Perturbative Theory for PDEs

Perturbative methods are used to tackle PDEs with variable coefficients, difficult geometries, or nonlinearities. The main idea is to introduce a small parameter \( \varepsilon \) multiplying the “hard” part of the equation. This provides a knob to control the irregularity: when \( \varepsilon = 0 \), the PDE reduces to a simpler problem; as \( \varepsilon \) increases toward \(1\), we recover the original PDE.

For example, recall Laplace’s equation:

\nabla^2 u = 0

If we wish to study a more complicated nonlinear PDE such as

\nabla^2 u + u^2 = 0,

our usual linear techniques (e.g., superposition) no longer apply. Instead, introduce a small parameter:

\nabla^2 u + \varepsilon\,u^2 = 0.

When \( \varepsilon = 0 \), we return to Laplace’s equation. We then analyze the problem for \( \varepsilon \in (0,1) \) and finally take \( \varepsilon = 1 \) to connect back to the original nonlinear PDE.

The Power Series Expansion

To solve the perturbed equation, we assume that the solution \( u(x, \varepsilon) \) can be expressed as a power series in \( \varepsilon \):

u(x, \varepsilon) = u_0(x) + \varepsilon u_1(x) + \varepsilon^2 u_2(x) + \dots

By substituting this series into the nonlinear PDE, we can group terms by powers of \( \varepsilon \). For the example \( \nabla^2 u + \varepsilon u^2 = 0 \), this looks like:

\nabla^2(u_0 + \varepsilon u_1 + \dots) + \varepsilon(u_0 + \varepsilon u_1 + \dots)^2 = 0

Solving the Hierarchy

For the equation to hold for any small \( \varepsilon \), the coefficient of each power of \( \varepsilon \) must independently equal zero. This creates a hierarchy of linear equations:

Order \( \varepsilon^0 \): \( \nabla^2 u_0 = 0 \). This is just the original Laplace equation. We solve this first to find the base state.
Order \( \varepsilon^1 \): \( \nabla^2 u_1 + u_0^2 = 0 \). Since we already found \( u_0 \), \( u_0^2 \) is just a known source term. This is now a Poisson equation, which is linear and solvable.
Order \( \varepsilon^2 \): \( \nabla^2 u_2 + 2u_0 u_1 = 0 \). Again, this is a linear equation for \( u_2 \) using the previous results.

By solving these one by one, we build an increasingly accurate approximation of the nonlinear behavior. This method is incredibly powerful because it turns one impossible nonlinear problem into a sequence of many solvable linear problems.

← Back to PDE Home