{d v^2}}& + (a - 2q\cos 2v)y  = 0, \cr
{{d^2 f}\over{d u^2}}& - (a - 2q\cosh 2u)f  = 0.
}
$$
\afterdisplay
@end tex
@ifinfo

@example
d^2y/dv^2 + (a - 2q\cos 2v)y = 0
d^2f/du^2 - (a - 2q\cosh 2u)f = 0
@end example

@end ifinfo
@noindent
The angular Mathieu functions @math{ce_r(x,q)}, @math{se_r(x,q)} are
the even and odd periodic solutions of the first equation, which is known as Mathieu's equation. These exist
only for the discrete sequence of  characteristic values @math{a=a_r(q)}
(even-periodic) and @math{a=b_r(q)} (odd-periodic).

The radial Mathieu functions @c{$Mc^{(j)}_{r}(z,q)$}
@math{Mc^@{(j)@}_@{r@}(z,q)}, @c{$Ms^{(j)}_{r}(z,q)$}
@math{Ms^@{(j)@}_@{r@}(z,q)} are the solutions of the second equation,
which is referred to as Mathieu's modified equation.  The
radial Mathieu functions of the first, second, third and fourth kind
are denoted by the parameter @math{j}, which takes the value 1, 2, 3
or 4.

@comment The angular Mathieu functions can be divided into four types as
@comment @tex
@comment \beforedisplay
@comment $$
@comment \eqalign{
@comment x & = \sum_{m=0}^\infty A_{2m+p} \cos(2m+p)\phi, \quad p = 0, 1, \cr
@comment x & = \sum_{m=0}^\infty B_{2m+p} \sin(2m+p)\phi, \quad p = 0, 1.
@comment }
@comment $$
@comment \afterdisplay
@comment @end tex
@comment @ifinfo

@comment @example
@comment x = \sum_(m=0)^\infty A_(2m+p) \cos(2m+p)\phi,   p = 0, 1,
@comment x = \sum_(m=0)^\infty B_(2m+p) \sin(2m+p)\phi,   p = 0, 1.
@comment @end example

@comment @end ifinfo
@comment @noindent
@comment The nomenclature used for the angular Mathieu functions is @math{ce_n}
@comment for the first solution and @math{se_n} for the second.

@comment Similar solutions exist for the radial Mathieu functions by replacing
@comment the trigonometric functions with their corresponding hyperbolic
@comment functions as shown below.
@comment @tex
@comment \beforedisplay
@comment $$
@comment \eqalign{
@comment x & = \sum_{m=0}^\infty A_{2m+p} \cosh(2m+p)u, \quad p = 0, 1, \cr
@comment x & = \sum_{m=0}^\infty B_{2m+p} \sinh(2m+p)u, \quad p = 0, 1.
@comment }
@comment $$
@comment \afterdisplay
@comment @end tex
@comment @ifinfo

@comment @example
@comment x = \sum_(m=0)^\infty A_(2m+p) \cosh(2m+p)u,   p = 0, 1,
@comment x = \sum_(m=0)^\infty B_(2m+p) \sinh(2m+p)u,   p = 0, 1.
@comment @end example

@comment @end ifinfo
@comment @noindent
@comment The nomenclature used for the radial Mathieu functions is @math{Mc_n}
@comment for the first solution and @math{Ms_n} for the second.  The hyperbolic
@comment series do not always converge at an acceptable rate.  Therefore most
@comment texts on the subject suggest using the following equivalent equations
@comment that are expanded in series of Bessel and Hankel functions.
@comment @tex
@comment \beforedisplay
@comment $$
@comment \eqalign{
@comment Mc_{2n}^{(j)}(x,q) & = \sum_{m=0}^\infty (-1)^{r+k}
@comment       A_{2m}^{2n}(q)\left[J_m(u_1)Z_m^{(j)}(u_2) +
@comment                           J_m(u_1)Z_m^{(j)}(u_2)\right]/A_2^{2n} \cr
@comment Mc_{2n+1}^{(j)}(x,q) & = \sum_{m=0}^\infty (-1)^{r+k}
@comment       A_{2m+1}^{2n+1}(q)\left[J_m(u_1)Z_{m+1}^{(j)}(u_2) +
@comment                               J_{m+1}(u_1)Z_m^{(j)}(u_2)\right]/A_1^{2n+1} \cr
@comment Ms_{2n}^{(j)}(x,q) & = \sum_{m=1}^\infty (-1)^{r+k}
@comment       B_{2m}^{2n}(q)\left[J_{m-1}(u_1)Z_{m+1}^{(j)}(u_2) +
@comment                           J_{m+1}(u_1)Z_{m-1}^{(j)}(u_2)\right]/B_2^{2n} \cr
@comment Ms_{2n+1}^{(j)}(x,q) & = \sum_{m=0}^\infty (-1)^{r+k}
@comment       B_{2m+1}^{2n+1}(q)\left[J_m(u_1)Z_{m+1}^{(j)}(u_2) +
@comment                               J_{m+1}(u_1)Z_m^{(j)}(u_2)\right]/B_1^{2n+1}
@comment }
@comment $$
@comment \afterdisplay
@comment @end tex
@comment @ifinfo

