Skip to content
Tutorials Center
My learning

Topics

Lesson 27 of 47 · Partial Differential Equations

Chapter 5 review: partial differential equations

A one-picture recap of the heat, wave and Laplace solvers, their stability rules, and a ten-question quiz covering the whole chapter.

You can now solve the three model PDEs on a grid. This review ties them together in one picture and one table, and then tests the details that cause the most mistakes: stability limits, boundary handling and choice of method.

Three strips, one for each PDE, with its equation, scheme and stability rule
Chapter 5 in one picture: match the equation to its scheme and its safety rule.

Recap table

Equation Type Scheme Rule to remember
u_t = α u_xx parabolic FTCS (explicit) r = αΔt/Δx² ≤ 1/2
u_t = α u_xx parabolic Crank-Nicolson (implicit) stable for any r; solve a tridiagonal system
u_tt = c² u_xx hyperbolic leapfrog r = cΔt/Δx ≤ 1; r = 1 is exact
∇²u = 0 elliptic Jacobi, Gauss-Seidel, SOR iterate to a tolerance; ω_opt = 2/(1 + sin(π/(N−1)))

Habits worth keeping

  • Test on a known answer first (decaying sine, d'Alembert pulse, symmetry value 0.25).
  • Check stability before accuracy. A solver that blows up has no accuracy to discuss.
  • Refine and compare. Halve the spacing and see whether the error falls by the expected factor.
  • Match the method to the physics. Information that travels at a finite speed needs a scheme that respects it.

Check your understanding

Quick check · 10 questions

of 10 answered You scored / 10 · perfect!

  1. 1. You use FTCS with α = 1e-4 m²/s and Δx = 0.01 m. What is the largest stable Δt?

    Δt ≤ Δx²/(2α) = 1e-4/(2e-4) = 0.5 s.
  2. 2. Which scheme is exact on the grid for the 1D wave equation when r = 1?

    At r = 1 the leapfrog update reproduces d'Alembert's solution exactly on the grid points.
  3. 3. Which method would you pick for a diffusion problem that must run for a very long physical time?

    An implicit method allows large time steps without losing stability.
  4. 4. Halving Δx while keeping FTCS stable makes the work grow by about...

    Twice as many points, and Δt must drop by 4, so the work grows by 2 × 4 = 8.
  5. 5. Which statement about the Laplace equation is true?

    An elliptic problem is fixed by what happens on the boundary of the region.
  6. 6. Which ω can SOR use and still converge?

    SOR converges for 0 < ω < 2; useful values lie between 1 and 2.
  7. 7. A narrow pulse on a coarse grid with r = 0.5 leaves ripples behind it. This is...

    Short wavelengths travel slightly slower on the grid; the ripples shrink as you refine it.
  8. 8. What does the CFL condition ensure?

    If a wave can move more than one cell per step, the stencil misses information it needs.
  9. 9. The classification B² − 4AC < 0 identifies which type of equation?

    A negative discriminant gives an elliptic equation such as Laplace's.
  10. 10. Why is the maximum principle useful for checking a Laplace solver?

    Any interior value outside the boundary range reveals a bug.

Summary

  • Heat: explicit FTCS needs r ≤ 1/2; implicit schemes remove this limit.
  • Wave: leapfrog needs r ≤ 1; watch for numerical dispersion at coarse resolution.
  • Laplace: relax until the change is tiny; SOR is the fast classic choice.
  • Validate with exact cases and refine to see the error fall.

Next: Chapter overview: molecular dynamics