Merge pull request #32 from JuliaAI/dev

ablaom · web-flow · commit 73d73ea434f6 · 2025-05-09T11:19:59.000+12:00
For a 0.1.5 release
diff --git a/.github/workflows/CI.yml b/.github/workflows/CI.yml
@@ -1,49 +1,44 @@
 name: CI
 on:
-  pull_request:
-    branches:
-      - master
-      - dev
   push:
     branches:
-      - master
       - dev
-    tags: '*'
+    tags: ['*']
+  pull_request:
+  workflow_dispatch:
+concurrency:
+  # Skip intermediate builds: always.
+  # Cancel intermediate builds: only if it is a pull request build.
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ startsWith(github.ref, 'refs/pull/') }}
 jobs:
   test:
-    name: Julia ${{ matrix.version }} - ${{ matrix.os }} - ${{ matrix.arch }} - ${{ github.event_name }}
+    name: Julia ${{ matrix.version }} - ${{ matrix.os }} - ${{ matrix.arch }}
     runs-on: ${{ matrix.os }}
+    timeout-minutes: 60
+    permissions: # needed to allow julia-actions/cache to proactively delete old caches that it has created
+      actions: write
+      contents: read
     strategy:
       fail-fast: false
       matrix:
         version:
-          - '1.6'
           - '1'
+          - 'lts'
         os:
           - ubuntu-latest
         arch:
           - x64
     steps:
-      - uses: actions/checkout@v2
-      - uses: julia-actions/setup-julia@v1
+      - uses: actions/checkout@v4
+      - uses: julia-actions/setup-julia@v2
         with:
           version: ${{ matrix.version }}
           arch: ${{ matrix.arch }}
-      - uses: actions/cache@v1
-        env:
-          cache-name: cache-artifacts
-        with:
-          path: ~/.julia/artifacts
-          key: ${{ runner.os }}-test-${{ env.cache-name }}-${{ hashFiles('**/Project.toml') }}
-          restore-keys: |
-            ${{ runner.os }}-test-${{ env.cache-name }}-
-            ${{ runner.os }}-test-
-            ${{ runner.os }}-
+      - uses: julia-actions/cache@v2
       - uses: julia-actions/julia-buildpkg@v1
       - uses: julia-actions/julia-runtest@v1
-        env:
-          JULIA_NUM_THREADS: '2'
       - uses: julia-actions/julia-processcoverage@v1
-      - uses: codecov/codecov-action@v1
+      - uses: codecov/codecov-action@v5
         with:
-          file: lcov.info
+          files: lcov.info
diff --git a/Project.toml b/Project.toml
@@ -1,7 +1,7 @@
 name = "MLJParticleSwarmOptimization"
 uuid = "17a086e9-ed03-4f30-ab88-8b63f0f6126c"
 authors = ["Long Nguyen <lhnguyen.vn@gmail.com> and contributors"]
-version = "0.1.3"
+version = "0.1.5"
 
 [deps]
 Distributions = "31c24e10-a181-5473-b8eb-7969acd0382f"
@@ -12,6 +12,19 @@ Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 
 [compat]
 Distributions = "0.25"
-MLJBase = "0.18, 0.19, 0.20, 0.21"
-MLJTuning = "0.6, 0.7"
-julia = "1.6"
+LinearAlgebra = "1"
+MLJBase = "1"
+MLJTuning = "0.8"
+Random = "1"
+julia = "1"
+
+[extras]
+ComputationalResources = "ed09eef8-17a6-5b46-8889-db040fac31e3"
+Distributed = "8ba89e20-285c-5b6f-9357-94700520ee1b"
+EvoTrees = "f6006082-12f8-11e9-0c9c-0d5d367ab1e5"
+MLJ = "add582a8-e3ab-11e8-2d5e-e98b27df1bc7"
+StableRNGs = "860ef19b-820b-49d6-a774-d7a799459cd3"
+Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
+
+[targets]
+test = ["ComputationalResources", "Distributed", "EvoTrees", "MLJ", "StableRNGs", "Test"]
diff --git a/src/MLJParticleSwarmOptimization.jl b/src/MLJParticleSwarmOptimization.jl
@@ -8,10 +8,9 @@ using MLJTuning
 
 export ParticleSwarm, AdaptiveParticleSwarm
 
-include("swarm.jl")
+include("interface.jl")
 include("parameters.jl")
 include("update.jl")
-include("strategies/abstract.jl")
 include("strategies/basic.jl")
 include("strategies/adaptive.jl")
 