@comment @example
@comment Mc_(2n)^(j)(x,q) = \sum_(m=0)^\infty (-1)^(r+k) A_(2m)^(2n)(q)
@comment     [J_m(u_1)Z_m^(j)(u_2) + J_m(u_1)Z_m^(j)(u_2)]/A_2^(2n)
@comment Mc_(2n+1)^(j)(x,q) = \sum_(m=0)^\infty (-1)^(r+k) A_(2m+1)^(2n+1)(q)
@comment     [J_m(u_1)Z_(m+1)^(j)(u_2) + J_(m+1)(u_1)Z_m^(j)(u_2)]/A_1^(2n+1)
@comment Ms_(2n)^(j)(x,q) = \sum_(m=1)^\infty (-1)^(r+k) B_(2m)^(2n)(q)
@comment     [J_(m-1)(u_1)Z_(m+1)^(j)(u_2) + J_(m+1)(u_1)Z_(m-1)^(j)(u_2)]/B_2^(2n)
@comment Ms_(2n+1)^(j)(x,q) = \sum_(m=0)^\infty (-1)^(r+k) B_(2m+1)^(2n+1)(q)
@comment     [J_m(u_1)Z_(m+1)^(j)(u_2) + J_(m+1)(u_1)Z_m^(j)(u_2)]/B_1^(2n+1)
@comment @end example

@comment @end ifinfo
@comment @noindent
@comment where @c{$u_1 = \sqrt{q} \exp(-x)$} 
@comment @math{u_1 = \sqrt@{q@} \exp(-x)} and @c{$u_2 = \sqrt@{q@} \exp(x)$}
@comment @math{u_2 = \sqrt@{q@} \exp(x)} and
@comment @tex
@comment \beforedisplay
@comment $$
@comment \eqalign{
@comment Z_m^{(1)}(u) & = J_m(u) \cr
@comment Z_m^{(2)}(u) & = Y_m(u) \cr
@comment Z_m^{(3)}(u) & = H_m^{(1)}(u) \cr
@comment Z_m^{(4)}(u) & = H_m^{(2)}(u)
@comment }
@comment $$
@comment \afterdisplay
@comment @end tex
@comment @ifinfo

@comment @example
@comment Z_m^(1)(u) = J_m(u)
@comment Z_m^(2)(u) = Y_m(u)
@comment Z_m^(3)(u) = H_m^(1)(u)
@comment Z_m^(4)(u) = H_m^(2)(u)
@comment @end example

@comment @end ifinfo
@comment @noindent
@comment where @math{J_m(u)}, @math{Y_m(u)}, @math{H_m^{(1)}(u)}, and
@comment @math{H_m^{(2)}(u)} are the regular and irregular Bessel functions and
@comment the Hankel functions, respectively.

For more information on the Mathieu functions, see Abramowitz and
Stegun, Chapter 20.  These functions are defined in the header file
@file{gsl_sf_mathieu.h}.

@menu
* Mathieu Function Workspace::  
* Mathieu Function Characteristic Values::  
* Angular Mathieu Functions::   
* Radial Mathieu Functions::    
@end menu

@node  Mathieu Function Workspace
@subsection  Mathieu Function Workspace

The Mathieu functions can be computed for a single order or
for multiple orders, using array-based routines.  The array-based
routines require a preallocated workspace.

@deftypefun {gsl_sf_mathieu_workspace *} gsl_sf_mathieu_alloc (size_t @var{n}, double @var{qmax})
@tindex gsl_sf_mathieu_workspace
This function returns a workspace for the array versions of the
Mathieu routines.  The arguments @var{n} and @var{qmax} specify the
maximum order and @math{q}-value of Mathieu functions which can be
computed with this workspace.  

@comment This is required in order to properly
@comment terminate the infinite eigenvalue matrix for high precision solutions.
@comment The characteristic values for all orders @math{0 \to n} are stored in
@comment the work structure array element @kbd{work->char_value}.
@end deftypefun

@deftypefun void gsl_sf_mathieu_free (gsl_sf_mathieu_workspace * @var{work})
This function frees the workspace @var{work}.
@end deftypefun

@node Mathieu Function Characteristic Values
@subsection Mathieu Function Characteristic Values
@cindex Mathieu Function Characteristic Values

@deftypefun int gsl_sf_mathieu_a (int @var{n}, double @var{q})
@deftypefunx int gsl_sf_mathieu_a_e (int @var{n}, double @var{q}, gsl_sf_result * @var{result})
@deftypefunx int gsl_sf_mathieu_b (int @var{n}, double @var{q})
@deftypefunx int gsl_sf_mathieu_b_e (int @var{n}, double @var{q}, gsl_sf_result * @var{result})
These routines compute the characteristic values @math{a_n(q)},
@math{b_n(q)} of the Mathieu functions @math{ce_n(q,x)} and
@math{se_n(q,x)}, respectively.
@end deftypefun

