You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
ϕₕ(1) = ϕ[1] # note: "indexing" with (x) is meant to convey physical coordinates different from array indexesϕₕ(1.5) = (ϕ[1] + ϕ[2])/2ϕₕ(2) = ϕ[2]
so
ϕₕ = P * ϕ
where
P = [10; 0.50.5; 01]
Now for restriction, just take the adjoint of P:
ϕᵣ = P'*ϕₕ
so that
ϕᵣ[1] = ϕₕ[1] + ϕₕ[2]/2
ϕᵣ[2] = ϕₕ[2]/2+ ϕₕ[3]
Note that
julia> P'*P
2×2 Matrix{Float64}:1.250.250.251.25
which, up to a scalar multiple (divide by 1.5), is "close to" the identity matrix. (Kind of a blurred version of it.)
For n-fold prolongation, we could just linearly interpolate at ϕₕ(1+1/n), ϕₕ(1+2/n), etc. Its adjoint is (up to a scalar multiple) how one should define n-fold restriction.
I'm not sure if this is possible, but I'm imagining a more efficient version of
imresize(img; ratio=1//n)
The text was updated successfully, but these errors were encountered: