Partial Differential Equations

Differential equations are fun!

Forward Time Centered Space

For dfdt=vdfdx, we write down the finite difference form 1

f(tn+1,xi)f(tn,xi)Δt=vf(tn,xi+1)f(tn,xi1)2Δx.

FTCS is an explicit method and is not stable.

Lax Method

Change the term f(tn,xi) in FTCS to (f(tn,xi+1)+f(tn,xi1))/2 1.

Stability condition is

|v|ΔtΔx1,

which is the Courant-Fridriches-Lewy stability criterion.

Staggered Leapfrog

f(tn+1,xi)f(tn1,xi)2Δt=vf(tn,xi+1)f(tn,xi1)2Δx

It’s kind of a Centered Space Centered Time method.

Two-Step Lax-Wendroff Scheme

Fully Implicit

f(tn+1,xi)f(tn,xi)Δt=vf(tn+1,xi+1)f(tn+1,xi1)2Δx.

It is called implicity because we can not simply iterate over the formula to get the solutions as like for the explicit method.

Crank-Nicholson

Crank-Nicholson is a average of the explicit and fully implicit method.

f(tn+1,xi)f(tn,xi)Δt=v2(f(tn+1,xi+1)f(tn+1,xi1))+(f(tn,xi+1)f(tn,xi1))2Δx.

References and Notes


  1. Numerical Recipes in C ↩︎ ↩︎

Planted: by ;

L Ma (2018). 'Partial Differential Equations', Datumorphism, 11 April. Available at: https://datumorphism.leima.is/wiki/dynamical-system/partial-difference-method/.