A regular Sturm-Liouville problem on an interval is of the form
Except for the various sign requirements, the distinguishing feature is that the coefficient of the f' term (-p') is the derivative of the coefficient for the f'' term (-p). Starting with an arbitrary second order linear O.D.E., you can achieve such a form by multiplying the entire O.D.E. with a suitable factor.
The boundary conditions may either be periodic ones,
If you have a Sturm-Liouville problem, simply (well, simply ...) solve
it. The solutions only exists for certain values of . Make
sure you find all solutions, or you are in trouble. They will
form an infinite sequence of `eigenfunctions', say f1(x), f2(x),
f3(x), ... with corresponding `eigenvalues'
,
,
, ... that go off to positive infinity.
Note: If you have to know why they are called eigenfunctions and eigenvectors, if you consider functions like f to play the role of (infinite-dimensional) vectors, and the operator
Any arbitrary function, say g(x), on the interval [a,b] can always be represented as a sum of the found eigenfunctions:
If you know g(x), you can find its Fourier coefficients Cn from the following formula:
Note: The above formula is referred to as orthogonality. If you have to know why, assume that our function g(x) happens to be an eigenfunction, say f5. In that case, C5 has to be one (which is automatic - a good way to check that you have your formula correct), and all other Cn need to be zero. From the formula, you see that then f5 must integrate to zero against any other function fn. Identifying such an integral as an inner product between ``vectors'', f5 would be orthogonal to all other eigenfunctions.