Merge pull request #202 from lxvm/mcintegration

wrap MCintegration.jl

Project.toml

@@ -19,6 +19,7 @@ Cuba = "8a292aeb-7a57-582c-b821-06e4c11590b1"
 Cubature = "667455a9-e2ce-5579-9412-b964f529a492"
 FastGaussQuadrature = "442a2c76-b920-505d-bb47-c5924d526838"
 ForwardDiff = "f6369f11-7733-5829-9624-2563aa707210"
+MCIntegration = "ea1e2de9-7db7-4b42-91ee-0cd1bf6df167"
 Zygote = "e88e6eb3-aa80-5325-afca-941959d7151f"
 
 [extensions]
@@ -27,6 +28,7 @@ IntegralsCubaExt = "Cuba"
 IntegralsCubatureExt = "Cubature"
 IntegralsFastGaussQuadratureExt = "FastGaussQuadrature"
 IntegralsForwardDiffExt = "ForwardDiff"
+IntegralsMCIntegrationExt = "MCIntegration"
 IntegralsZygoteExt = ["Zygote", "ChainRulesCore"]
 
 [compat]
@@ -42,6 +44,7 @@ FiniteDiff = "2.12"
 ForwardDiff = "0.10.19"
 HCubature = "1.5"
 LinearAlgebra = "1.9"
+MCIntegration = "0.4.2"
 MonteCarloIntegration = "0.0.3, 0.1"
 Pkg = "1"
 QuadGK = "2.9"
@@ -63,11 +66,12 @@ Distributions = "31c24e10-a181-5473-b8eb-7969acd0382f"
 FastGaussQuadrature = "442a2c76-b920-505d-bb47-c5924d526838"
 FiniteDiff = "6a86dc24-6348-571c-b903-95158fe2bd41"
 ForwardDiff = "f6369f11-7733-5829-9624-2563aa707210"
+MCIntegration = "ea1e2de9-7db7-4b42-91ee-0cd1bf6df167"
 Pkg = "44cfe95a-1eb2-52ea-b672-e2afdf69b78f"
 SafeTestsets = "1bc83da4-3b8d-516f-aca4-4fe02f6d838f"
 StaticArrays = "90137ffa-7385-5640-81b9-e52037218182"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 Zygote = "e88e6eb3-aa80-5325-afca-941959d7151f"
 
 [targets]
-test = ["Aqua", "Arblib", "StaticArrays", "FiniteDiff", "Pkg", "SafeTestsets", "Test", "Distributions", "ForwardDiff", "Zygote", "ChainRulesCore", "FastGaussQuadrature", "Cuba", "Cubature"]
+test = ["Aqua", "Arblib", "StaticArrays", "FiniteDiff", "Pkg", "SafeTestsets", "Test", "Distributions", "ForwardDiff", "Zygote", "ChainRulesCore", "FastGaussQuadrature", "Cuba", "Cubature", "MCIntegration"]

docs/src/solvers/IntegralSolvers.md

@@ -4,7 +4,9 @@ The following algorithms are available:
 
  - `QuadGKJL`: Uses QuadGK.jl. Requires `nout=1` and `batch=0`, in-place is not allowed.
  - `HCubatureJL`: Uses HCubature.jl. Requires `batch=0`.
- - `VEGAS`: Uses MonteCarloIntegration.jl. Requires `nout=1`. Works only for `>1`-dimensional integrations.
+ - `VEGAS`: Uses MonteCarloIntegration.jl. Requires `nout=1`. Works only for
+ `>1`-dimensional integrations.
+ - `VEGASMC`: Uses MCIntegration.jl. Requires `using MCIntegration`. Doesn't support batching.
  - `CubatureJLh`: h-Cubature from Cubature.jl. Requires `using Cubature`.
  - `CubatureJLp`: p-Cubature from Cubature.jl. Requires `using Cubature`.
  - `CubaVegas`: Vegas from Cuba.jl. Requires `using Cuba`, `nout=1`.
@@ -20,6 +22,7 @@ The following algorithms are available:
 QuadGKJL
 HCubatureJL
 VEGAS
+VEGASMC
 CubaVegas
 CubaSUAVE
 CubaDivonne

