Learn faster with bite‑sized practice that actually sticks.

Sturm–Liouville Problems (via a friendly eigenvalue story)

Second-order differential equations can feel like a maze… until you learn the “map.” The map is: solve the ODE, then let the boundary conditions choose which solutions are allowed. And surprise: those “allowed” choices often come in a discrete list (like notes on a guitar string). That’s the heart of Sturm–Liouville.

---

1) Quick refresher: constant-coefficient ODEs love exponentials

Take the classic constant-coefficient ODE:

$y'' + a y' + b y = 0.$

A reliable trick is to try an exponential:

$y = e^{rx}.$

Then

$y' = r e^{rx}, \quad y'' = r^2 e^{rx}.$

Plugging in gives:

$r^2 e^{rx} + a r e^{rx} + b e^{rx} = 0 \quad \Rightarrow \quad r^2 + a r + b = 0.$

That polynomial is the characteristic equation.

When complex numbers show up (and that’s okay)

If you get roots like

$r = \alpha \pm i\beta,$

then

$e^{(\alpha \pm i\beta)x} = e^{\alpha x}(\cos \beta x \pm i\sin \beta x).$

The key idea: solutions superpose (because the ODE is linear). So your real solution becomes:

$y(x) = e^{\alpha x}\left(C\cos(\beta x) + D\sin(\beta x)\right).$

That’s the whole “complex exponentials → sines/cosines” bridge in one move.

---

2) A canonical boundary value problem: a vibrating-string-style BVP

Now for the star example (this one basically is the poster child for Sturm–Liouville):

$y'' + \lambda y = 0 \quad \text{on } [0,L], \qquad y(0)=0,\; y(L)=0.$

Here (\lambda) is a parameter. The boundary conditions will decide which (\lambda) values allow a nontrivial solution (meaning: not the boring solution (y\equiv 0)).

We split by cases.

Case A: (\lambda = 0)

Then

$y''=0 \Rightarrow y = Ax + B.$

Apply boundaries:

• (y(0)=0 \Rightarrow B=0)
• (y(L)=0 \Rightarrow A L = 0 \Rightarrow A=0)

So only (y\equiv 0). No nontrivial solution here.

Case B: (\lambda < 0)

Write (\lambda = -\mu^2) with (\mu>0). Then

$y'' - \mu^2 y = 0 \Rightarrow y = C e^{\mu x} + D e^{-\mu x}.$

Apply (y(0)=0):

$C + D = 0 \Rightarrow D = -C.$

So

$y(x) = C(e^{\mu x} - e^{-\mu x}) = 2C\sinh(\mu x).$

Apply (y(L)=0):

$2C\sinh(\mu L)=0 \Rightarrow C=0.$

Again only (y\equiv 0). No nontrivial solution for (\lambda<0).

Case C: (\lambda > 0)

Write (\lambda = k^2) with (k>0). Then

$y'' + k^2 y = 0 \Rightarrow y = A\cos(kx) + B\sin(kx).$

Apply (y(0)=0):

$A\cos(0)+B\sin(0)=A=0.$

So

$y(x)=B\sin(kx).$

Apply (y(L)=0):

$B\sin(kL)=0.$

For a nontrivial solution, we need (B\neq 0), so we must have:

$\sin(kL)=0 \Rightarrow kL = n\pi \quad (n=1,2,3,\dots).$

So

$k_n = \frac{n\pi}{L}, \qquad \lambda_n = k_n^2 = \left(\frac{n\pi}{L}\right)^2.$

And the corresponding eigenfunctions (up to scaling) are:

$y_n(x) = \sin\left(\frac{n\pi x}{L}\right).$

Big takeaway: the boundary conditions act like a “filter” that allows only a discrete set of (\lambda) values:

$\{\lambda_n\}_{n=1}^\infty = \left\{\left(\frac{n\pi}{L}\right)^2\right\}.$

Tiny visual prompt (optional, but clarifying)

Imagine (or sketch if you like) the first three shapes:

• (y_1): one smooth hump (no interior zeros)
• (y_2): two humps (one interior zero at (x=L/2))
• (y_3): three humps (two interior zeros)

Each higher (n) wiggles more.

---

3) The general idea: Sturm–Liouville form

That example isn’t a one-off. It’s part of a whole family called Sturm–Liouville problems, which (in one common form) look like:

$\big(p(x) y'(x)\big)' + \big(\lambda\, w(x) - q(x)\big) y(x) = 0,$

on an interval ([a,b]), plus some boundary conditions (often things like (y(a)=0), (y(b)=0), or more general linear conditions).

What are these pieces?

• $p(x)$: a coefficient (usually positive) that can vary with (x)
• $q(x)$: another coefficient (a “potential-like” term)
• $w(x)$: the weight function (usually positive)
• $\lambda$: the eigenvalue parameter (the thing we’re “tuning”)

In our earlier example:

• (p(x)=1)
• (q(x)=0)
• (w(x)=1)
• interval ([0,L])

So it fits perfectly.

Weighted inner product and orthogonality

Sturm–Liouville problems come with a natural notion of “dot product,” called a weighted inner product:

$\langle f, g\rangle = \int_a^b f(x)\, g(x)\, w(x)\, dx.$

(For real-valued functions we usually don’t need complex conjugates; if functions are complex, you’d use (\overline{f},g).)

A beautiful result: eigenfunctions from different eigenvalues are orthogonal under this weighted inner product:

If (\lambda_m \neq \lambda_n), then

$\langle y_m, y_n\rangle = \int_a^b y_m(x)\, y_n(x)\, w(x)\, dx = 0.$

For the string example with (w(x)=1), this says:

$\int_0^L \sin\left(\frac{m\pi x}{L}\right)\sin\left(\frac{n\pi x}{L}\right)dx = 0 \quad (m\neq n).$

So these sine functions behave like perpendicular vectors—just in “function space.”

---

4) Bridge to quantum mechanics: self-adjoint operators are the grown-up version of Hermitian matrices

In quantum mechanics, lots of problems boil down to an eigenvalue equation for an operator (a machine that eats a function and spits out another function). The magic phrase you’ll hear is self-adjoint (closely related to “Hermitian”).

• A Hermitian matrix has:

  • real eigenvalues
  • orthogonal eigenvectors

• A self-adjoint differential operator (like the Sturm–Liouville operator) similarly has:

  • real eigenvalues
  • orthogonal eigenfunctions (with the right inner product, often weighted)

So you can think of Sturm–Liouville theory as the function-space version of the linear algebra fact:

“Nice symmetric/Hermitian things have nice real spectra and perpendicular eigen-directions.”

That’s why Sturm–Liouville shows up everywhere: heat flow, waves, and especially the Schrödinger equation, where allowed energies become a discrete set of eigenvalues under boundary/normalization conditions.

---

Takeaway

Constant-coefficient ODEs teach you how to build solution families; boundary conditions then select which members survive. In Sturm–Liouville problems, that selection often produces a discrete spectrum ({\lambda_n}), with eigenfunctions that are orthogonal under a weighted inner product. It’s like linear algebra—but with functions—so the same “Hermitian → real + orthogonal” vibe carries straight into quantum mechanics.