62
mV. Here A and B are pointwise functions whose actions are defined, recall (2.18), by
A(v)i = uλ,oo(vi), B(v)i =-r(vi), z = l,..., N. (3.4)
We discretize (3.2) (3.3) following the second-order staggered Euler scheme of
(Hines, 1984). More precisely, for a fixed time-step ∆i, we evaluate
ɑɑ) = G((J-3∕2)∆i), w^ɔ ≈ w((j-3∕2)∆f) and vɑɔ ≈ v((j-l)∆t), j = 1,2,...
via the marching scheme: Given wɑ-1) and v^-1∖ evaluate
wɑŋ = [(2B(√j~1>) - ∆i) .w(j-ɪ) + 2A(√j-1))Δt]./ [2B(v°-ŋ) + ∆t] (3.5)
and
vω = 2vmjd - v0~1> (3.6)
where vm∣d is the solution to the linear system
(2∕∕∆i — H + diag(Φ(w^)e + G^))vmjd = 2v^-1^∕∆t + Φ(w^)Ei + G^.Es, (3.7)
where the ‘diag’ operator takes a vector and transforms it into a diagonal matrix. In