Skip to content
Tutorials Center
My learning

Topics

Lesson 24 of 39 ยท Series and Special Functions

The Frobenius Method

Solve ODEs at regular singular points with the Frobenius method: indicial equation, the three root cases, recurrence relations and Python checks.

Some of the most important equations in physics have a coefficient that blows up exactly where you most want to expand: at x = 0. Bessel's equation, which describes vibrating drums, has a term y'/x. The simple power series ฮฃ a_n x^n fails there, because solutions can behave like sqrt(x), x^(-1) or ln x.

The Frobenius method repairs the series by multiplying it by a power x^r with a number r that you solve for. It works whenever the singular point is regular, which covers almost every equation you meet in practice.

Regular singular points

Write the equation as y'' + P(x) y' + Q(x) y = 0. The point x = 0 is a regular singular point when P and Q are singular there, but only mildly:

TEXT
x P(x)  and  x^2 Q(x)  are analytic at 0

In other words P has at most a simple pole and Q at most a double pole. Say x P(x) -> p0 and x^2 Q(x) -> q0 as x -> 0. If the singularity is worse than this, the point is irregular and the method does not apply as is.

The method

Assume a solution of the form

TEXT
y = x^r ฮฃ a_n x^n = ฮฃ a_n x^(n+r),     a0 โ‰  0

Substitute and look at the lowest power of x. Its coefficient gives the indicial equation:

TEXT
r(r - 1) + p0 r + q0 = 0

Its two roots r1 >= r2 (real case) decide how the solution looks. All the higher powers give a recurrence relation for a_1, a_2, ... in terms of a_0, just like in power series solutions, except that r appears in it.

The three cases

Roots of the indicial equation

Not an integer apart Two Frobenius series, one per root.
Equal roots Second solution has a ln x term.
Differ by an integer Larger root always works; the smaller may need ln x.

In every case the larger root r1 gives a solution of Frobenius form.

1 / 2
  1. r1 - r2 is not an integer. Both y1 = x^(r1) ฮฃ a_n x^n and y2 = x^(r2) ฮฃ b_n x^n are valid, and they are independent.
  2. r1 = r2. Only one pure Frobenius series exists. The second solution is y2 = y1 ln x + x^(r1) ฮฃ b_n x^n.
  3. r1 - r2 is a positive integer. Always y1 exists. The recurrence for the smaller root may run into a division by zero. If it does not, you get y2 as a pure series; otherwise y2 = C y1 ln x + x^(r2) ฮฃ b_n x^n, with C โ‰  0.

Example: 2x y'' + y' + y = 0

Divide by 2x: P = 1/(2x), Q = 1/(2x). Then p0 = 1/2 and q0 = 0, so the indicial equation is r(r-1) + r/2 = 0, i.e. r (2r - 1) = 0, with roots r = 1/2 and r = 0. They differ by 1/2, not an integer, so we are in case 1.

Substituting y = ฮฃ a_n x^(n+r) into the original equation gives

TEXT
(n + r)(2(n + r) - 1) a_n + a_{n-1} = 0

For r = 0 this reads a_n = -a_{n-1} / (n (2n - 1)). For r = 1/2 it reads a_n = -a_{n-1} / (n (2n + 1)). These two series happen to be cos(sqrt(2x)) and sin(sqrt(2x)), as the next demonstration shows.

Python Frobenius series versus closed forms

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Both match. Note how the solution sin(sqrt(2x)) behaves like sqrt(x) near 0, which no ordinary power series could represent. That is why the extra x^r matters.

Computing the indicial equation automatically

You can let SymPy do the algebra. Given p0 and q0 the indicial polynomial is r(r-1) + p0 r + q0. Try several equations and see which case they fall into.

Python Indicial roots and their case

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Euler equations are the model case

The equation x^2 y'' + p0 x y' + q0 y = 0 has constant p0, q0, and it is solved exactly by y = x^r with r a root of the indicial equation. Every Frobenius problem is, near x = 0, a perturbation of such an Euler equation, which explains why the indicial equation controls the leading behavior. When the roots are equal, y = x^r ln x appears, exactly as in the repeated-root case for constant-coefficient equations.

Where the series converges

Frobenius series converge at least out to the nearest other singularity of the coefficients. For Bessel's equation, the only singularity is x = 0, so the series converges for all x.

Common mistakes

  • Starting with a_0 = 0. The point of the indicial equation is that a_0 โ‰  0.
  • Reading p0 and q0 before dividing the equation by its leading coefficient, so they come out wrong.
  • Forgetting the logarithm in the equal-root case.
  • Applying the method at an irregular singular point (for instance x^3 y'' + y = 0 near 0).

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. The indicial equation of x^2 y'' + x y' - 9 y = 0 is:

    Here p0 = 1 and q0 = -9, so r(r-1) + r - 9 = r^2 - 9 = 0, giving r = ยฑ3.
  2. 2. When the indicial roots are equal, the second solution contains:

    Like repeated roots for constant coefficients, equal Frobenius exponents force a logarithm.
  3. 3. A point x = 0 is a regular singular point of y'' + P y' + Q y = 0 when:

    Regular singular points allow a simple pole in P and a double pole in Q, no worse.
  4. 4. Why do we assume a0 is nonzero?

    The lowest-power coefficient is a0 times the indicial polynomial; with a0 โ‰  0 it forces the indicial polynomial to vanish.

Summary

  • At a regular singular point, use y = x^r ฮฃ a_n x^n with a_0 โ‰  0.
  • The lowest power gives the indicial equation r(r-1) + p0 r + q0 = 0.
  • Distinct non-integer-spaced roots give two Frobenius series; equal or integer-spaced roots may require a logarithm.
  • Euler equations are the exactly solvable model for the leading behavior.
  • Bessel and Legendre equations are the famous applications.

Next: Bessel and Legendre Functions