From 1d798170dcc7a65afa085ce4d24452891b95978a Mon Sep 17 00:00:00 2001
From: Ray Kim <msca8h@naver.com>
Date: Sat, 10 Aug 2024 02:42:07 +0100
Subject: [PATCH] add missing file

---
 src/optimization/optrules.jl | 86 ++++++++++++++++++++++++++++++++++++
 1 file changed, 86 insertions(+)
 create mode 100644 src/optimization/optrules.jl

diff --git a/src/optimization/optrules.jl b/src/optimization/optrules.jl
new file mode 100644
index 00000000..f1c3fa2d
--- /dev/null
+++ b/src/optimization/optrules.jl
@@ -0,0 +1,86 @@
+
+"""
+    DoWG(repsilon = 1e-8)
+
+[DoWG](https://arxiv.org/abs/2305.16284) optimizer. It's only parameter is the 
+initial guess of the Euclidean distance to the optimum repsilon.
+The [DoG](https://arxiv.org/abs/2302.12022) paper recommends 1e-4*(1 + norm(x0)).
+
+# Parameters
+- repsilon: Initial guess of the Euclidean distance between the initial point and
+            the optimum.
+"""
+Optimisers.@def struct DoWG <: Optimisers.AbstractRule
+    repsilon = 1e-8
+end
+
+Optimisers.init(o::DoWG, x::AbstractArray{T}) where {T} = (copy(x), zero(T), T(o.repsilon))
+
+function Optimisers.apply!(::DoWG, state, x::AbstractArray{T}, dx) where {T}
+    x0, v, r = state
+
+    r = max(sqrt(sum(abs2, x - x0)), r)
+    r2 = r * r
+    v = v + r2 * sum(abs2, dx)
+    η = r2 / sqrt(v)
+    dx′ = Optimisers.@lazy dx * η
+    return (x0, v, r), dx′
+end
+
+"""
+    DoG(repsilon = 1e-8)
+
+[DoG](https://arxiv.org/abs/2305.16284) optimizer. It's only parameter is the 
+initial guess of the Euclidean distance to the optimum repsilon.
+The [DoG](https://arxiv.org/abs/2302.12022) paper recommends 1e-4*(1 + norm(x0)).
+
+# Parameters
+- repsilon: Initial guess of the Euclidean distance between the initial point and
+            the optimum.
+"""
+
+Optimisers.@def struct DoG <: Optimisers.AbstractRule
+    repsilon = 1e-8
+end
+
+Optimisers.init(o::DoG, x::AbstractArray{T}) where {T} = (copy(x), zero(T), T(o.repsilon))
+
+function Optimisers.apply!(::DoG, state, x::AbstractArray{T}, dx) where {T}
+    x0, v, r = state
+
+    r = max(sqrt(sum(abs2, x - x0)), r)
+    v = v + sum(abs2, dx)
+    η = r / sqrt(v)
+    dx′ = Optimisers.@lazy dx * η
+    return (x0, v, r), dx′
+end
+
+"""
+    COCOB(α = 100)
+
+[Continuous Coin Betting](https://arxiv.org/abs/1705.07795) optimizer.
+It's only parameter is the maximum change per parameter α, which shouldn't need much tuning.
+The paper suggests α = 100 as a generally default value.
+
+# Parameters
+- alpha (α): Scaling parameter.
+"""
+Optimisers.@def struct COCOB <: Optimisers.AbstractRule
+    alpha = 100
+end
+
+function Optimisers.init(::COCOB, x::AbstractArray{T}) where {T}
+    return (zero(x), zero(x), zero(x), zero(x), copy(x))
+end
+
+function Optimisers.apply!(o::COCOB, state, x::AbstractArray{T}, dx) where {T}
+    α = T(o.alpha)
+    L, G, R, θ, x1 = state
+
+    Optimisers.@.. L = max(L, abs(dx))
+    Optimisers.@.. G = G + abs(dx)
+    Optimisers.@.. R = max(R + (x - x1) * -dx, 0)
+    Optimisers.@.. θ = θ + -dx
+    dx′ = Optimisers.@lazy -(x1 - x) - (θ / (L * max(G + L, α * L)) * (L + R))
+    return (L, G, R, θ, x1), dx′
+end