Merge pull request #933 from mxpoch/master

PyCMA Wrapper

Author: ChrisRackauckas

.github/workflows/CI.yml

@@ -33,6 +33,7 @@ jobs:
           - OptimizationOptimJL
           - OptimizationOptimisers
           - OptimizationPRIMA
+          - OptimizationPyCMA
           - OptimizationQuadDIRECT
           - OptimizationSciPy
           - OptimizationSpeedMapping

lib/OptimizationPyCMA/CondaPkg.toml

@@ -0,0 +1,3 @@
+[deps]
+matplotlib = ""
+cma = ""

lib/OptimizationPyCMA/Project.toml

@@ -0,0 +1,18 @@
+name = "OptimizationPyCMA"
+uuid = "fb0822aa-1fe5-41d8-99a6-e7bf6c238d3b"
+authors = ["Maximilian Pochapski <[email protected]>"]
+version = "0.1.0"
+
+[deps]
+CondaPkg = "992eb4ea-22a4-4c89-a5bb-47a3300528ab"
+Optimization = "7f7a1694-90dd-40f0-9382-eb1efda571ba"
+PythonCall = "6099a3de-0909-46bc-b1f4-468b9a2dfc0d"
+Reexport = "189a3867-3050-52da-a836-e630ba90ab69"
+Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
+
+[compat]
+CondaPkg = "0.2.29"
+Optimization = "4.4.0"
+PythonCall = "0.9.25"
+Reexport = "1.2.2"
+Test = "1.11.0"

lib/OptimizationPyCMA/src/OptimizationPyCMA.jl

@@ -0,0 +1,152 @@
+module OptimizationPyCMA
+
+using Reexport
+@reexport using Optimization
+using PythonCall, Optimization.SciMLBase
+
+export PyCMAOpt
+
+struct PyCMAOpt end
+
+# importing PyCMA
+const cma = Ref{Py}()
+function get_cma()
+    if !isassigned(cma) || cma[] === nothing
+        cma[] = pyimport("cma")
+    end
+    return cma[]
+end
+
+# Defining the SciMLBase interface for PyCMAOpt
+
+SciMLBase.allowsbounds(::PyCMAOpt) = true
+SciMLBase.supports_opt_cache_interface(opt::PyCMAOpt) = true
+SciMLBase.requiresgradient(::PyCMAOpt) = false
+SciMLBase.requireshessian(::PyCMAOpt) = false
+SciMLBase.requiresconsjac(::PyCMAOpt) = false
+SciMLBase.requiresconshess(::PyCMAOpt) = false
+
+# wrapping Optimization.jl args into a python dict as arguments to PyCMA opts
+function __map_optimizer_args(prob::OptimizationCache, opt::PyCMAOpt;
+        maxiters::Union{Number, Nothing} = nothing,
+        maxtime::Union{Number, Nothing} = nothing,
+        abstol::Union{Number, Nothing} = nothing,
+        reltol::Union{Number, Nothing} = nothing)
+    if !isnothing(reltol)
+        @warn "common reltol is currently not used by $(opt)"
+    end
+
+    mapped_args = Dict(
+        "verbose" => -5,
+        "bounds" => (prob.lb, prob.ub),
+        )
+
+    if !isnothing(abstol)
+        mapped_args["tolfun"] = abstol
+    end
+
+    if !isnothing(reltol)
+        mapped_args["tolfunrel"] = reltol 
+    end
+
+    if !isnothing(maxtime)
+        mapped_args["timeout"] = maxtime
+    end 
+
+    if !isnothing(maxiters)
+        mapped_args["maxiter"] = maxiters
+    end
+
+    return mapped_args
+end
+
+function __map_pycma_retcode(stop_dict::Dict{String, Any})
+    # mapping termination conditions to SciMLBase return codes
+    if any(k ∈ keys(stop_dict) for k in ["ftarget", "tolfun", "tolx"])
+        return ReturnCode.Success
+    elseif any(k ∈ keys(stop_dict) for k in ["maxiter", "maxfevals"])
+        return ReturnCode.MaxIters
+    elseif "timeout" ∈ keys(stop_dict)
+        return ReturnCode.MaxTime
+    elseif "callback" ∈ keys(stop_dict)
+        return ReturnCode.Terminated
+    elseif any(k ∈ keys(stop_dict) for k in ["tolupsigma", "tolconditioncov", "noeffectcoord", "noeffectaxis", "tolxstagnation", "tolflatfitness", "tolfacupx", "tolstagnation"])
+        return ReturnCode.Failure
+    else
+        return ReturnCode.Default
+    end
+end
+
+function SciMLBase.__solve(cache::OptimizationCache{
+        F,
+        RC,
+        LB,
+        UB,
+        LC,
+        UC,
+        S,
+        O,
+        D,
+        P,
+        C
+}) where {
+        F,
+        RC,
+        LB,
+        UB,
+        LC,
+        UC,
+        S,
+        O <:
+        PyCMAOpt,
+        D,
+        P,
+        C
+}
+    local x
+
+    # doing conversions
+    maxiters = Optimization._check_and_convert_maxiters(cache.solver_args.maxiters)
+    maxtime = Optimization._check_and_convert_maxtime(cache.solver_args.maxtime)
+
+    # wrapping the objective function
+    _loss = function (θ)
+        x = cache.f(θ, cache.p)
+        return first(x)
+    end
+
+    # converting the Optimization.jl Args to PyCMA format
+    opt_args = __map_optimizer_args(cache, cache.opt; cache.solver_args...,
+        maxiters = maxiters,
+        maxtime = maxtime)
+    
+    # init the CMAopt class
+    es = get_cma().CMAEvolutionStrategy(cache.u0, 1, pydict(opt_args))
+    logger = es.logger
+
+    # running the optimization
+    t0 = time()
+    opt_res = es.optimize(_loss)
+    t1 = time()
+
+    # loading logged files from disk
+    logger.load()
+
+    # reading the results
+    opt_ret_dict = opt_res.stop()
+    retcode = __map_pycma_retcode(pyconvert(Dict{String, Any}, opt_ret_dict))
+    
+    # logging and returning results of the optimization
+    stats = Optimization.OptimizationStats(;
+        iterations = length(logger.xmean),
+        time = t1 - t0,
+        fevals = length(logger.xmean))
+    
+    SciMLBase.build_solution(cache, cache.opt,
+        pyconvert(Float64, logger.xrecent[-1][-1]),
+        pyconvert(Float64, logger.f[-1][-1]); original = opt_res,
+        retcode = retcode,
+        stats = stats)
+end
+
+end # module OptimizationPyCMA

lib/OptimizationPyCMA/test/runtests.jl

@@ -0,0 +1,14 @@
+using OptimizationPyCMA
+using Test
+
+@testset "OptimizationPyCMA.jl" begin
+    rosenbrock(x, p) = (p[1] - x[1])^2 + p[2] * (x[2] - x[1]^2)^2
+    x0 = zeros(2)
+    _p = [1.0, 100.0]
+    l1 = rosenbrock(x0, _p)
+    f = OptimizationFunction(rosenbrock)
+    prob = OptimizationProblem(f, x0, _p, lb = [-1.0, -1.0], ub = [0.8, 0.8])
+    sol = solve(prob, PyCMAOpt())
+    @test 10 * sol.objective < l1
+    sol = solve(prob, PyCMAOpt(), maxiters = 100)
+end