28.1 Special Notation28.3 Graphics

§28.2 Definitions and Basic Properties

Contents

§28.2(i) Mathieu’s Equation

The standard form of Mathieu’s equation with parameters (a,q) is

28.2.1w^{{\prime\prime}}+(a-2q\mathop{\cos\/}\nolimits\!\left(2z\right))w=0.

With \zeta={\mathop{\sin\/}\nolimits^{{2}}}z we obtain the algebraic form of Mathieu’s equation

28.2.2\zeta(1-\zeta)w^{{\prime\prime}}+\tfrac{1}{2}\left(1-2\zeta)w^{{\prime}}+\tfrac{1}{4}(a-2q(1-2\zeta)\right)w=0.

This equation has regular singularities at 0 and 1, both with exponents 0 and \frac{1}{2}, and an irregular singular point at \infty. With \zeta=\mathop{\cos\/}\nolimits z we obtain another algebraic form:

28.2.3(1-\zeta^{2})w^{{\prime\prime}}-\zeta w^{{\prime}}+\left(a+2q-4q\zeta^{2}\right)w=0.

§28.2(ii) Basic Solutions w_{{\mbox{\rm\tiny I}}}, w_{{\mbox{\rm\tiny II}}}

Since (28.2.1) has no finite singularities its solutions are entire functions of z. Furthermore, a solution w with given initial constant values of w and w^{{\prime}} at a point z_{0} is an entire function of the three variables z, a, and q.

The following three transformations

28.2.4
z\to-z;
z\to z\pm\pi;
z\to z\pm\tfrac{1}{2}\pi,q\to-q;

each leave (28.2.1) unchanged. (28.2.1) possesses a fundamental pair of solutions w_{{\mbox{\tiny I}}}(z;a,q),w_{{\mbox{\tiny II}}}(z;a,q) called basic solutions with

28.2.5\begin{bmatrix}w_{{\mbox{\tiny I}}}(0;a,q)&w_{{\mbox{\tiny II}}}(0;a,q)\\
w^{{\prime}}_{{\mbox{\tiny I}}}(0;a,q)&w^{{\prime}}_{{\mbox{\tiny II}}}(0;a,q)\end{bmatrix}=\begin{bmatrix}1&0\\
0&1\end{bmatrix}.

w_{{\mbox{\tiny I}}}(z;a,q) is even and w_{{\mbox{\tiny II}}}(z;a,q) is odd. Other properties are as follows.

28.2.6\mathop{\mathscr{W}\/}\nolimits\left\{ w_{{\mbox{\tiny I}}},w_{{\mbox{\tiny II}}}\right\}=1,

§28.2(iii) Floquet’s Theorem and the Characteristic Exponents

Let \nu be any real or complex constant. Then Mathieu’s equation (28.2.1) has a nontrivial solution w(z) such that

28.2.14w(z+\pi)=e^{{\pi i\nu}}w(z),

iff e^{{\pi i\nu}} is an eigenvalue of the matrix

28.2.15\begin{bmatrix}w_{{\mbox{\tiny I}}}(\pi;a,q)&w_{{\mbox{\tiny II}}}(\pi;a,q)\\
w^{{\prime}}_{{\mbox{\tiny I}}}(\pi;a,q)&w^{{\prime}}_{{\mbox{\tiny II}}}(\pi;a,q)\end{bmatrix}.

Equivalently,

28.2.16\mathop{\cos\/}\nolimits\!\left(\pi\nu\right)=w_{{\mbox{\tiny I}}}(\pi;a,q)=w_{{\mbox{\tiny I}}}(\pi;a,-q).

This is the characteristic equation of Mathieu’s equation (28.2.1). \mathop{\cos\/}\nolimits\!\left(\pi\nu\right) is an entire function of a,q^{2}. The solutions of (28.2.16) are given by \nu=\pi^{{-1}}\mathop{\mathrm{arccos}\/}\nolimits\!\left(w_{{\mbox{\tiny I}}}(\pi;a,q)\right). If the inverse cosine takes its principal value (§4.23(ii)), then \nu=\widehat{\nu}, where 0\leq\realpart{\widehat{\nu}}\leq 1. The general solution of (28.2.16) is \nu=\pm\widehat{\nu}+2n, where n\in\Integer. Either \widehat{\nu} or \nu is called a characteristic exponent of (28.2.1). If \widehat{\nu}=0 or 1, or equivalently, \nu=n, then \nu is a double root of the characteristic equation, otherwise it is a simple root.

§28.2(iv) Floquet Solutions

A solution with the pseudoperiodic property (28.2.14) is called a Floquet solution with respect to \nu. (28.2.9), (28.2.16), and (28.2.7) give for each solution w(z) of (28.2.1) the connection formula

28.2.17w(z+\pi)+w(z-\pi)=2\mathop{\cos\/}\nolimits\!\left(\pi\nu\right)w(z).

Therefore a nontrivial solution w(z) is either a Floquet solution with respect to \nu, or w(z+\pi)-e^{{i\nu\pi}}w(z) is a Floquet solution with respect to -\nu.

If q\neq 0, then for a given value of \nu the corresponding Floquet solution is unique, except for an arbitrary constant factor (Theorem of Ince; see also 28.5(i)).

The Fourier series of a Floquet solution

28.2.18w(z)=\sum _{{n=-\infty}}^{{\infty}}c_{{2n}}e^{{i(\nu+2n)z}}

converges absolutely and uniformly in compact subsets of \Complex. The coefficients c_{{2n}} satisfy

28.2.19{qc_{{2n+2}}-\left(a-(\nu+2n)^{2}\right)c_{{2n}}+qc_{{2n-2}}=0,}n\in\Integer.

