Heat Equation: Separation of Variables
Ch 19 — Part IV — PDEs & Fourier Series
Separation of variables assumes \(u(x,t) = X(x) T(t)\), reducing a PDE to ODEs. It is the entry point to Fourier series.
Derivation
For \(u_t = \alpha u_{xx}\) on \(0 < x < L\) with \(u(0,t) = u(L,t) = 0\):
Substitute \(u = X(x) T(t)\): $$X T' = \alpha X'' T\Rightarrow \frac{T'}{\alpha T} = \frac{X''}{X} = -\lambda \quad (\text{constant}).$$ The BCs force \(X(0) = X(L) = 0\), leading to eigenvalues \(\lambda_n = (n\pi/L)^2\) and eigenfunctions \(X_n(x) = \sin(n\pi x / L)\).
Time part
\(T_n' = -\alpha \lambda_n T_n\Rightarrow T_n(t) = e^{-\alpha \lambda_n t}\).
Superposition and initial data
$$u(x,t) = \sum_{n=1}^\infty b_n \sin\!\left(\frac{n\pi x}{L}\right) e^{-\alpha (n\pi/L)^2 t},$$ with coefficients \(b_n\) fixed by expanding the initial data \(u(x,0) = f(x)\) in a sine Fourier series (next concept).
Practice Problems
Hint
Solution
\(u(x,t) = \sin(3x) e^{-9t}\).
Answer: \(\sin(3x) e^{-9t}.\)
Hint
Solution
\(\lambda_n = (n\pi/L)^2\), \(n=0,1,2,\ldots\) (including zero with constant eigenfunction).
Answer: \(\lambda_n = (n\pi/L)^2,\; n \ge 0.\)
Hint
Solution
Mode \(n\) decays like \(e^{-\alpha (n\pi/L)^2 t}\). Larger \(n\) ⇒ faster decay. High frequencies smoothed away, explaining the smoothing effect of diffusion.
Answer: \(n^2\) dependence in exponent.