From d9d7d55bdfc68553edcf95da3b993afc35fa4570 Mon Sep 17 00:00:00 2001
From: Stephan Hilb <stephan@ecshi.net>
Date: Mon, 28 Feb 2022 19:44:41 +0100
Subject: [PATCH] update denoise_approximation experiment
---
scripts/run_experiments.jl | 48 +++++++++++++++++++++++---------------
1 file changed, 29 insertions(+), 19 deletions(-)
diff --git a/scripts/run_experiments.jl b/scripts/run_experiments.jl
index 8a316a2..c573932 100644
--- a/scripts/run_experiments.jl
+++ b/scripts/run_experiments.jl
@@ -693,7 +693,7 @@ function estimate_res!(st::L1L2TVState)
fj = mesh.cells[j, cell]
gdofs = space.dofmap[:, 1, cell]
- st.est.data[gdofs] .+= (isSnabla ? hfacet : inv(hfacet)) *
+ st.est.data[gdofs] .+= (isSnabla ? hfacet : inv(hfacet)) *
norm2(facetv[(fi, fj)]) * intel
end
end
@@ -921,8 +921,8 @@ function denoise_pd(ctx)
#est = estimate_error(st),
primal_energy = primal_energy(st))
- #norm_residual_ < params.tol && norm_step_ < params.tol && break
- haskey(params, :tol) && norm_step_ < params.tol && break
+ #norm_residual_ < params.eps_newton && norm_step_ < params.eps_newton && break
+ haskey(params, :eps_newton) && norm_step_ < params.eps_newton && break
haskey(params, :max_iters) && k >= params.max_iters && break
end
@@ -941,7 +941,7 @@ function experiment_convergence_rate(ctx)
algparams = (
alpha1=0., alpha2=30., lambda=1., beta=0.,
gamma1=1e-2, gamma2=1e-3,
- tol = 1e-10,
+ eps_newton = 1e-10,
max_iters = 10000,
)
algctx = Util.Context(ctx; g_arr, mesh, algparams...)
@@ -988,7 +988,7 @@ function experiment_convergence_difference(ctx)
init_zero = true,
alpha1 = 0.5, alpha2 = 0., lambda = 0., beta = 1.,
gamma1 = 1e-7, gamma2 = 1e-7,
- tol = 1e-10,
+ eps_newton = 1e-10,
max_iters = 10000,
)
algctx = Util.Context(ctx; g_arr, mesh, algparams...)
@@ -1024,7 +1024,7 @@ function experiment_convergence_oscillations(ctx)
alpha1 = 0.5, alpha2 = 0., lambda = 1., beta = 1., # oscillating
# DP0 for p1 has less oscillations
gamma1 = 1e-7, gamma2 = 1e-7,
- tol = 1e-10,
+ eps_newton = 1e-10,
max_iters = 10000,
)
algctx = Util.Context(ctx; g_arr, mesh, algparams...)
@@ -1053,8 +1053,8 @@ function denoise_approximation(ctx)
m = 1
T(tdata, u) = u
T(::typeof(adjoint), tdata, v) = v
- S(u, nablau) = u
- #S(u, nablau) = nablau
+ #S(u, nablau) = u
+ S(u, nablau) = nablau'
st = L1L2TVState{m}(ctx.params.mesh;
T, tdata = nothing, S,
@@ -1066,8 +1066,15 @@ function denoise_approximation(ctx)
w = FeFunction(st.u.space, name = "w")
w.data .= 0
- #project_l2_lagrange!(st.g, img)
- interpolate!(st.g, x -> evaluate_bilinear(ctx.params.g_arr, x))
+ #project_image! = project_l2_lagrange!
+ project_image!(f, img) = interpolate!(f, x -> evaluate_bilinear(img, x))
+
+ function interpolate_data!()
+ project_image!(st.g, ctx.params.g_arr)
+ interpolate!(st.g, x -> float(norm(x .- 1., Inf) <= 3/4))
+ end
+
+ interpolate_data!()
st.u.data .= st.g.data
#st.u.data .= rand(size(st.u.data))
@@ -1107,13 +1114,14 @@ function denoise_approximation(ctx)
k = 0
k_refine = 0
- while true && ndofs(st.u.space) < 1000 && k < 300
+ while true && ndofs(st.u.space) < 2000 && k < 300
+ k += 1
+
step!(st)
+
solve_primal!(w, st)
norm_step_ = norm_step(st) * domain_factor
-
- k += 1
norm_residual_ = norm_residual(st) * domain_factor
estimate_res!(st)
@@ -1126,7 +1134,7 @@ function denoise_approximation(ctx)
dual_energy = dual_energy(st),
est = estimate_error(st),
)
- if haskey(ctx.params, :tol) && norm_step_ < ctx.params.tol
+ if haskey(ctx.params, :eps_newton) && norm_step_ < ctx.params.eps_newton
k += 1
k_refine += 1
norm_residual_ = norm_residual(st) * domain_factor
@@ -1151,6 +1159,8 @@ function denoise_approximation(ctx)
fs.est, fs.g, fs.u, fs.p1, fs.p2, fs.du, fs.dp1, fs.dp2)
w = fs.w
+ interpolate_data!()
+
norm_residual_ = norm_residual(st) * domain_factor
estimate_res!(st)
@@ -1173,18 +1183,18 @@ end
function experiment_approximation(ctx)
phi = (sqrt(5) + 1) / 2
- #g_arr = from_img([1. 0.; 0. 0.])
g_arr = from_img([0. 0.; 1. 0.])
+ #g_arr = rand(30, 30)
mesh = init_grid(g_arr;)
df = DataFrame()
denoise_approximation(Util.Context(ctx; name = "test", df,
g_arr, mesh,
- alpha1 = 0., alpha2 = 20., lambda = 1., beta = 0.,
+ alpha1 = 0., alpha2 = 100., lambda = 1., beta = 1e-5,
#alpha1 = 1. - inv(phi), alpha2 = 0., lambda = 0., beta = 1.,
- #alpha1 = 1., alpha2 = 0., lambda = 1., beta = 1e-5,
- gamma1 = 1e-5, gamma2 = 1e-5,
- tol = 1e-10, adaptive = true,
+ #alpha1 = 2., alpha2 = 0., lambda = 1., beta = 1e-5,
+ gamma1 = 1e-4, gamma2 = 1e-4,
+ eps_newton = 1e-5, adaptive = true,
))
end
--
GitLab