@deftypefun int gsl_sf_mathieu_a_array (int @var{order_min}, int @var{order_max}, double @var{q}, gsl_sf_mathieu_workspace * @var{work}, double @var{result_array}[])
@deftypefunx int gsl_sf_mathieu_b_array (int @var{order_min}, int @var{order_max}, double @var{q}, gsl_sf_mathieu_workspace * @var{work}, double @var{result_array}[])
These routines compute a series of Mathieu characteristic values
@math{a_n(q)}, @math{b_n(q)} for @math{n} from @var{order_min} to
@var{order_max} inclusive, storing the results in the array @var{result_array}.
@end deftypefun

@node Angular Mathieu Functions
@subsection Angular Mathieu Functions
@cindex Angular Mathieu Functions
@cindex @math{ce(q,x)}, Mathieu function
@cindex @math{se(q,x)}, Mathieu function

@deftypefun int gsl_sf_mathieu_ce (int @var{n}, double @var{q}, double @var{x})
@deftypefunx int gsl_sf_mathieu_ce_e (int @var{n}, double @var{q}, double @var{x}, gsl_sf_result * @var{result})
@deftypefunx int gsl_sf_mathieu_se (int @var{n}, double @var{q}, double @var{x})
@deftypefunx int gsl_sf_mathieu_se_e (int @var{n}, double @var{q}, double @var{x}, gsl_sf_result * @var{result})
These routines compute the angular Mathieu functions @math{ce_n(q,x)}
and @math{se_n(q,x)}, respectively.
@end deftypefun

@deftypefun int gsl_sf_mathieu_ce_array (int @var{nmin}, int @var{nmax}, double @var{q}, double @var{x}, gsl_sf_mathieu_workspace * @var{work}, double @var{result_array}[])
@deftypefunx int gsl_sf_mathieu_se_array (int @var{nmin}, int @var{nmax}, double @var{q}, double @var{x}, gsl_sf_mathieu_workspace * @var{work}, double @var{result_array}[])
These routines compute a series of the angular Mathieu functions
@math{ce_n(q,x)} and @math{se_n(q,x)} of order @math{n} from
@var{nmin} to @var{nmax} inclusive, storing the results in the array
@var{result_array}.
@end deftypefun

@node Radial Mathieu Functions
@subsection Radial Mathieu Functions
@cindex Radial Mathieu Functions

@deftypefun int gsl_sf_mathieu_Mc (int @var{j}, int @var{n}, double @var{q}, double @var{x})
@deftypefunx int gsl_sf_mathieu_Mc_e (int @var{j}, int @var{n}, double @var{q}, double @var{x}, gsl_sf_result * @var{result})
@deftypefunx int gsl_sf_mathieu_Ms (int @var{j}, int @var{n}, double @var{q}, double @var{x})
@deftypefunx int gsl_sf_mathieu_Ms_e (int @var{j}, int @var{n}, double @var{q}, double @var{x}, gsl_sf_result * @var{result})
These routines compute the radial @var{j}-th kind Mathieu functions
@c{$Mc_n^{(j)}(q,x)$}
@math{Mc_n^@{(j)@}(q,x)} and 
@c{$Ms_n^{(j)}(q,x)$}
@math{Ms_n^@{(j)@}(q,x)} of order @var{n}.

The allowed values of @var{j} are 1 and 2.
The functions for @math{j = 3,4} can be computed as 
@c{$M_n^{(3)} = M_n^{(1)} + iM_n^{(2)}$}
@math{M_n^@{(3)@} = M_n^@{(1)@} + iM_n^@{(2)@}} and 
@c{$M_n^{(4)} = M_n^{(1)} - iM_n^{(2)}$}
@math{M_n^@{(4)@} = M_n^@{(1)@} - iM_n^@{(2)@}},
where 
@c{$M_n^{(j)} = Mc_n^{(j)}$}
@math{M_n^@{(j)@} = Mc_n^@{(j)@}} or 
@c{$Ms_n^{(j)}$}
@math{Ms_n^@{(j)@}}.
@end deftypefun

@deftypefun int gsl_sf_mathieu_Mc_array (int @var{j}, int @var{nmin}, int @var{nmax}, double @var{q}, double @var{x}, gsl_sf_mathieu_workspace * @var{work}, double @var{result_array}[])
@deftypefunx int gsl_sf_mathieu_Ms_array (int @var{j}, int @var{nmin}, int @var{nmax}, double @var{q}, double @var{x}, gsl_sf_mathieu_workspace * @var{work}, double @var{result_array}[])
These routines compute a series of the radial Mathieu functions of
kind @var{j}, with order from @var{nmin} to @var{nmax} inclusive, storing the
results in the array @var{result_array}.
@end deftypefun