diff --git a/src/interface.jl b/src/interface.jl
@@ -0,0 +1,47 @@
+abstract type AbstractParticleSwarm <: MLJTuning.TuningStrategy end
+
+struct ParticleSwarmState{T, R, P, I}
+    ranges::R
+    parameters::P
+    indices::I
+    X::Matrix{T}
+    V::Matrix{T}
+    pbest_X::Matrix{T}
+    gbest_X::Matrix{T}
+    pbest::Vector{T}
+    gbest::Vector{T}
+end
+
+mutable struct ParticleSwarm <: AbstractParticleSwarm
+    n_particles::Integer
+    w::Float64
+    c1::Float64
+    c2::Float64
+    prob_shift::Float64
+    rng::AbstractRNG
+    # TODO: topology
+end
+
+mutable struct AdaptiveParticleSwarm <: AbstractParticleSwarm
+    n_particles::Integer
+    c1::Float64
+    c2::Float64
+    prob_shift::Float64
+    rng::AbstractRNG
+end
+
+get_n_particles(tuning::AbstractParticleSwarm) = tuning.n_particles
+get_prob_shift(tuning::AbstractParticleSwarm) = tuning.prob_shift
+get_rng(tuning::AbstractParticleSwarm) = tuning.rng
+
+function initialize(r, tuning::AbstractParticleSwarm)
+    return initialize(get_rng(tuning), r, get_n_particles(tuning))
+end
+
+function retrieve!(state::ParticleSwarmState, tuning::AbstractParticleSwarm)
+    return retrieve!(get_rng(tuning), state)
+end
+
+function pbest!(state::ParticleSwarmState, measurements, tuning::AbstractParticleSwarm)
+    return pbest!(state, measurements, get_prob_shift(tuning))
+end
diff --git a/src/strategies/abstract.jl b/src/strategies/abstract.jl
diff --git a/src/strategies/adaptive.jl b/src/strategies/adaptive.jl
@@ -40,15 +40,6 @@ copy of `model`. If the corresponding range has a specified `scale` function,
 then the transformation is applied before the hyperparameter is returned. If
 `scale` is a symbol (eg, `:log`), it is ignored.
 """
-mutable struct AdaptiveParticleSwarm{R<:AbstractRNG} <: AbstractParticleSwarm
-    n_particles::Int
-    c1::Float64
-    c2::Float64
-    prob_shift::Float64
-    rng::R
-end
-
-# Constructor
 
 function AdaptiveParticleSwarm(;
     n_particles=3,
@@ -57,7 +48,7 @@ function AdaptiveParticleSwarm(;
     prob_shift=0.25,
     rng::R=Random.GLOBAL_RNG
 ) where {R}
-    swarm = AdaptiveParticleSwarm{R}(n_particles, c1, c2, prob_shift, rng)
+    swarm = AdaptiveParticleSwarm(n_particles, c1, c2, prob_shift, rng)
     message = MLJTuning.clean!(swarm)
     isempty(message) || @warn message
     return swarm
diff --git a/src/strategies/basic.jl b/src/strategies/basic.jl
@@ -103,15 +103,6 @@ where pₛ is the probability of the sampled hyperparameter value. For more
 information, refer to "A New Discrete Particle Swarm Optimization Algorithm" by
 Strasser, Goodman, Sheppard, and Butcher.
 """
-mutable struct ParticleSwarm{R<:AbstractRNG} <: AbstractParticleSwarm
-    n_particles::Int
-    w::Float64
-    c1::Float64
-    c2::Float64
-    prob_shift::Float64
-    rng::R
-    # TODO: topology
-end
 
 function ParticleSwarm(;
     n_particles=3,
@@ -121,7 +112,7 @@ function ParticleSwarm(;
     prob_shift=0.25,
     rng::R=Random.GLOBAL_RNG
 ) where {R}
-    swarm = ParticleSwarm{R}(n_particles, w, c1, c2, prob_shift, rng)
+    swarm = ParticleSwarm(n_particles, w, c1, c2, prob_shift, rng)
     message = MLJTuning.clean!(swarm)
     isempty(message) || @warn message
     return swarm
diff --git a/src/swarm.jl b/src/swarm.jl
diff --git a/test/Project.toml b/test/Project.toml
diff --git a/test/runtests.jl b/test/runtests.jl
@@ -8,7 +8,7 @@ addprocs(2)
     using ComputationalResources
     using Distributions
     using EvoTrees
-    using MLJBase
+    using MLJ
     using MLJTuning
     using StableRNGs
 end
diff --git a/test/strategies/adaptive.jl b/test/strategies/adaptive.jl
@@ -82,7 +82,7 @@ for acceleration in modes
     @testset "EvoTree Tuning with AdaptiveParticleSwarm and $(typeof(acceleration))" begin
         rng = StableRNG(123)
         features = rand(rng, 10_000) .* 5 .- 2
-        X = MLJBase.table(reshape(features, (size(features)[1], 1)))
+        X = MLJ.table(reshape(features, (size(features)[1], 1)))
         y = sin.(features) .* 0.5 .+ 0.5
         y = EvoTrees.logit(y) + randn(rng, size(y))
         y = EvoTrees.sigmoid(y)
@@ -128,4 +128,4 @@ for acceleration in modes
 end
 
 println("Adaptive PSO losses (see Issue #14):")
-(; modes=modes, losses=losses) |> MLJBase.pretty
+(; modes=modes, losses=losses) |> MLJ.pretty
diff --git a/test/strategies/basic.jl b/test/strategies/basic.jl
@@ -18,7 +18,7 @@ for acceleration in (CPU1(), CPUProcesses(), CPUThreads())
     @testset "EvoTree Tuning with ParticleSwarm and $(typeof(acceleration))" begin
         rng = StableRNG(123)
         features = rand(rng, 10_000) .* 5 .- 2
-        X = MLJBase.table(reshape(features, (size(features)[1], 1)))
+        X = MLJ.table(reshape(features, (size(features)[1], 1)))
         y = sin.(features) .* 0.5 .+ 0.5
         y = EvoTrees.logit(y) + randn(rng, size(y))
         y = EvoTrees.sigmoid(y)
@@ -32,7 +32,7 @@ for acceleration in (CPU1(), CPUProcesses(), CPUThreads())
             tuning=RandomSearch(rng=StableRNG(1234)),
             resampling=CV(nfolds=5, rng=StableRNG(8888)),
             range=[r1, r2],
-            measure=mae,
+            measure=MLJ.mae,
             n=15,
             acceleration=acceleration
         )