Conversely, a nontrivial solution c_{{2n}} of (28.2.19) that satisfies

28.2.20\lim _{{n\to\pm\infty}}|c_{{2n}}|^{{1/|n|}}=0

leads to a Floquet solution.

§28.2(v) Eigenvalues \mathop{a_{{n}}\/}\nolimits, \mathop{b_{{n}}\/}\nolimits

For given \nu and q, equation (28.2.16) determines an infinite discrete set of values of a, the eigenvalues or characteristic values, of Mathieu’s equation. When \widehat{\nu}=0 or 1, the notation for the two sets of eigenvalues corresponding to each \widehat{\nu} is shown in Table 28.2.1, together with the boundary conditions of the associated eigenvalue problem. In Table 28.2.1 n=0,1,2,\dots.

Table 28.2.1: Eigenvalues of Mathieu’s equation.
\widehat{\nu} Boundary Conditions Eigenvalues
0 w^{{\prime}}(0)=w^{{\prime}}(\tfrac{1}{2}\pi)=0 \mathop{a_{{2n}}\/}\nolimits\!\left(q\right)
1 w^{{\prime}}(0)=w(\tfrac{1}{2}\pi)=0 \mathop{a_{{2n+1}}\/}\nolimits\!\left(q\right)
1 w(0)=w^{{\prime}}(\tfrac{1}{2}\pi)=0 \mathop{b_{{2n+1}}\/}\nolimits\!\left(q\right)
0 w(0)=w(\tfrac{1}{2}\pi)=0 \mathop{b_{{2n+2}}\/}\nolimits\!\left(q\right)

An equivalent formulation is given by

28.2.21\begin{array}[]{ll}w^{{\prime}}_{{\mbox{\tiny I}}}(\tfrac{1}{2}\pi;a,q)=0,&a=\mathop{a_{{2n}}\/}\nolimits\!\left(q\right),\\
w_{{\mbox{\tiny I}}}(\tfrac{1}{2}\pi;a,q)=0,&a=\mathop{a_{{2n+1}}\/}\nolimits\!\left(q\right),\end{array}

and

28.2.22\begin{array}[]{ll}w^{{\prime}}_{{\mbox{\tiny II}}}(\tfrac{1}{2}\pi;a,q)=0,&a=\mathop{b_{{2n+1}}\/}\nolimits\!\left(q\right),\\
w_{{\mbox{\tiny II}}}(\tfrac{1}{2}\pi;a,q)=0,&a=\mathop{b_{{2n+2}}\/}\nolimits\!\left(q\right),\end{array}

where n=0,1,2,\dots. When q=0,

Near q=0, \mathop{a_{{n}}\/}\nolimits\!\left(q\right) and \mathop{b_{{n}}\/}\nolimits\!\left(q\right) can be expanded in power series in q (see §28.6(i)); elsewhere they are determined by analytic continuation (see §28.7). For nonnegative real values of q, see Figure 28.2.1.

See accompanying text
Figure 28.2.1: Eigenvalues \mathop{a_{{n}}\/}\nolimits\!\left(q\right), \mathop{b_{{n}}\/}\nolimits\!\left(q\right) of Mathieu’s equation as functions of q for 0\leq q\leq 10, n=0,1,2,3,4 (a’s), n=1,2,3,4 (b’s). Magnify

Change of Sign of q

28.2.26\mathop{a_{{2n}}\/}\nolimits\!\left(-q\right)=\mathop{a_{{2n}}\/}\nolimits\!\left(q\right),
28.2.27\mathop{a_{{2n+1}}\/}\nolimits\!\left(-q\right)=\mathop{b_{{2n+1}}\/}\nolimits\!\left(q\right),
28.2.28\mathop{b_{{2n+2}}\/}\nolimits\!\left(-q\right)=\mathop{b_{{2n+2}}\/}\nolimits\!\left(q\right).

§28.2(vi) Eigenfunctions

Table 28.2.2 gives the notation for the eigenfunctions corresponding to the eigenvalues in Table 28.2.1. Period \pi means that the eigenfunction has the property w(z+\pi)=w(z), whereas antiperiod \pi means that w(z+\pi)=-w(z). Even parity means w(-z)=w(z), and odd parity means w(-z)=-w(z).

Table 28.2.2: Eigenfunctions of Mathieu’s equation.
Eigenvalues Eigenfunctions Periodicity Parity
\mathop{a_{{2n}}\/}\nolimits\!\left(q\right) \mathop{\mathrm{ce}_{{2n}}\/}\nolimits\!\left(z,q\right) Period \pi Even
\mathop{a_{{2n+1}}\/}\nolimits\!\left(q\right) \mathop{\mathrm{ce}_{{2n+1}}\/}\nolimits\!\left(z,q\right) Antiperiod \pi Even
\mathop{b_{{2n+1}}\/}\nolimits\!\left(q\right) \mathop{\mathrm{se}_{{2n+1}}\/}\nolimits\!\left(z,q\right) Antiperiod \pi Odd
\mathop{b_{{2n+2}}\/}\nolimits\!\left(q\right) \mathop{\mathrm{se}_{{2n+2}}\/}\nolimits\!\left(z,q\right) Period \pi Odd

For simple roots q of the corresponding equations (28.2.21) and (28.2.22), the functions are made unique by the normalizations

28.2.30
\int _{0}^{{2\pi}}\left(\mathop{\mathrm{ce}_{{n}}\/}\nolimits\!\left(x,q\right)\right)^{2}dx=\pi,
\int _{0}^{{2\pi}}\left(\mathop{\mathrm{se}_{{n}}\/}\nolimits\!\left(x,q\right)\right)^{2}dx=\pi,

the ambiguity of sign being resolved by (28.2.29) when q=0 and by continuity for the other values of q.