ext/IntegralsMCIntegrationExt.jl

@@ -0,0 +1,48 @@
+module IntegralsMCIntegrationExt
+
+using MCIntegration, Integrals
+
+function Integrals.__solvebp_call(prob::IntegralProblem, alg::VEGASMC, sensealg, domain, p;
+ reltol = nothing, abstol = nothing, maxiters = 1000)
+ lb, ub = domain
+ mid = vec(collect((lb + ub) / 2))
+ vars = Continuous(vec([tuple(a,b) for (a,b) in zip(lb, ub)]))
+
+ if prob.f isa BatchIntegralFunction
+ error("VEGASMC doesn't support batching. See https://github.com/numericalEFT/MCIntegration.jl/issues/29")
+ else
+ if isinplace(prob)
+ f0 = similar(prob.f.integrand_prototype)
+ f_ = (x, f, c) -> begin
+ n = 0
+ for v in x
+ mid[n+=1] = first(v)
+ end
+ prob.f(f0, mid, p)
+ f .= vec(f0)
+ end
+ else
+ f0 = prob.f(mid, p)
+ f_ = (x, c) -> begin
+ n = 0
+ for v in x
+ mid[n+=1] = first(v)
+ end
+ fx = prob.f(mid, p)
+ fx isa AbstractArray ? vec(fx) : fx
+ end
+ end
+ dof = ones(Int, length(f0)) # each composite Continuous var gets 1 dof
+ res = integrate(f_, inplace=isinplace(prob), var=vars, dof=dof, solver=:vegasmc,
+ neval=alg.neval, niter=min(maxiters,alg.niter), block=alg.block, adapt=alg.adapt,
+ gamma=alg.gamma, verbose=alg.verbose, debug=alg.debug, type=eltype(f0), print=-2)
+ out, err, chi = if f0 isa Number
+ map(only, (res.mean, res.stdev, res.chi2))
+ else
+ map(a -> reshape(a, size(f0)), (res.mean, res.stdev, res.chi2))
+ end
+ SciMLBase.build_solution(prob, VEGASMC(), out, err, chi=chi, retcode = ReturnCode.Success)
+ end
+end
+
+end

src/Integrals.jl

@@ -217,7 +217,8 @@ function __solvebp_call(prob::IntegralProblem, alg::VEGAS, sensealg, domain, p;
  SciMLBase.build_solution(prob, alg, val, err, chi = chi, retcode = ReturnCode.Success)
 end
 
-export QuadGKJL, HCubatureJL, VEGAS, GaussLegendre, QuadratureRule, TrapezoidalRule
+
+export QuadGKJL, HCubatureJL, VEGAS, VEGASMC, GaussLegendre, QuadratureRule, TrapezoidalRule
 export CubaVegas, CubaSUAVE, CubaDivonne, CubaCuhre
 export CubatureJLh, CubatureJLp
 export ArblibJL

src/algorithms.jl

@@ -370,3 +370,20 @@ end
 function ArblibJL(; check_analytic=false, take_prec=false, warn_on_no_convergence=false, opts=C_NULL)
  return ArblibJL(check_analytic, take_prec, warn_on_no_convergence, opts)
 end
+
+"""
+ VEGASMC(; neval=10^4, niter=20, block=16, adapt=true, gamma=1.0, verbose=-2, debug=false)
+
+Markov-chain based Vegas algorithm from MCIntegration.jl
+"""
+struct VEGASMC <: SciMLBase.AbstractIntegralAlgorithm
+ neval::Int
+ niter::Int
+ block::Int
+ adapt::Bool
+ gamma::Float64
+ verbose::Int
+ debug::Bool
+end
+VEGASMC(; neval=10^4, niter=20, block=16, adapt=true, gamma=1.0, verbose=-2, debug=false) =
+ VEGASMC(neval, niter, block, adapt, gamma, verbose, debug)

test/interface_tests.jl

@@ -1,22 +1,22 @@
 using Integrals
-using Cuba, Cubature, Arblib
+using Cuba, Cubature, Arblib, MCIntegration
 using Test
 
 max_dim_test = 2
 max_nout_test = 2
 reltol = 1e-3
 abstol = 1e-3
 
-
-algs = [QuadGKJL, HCubatureJL, CubatureJLh, CubatureJLp, #VEGAS, #CubaVegas,
- CubaSUAVE, CubaDivonne, CubaCuhre]
+algs = [QuadGKJL, HCubatureJL, CubatureJLh, CubatureJLp, VEGAS, VEGASMC, #CubaVegas,
+ CubaSUAVE, CubaDivonne, CubaCuhre, ArblibJL]
 
 alg_req = Dict(QuadGKJL => (nout = 1, allows_batch = true, min_dim = 1, max_dim = 1,
  allows_iip = true),
  HCubatureJL => (nout = Inf, allows_batch = false, min_dim = 1,
  max_dim = Inf, allows_iip = true),
  VEGAS => (nout = 1, allows_batch = true, min_dim = 2, max_dim = Inf,
  allows_iip = true),
+ VEGASMC => (nout = Inf, allows_batch = false, min_dim = 1, max_dim = Inf, allows_iip = true),
  CubatureJLh => (nout = Inf, allows_batch = true, min_dim = 1,
  max_dim = Inf, allows_iip = true),
  CubatureJLp => (nout = Inf, allows_batch = true, min_dim = 1,