The Lax-Friedrichs scheme for with is u_j^ n+1 = 12 (u_ j+1 ^n + u_ j-1 ^n ) - C 2…

PHYSICS · Multi-part · Fermi University Physics · Mathematical Methods for Physics · Oscillations, Waves & Acoustics · Wave motion, the wave equation and wave energy

Error, Chance and the Machine, Ch Partial differential equations: stability, convergence and solvers

Problem

The Lax-Friedrichs scheme for ut+cux=0u_t + cu_x = 0 with c>0c > 0 is ujn+1=12(uj+1n+uj−1n)−C2(uj+1n−uj−1n),C=c ΔtΔx.u_j^{n+1} = \tfrac12\bigl(u_{j+1}^n + u_{j-1}^n\bigr) - \tfrac{C}{2}\bigl(u_{j+1}^n - u_{j-1}^n\bigr),\qquad C = \frac{c\,\Delta t}{\Delta x}. When Δx\Delta x and Δt\Delta t are refined together at fixed CC it is consistent, with a residual of order Δx\Delta x. ((a)) Substitute ujn=Gneijθu_j^n = G^ne^{ij\theta} to find G(θ)G(\theta) and ∣G∣2|G|^2. Show that ∣G∣≤1|G| \le 1 for every θ\theta exactly when C≤1C \le 1, and find the smallest ∣G∣|G| at C=0.9C = 0.9. ((b)) Three runs are refined with CC held fixed and Δx,Δt→0\Delta x, \Delta t \to 0 together: forward-time centred-space advection at C=0.01C = 0.01, upwind at C=0.8C = 0.8 and Lax-Friedrichs at C=0.9C = 0.9. For each, say whether it converges to the exact solution at a fixed time TT, and give the reason. ((c)) A colleague runs the centred scheme at C=0.01C = 0.01 with c=1 m s−1c = 1\,\mathrm{m\,s^{-1}} on the grid Δx=0.01 m\Delta x = 0.01\,\mathrm{m}. The worst mode, θ=π/2\theta = \pi/2, grows by only 1.651.65 over T=1 sT = 1\,\mathrm{s}, and concludes that the scheme converges. Find the growth of that mode over T=1 sT = 1\,\mathrm{s} on the grid Δx=2×10−4 m\Delta x = 2\times10^{-4}\,\mathrm{m}, still at C=0.01C = 0.01, and say what the coarse-grid test could not show.

Browse the problem library