From 38caca2c1e079a2f97d8d34fcc51abb5145f59df Mon Sep 17 00:00:00 2001
From: Stephan Hilb <stephan@ecshi.net>
Date: Mon, 25 May 2020 17:23:30 +0200
Subject: [PATCH] chambolle: compensate norm(B) in iteration
---
src/chambolle.jl | 12 ++++++++----
1 file changed, 8 insertions(+), 4 deletions(-)
diff --git a/src/chambolle.jl b/src/chambolle.jl
index 0c41a78..0a2cfa0 100644
--- a/src/chambolle.jl
+++ b/src/chambolle.jl
@@ -38,11 +38,11 @@ struct ChambolleContext{M,A,G,Λ,T,R,S,Sv,K1,K2} <: Context
s::S
"dual variable as vector"
rv::Sv
- "scalar temporary"
+ "scalar temporary variable"
sv::Sv
"div(p) + g kernel"
k1::K1
- "(p + τ*grad(q))/(1 + τ/λ|grad(q)|) kernel"
+ "(p + (τ/normB)*grad(q))/(1 + (τ/normB)/λ|grad(q)|) kernel"
k2::K2
end
@@ -61,9 +61,11 @@ function init(md::OpROFModel, alg::ChambolleAlgorithm)
@inline kf1(pw, gw) = @inbounds divergence(pw) + gw[z]
k1 = Kernel{ntuple(_->-1:1, d)}(kf1)
+ normB = norm(md.B)
+
@inline function kf2(pw, sw)
- iszero(λ[pw.position]) && return zero(pw[z])
- sgrad = alg.τ * gradient(sw)
+ @inbounds iszero(λ[pw.position]) && return zero(pw[z])
+ sgrad = (alg.τ / normB) * gradient(sw)
return @inbounds (pw[z] + sgrad) / (1 + norm(sgrad) / λ[pw.position])
end
k2 = Kernel{ntuple(_->0:1, d)}(kf2)
@@ -111,6 +113,8 @@ function recover_u!(ctx)
map!(ctx.k1, ctx.r, ctx.p, ctx.g)
# s = B * r
mul!(ctx.sv, ctx.model.B, ctx.rv)
+
+ return ctx.s
end
#
--
GitLab