Merge pull request #216 from SciML/compathelper/new_version/2021-06-0…

…3-02-42-43-659-1397580942 CompatHelper: bump compat for "Turing" to "0.16"
SciML · Jul 12, 2021 · ea7fa7c · ea7fa7c
2 parents 1b95523 + 9ed7f80
commit ea7fa7c
Show file tree

Hide file tree

Showing 7 changed files with 113 additions and 127 deletions.
diff --git a/Project.toml b/Project.toml
@@ -4,7 +4,6 @@ authors = ["Vaibhavdixit02 <vaibhavyashdixit@gmail.com>"]
 version = "2.25.0"
 
 [deps]
-ApproxBayes = "f5f396d3-230c-5e07-80e6-9fadf06146cc"
 DiffEqBase = "2b5f629d-d688-5b77-993f-72d75c75574e"
 DiffResults = "163ba53b-c6d8-5494-b064-1a9d43ac40c5"
 Distances = "b4f34e82-e78d-54a5-968a-f98e89d6e8f7"
@@ -26,12 +25,12 @@ Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 RecursiveArrayTools = "731186ca-8d62-57ce-b412-fbd966d074cd"
 Reexport = "189a3867-3050-52da-a836-e630ba90ab69"
 Requires = "ae029012-a4dd-5104-9daa-d747884805df"
+StanSample = "c1514b29-d3a0-5178-b312-660c88baa699"
 StructArrays = "09ab397b-f2b6-538f-b94a-2f83cf4a842a"
 TransformVariables = "84d833dd-6860-57f9-a1a7-6da5db126cff"
 Turing = "fce5fe82-541a-59a6-adf8-730c64b5f9a0"
 
 [compat]
-ApproxBayes = "0.3"
 DiffEqBase = "6.36"
 DiffResults = "0.0.4, 1.0"
 Distances = "0.8, 0.9, 0.10"
@@ -46,14 +45,13 @@ Missings = "0.4, 1.0"
 ModelingToolkit = "5.6"
 Optim = "0.19, 0.20, 0.21, 0.22, 1.0"
 PDMats = "0.9, 0.10, 0.11"
-ParameterizedFunctions = "5"
 Parameters = "0.12"
 RecursiveArrayTools = "1,2"
 Reexport = "0.2, 1.0"
 Requires = "0.5, 1.0"
 StructArrays = "0.4, 0.5"
 TransformVariables = "0.3, 0.4"
-Turing = "0.12, 0.13, 0.14, 0.15"
+Turing = "0.12, 0.13, 0.14, 0.15, 0.16"
 julia = "1.3"
 
 [extras]

diff --git a/src/DiffEqBayes.jl b/src/DiffEqBayes.jl
@@ -7,29 +7,23 @@ using DocStringExtensions
 using DiffEqBase, Distributions, Turing, MacroTools
 using RecursiveArrayTools, ModelingToolkit
 using Parameters, Distributions, Optim, Requires
-using Distances, ApproxBayes, DocStringExtensions, Random
+using Distances, DocStringExtensions, Random, StanSample
 
 STANDARD_PROB_GENERATOR(prob,p) = remake(prob;u0=eltype(p).(prob.u0),p=p)
 STANDARD_PROB_GENERATOR(prob::EnsembleProblem,p) = EnsembleProblem(remake(prob.prob;u0=eltype(p).(prob.prob.u0),p=p))
 
 include("turing_inference.jl")
-include("abc_inference.jl")
+# include("abc_inference.jl")
+include("stan_string.jl")
+include("stan_inference.jl")
 
 function __init__()
-    @require CmdStan="593b3428-ca2f-500c-ae53-031589ec8ddd" begin
-        using .CmdStan
-        include("stan_inference.jl")
-        include("stan_string.jl")
-        export stan_inference, stan_string
-    end
-
     @require DynamicHMC="bbc10e6e-7c05-544b-b16e-64fede858acb" begin
         using .DynamicHMC, TransformVariables, LogDensityProblems
         include("dynamichmc_inference.jl")
         export dynamichmc_inference
     end
 end
 
-export turing_inference, abc_inference
-
+export turing_inference, stan_inference ,abc_inference
 end # module
diff --git a/src/dynamichmc_inference.jl b/src/dynamichmc_inference.jl
@@ -103,5 +103,5 @@ function dynamichmc_inference(problem::DiffEqBase.DEProblem, algorithm, t, data,
     ℓ = TransformedLogDensity(trans, P)
     ∇ℓ = LogDensityProblems.ADgradient(AD_gradient_kind, ℓ)
     results = mcmc_with_warmup(rng, ∇ℓ, num_samples; mcmc_kwargs...)
-    merge((posterior = transform.(Ref(trans), results.chain), ), results)
+    merge((posterior = TransformVariables.transform.(Ref(trans), results.chain), ), results)
 end
diff --git a/src/stan_inference.jl b/src/stan_inference.jl
@@ -1,22 +1,25 @@
-struct StanModel{M,R,C,N}
+struct StanResult{M,R,C}
   model::M
   return_code::R
   chains::C
-  cnames::N
+end
+
+function Base.show(io::IO, mime::MIME"text/plain", res::StanResult)
+  show(io, mime, res.chains)
 end
 
 struct StanODEData
 end
 
 function generate_priors(n,priors)
   priors_string = ""
-  if priors==nothing
+  if priors===nothing
     for i in 1:n
-      priors_string = string(priors_string,"theta[$i] ~ normal(0, 1)", " ; ")
+      priors_string = string(priors_string,"theta_$i ~ normal(0, 1)", " ; ")
     end
   else
     for i in 1:n
-      priors_string = string(priors_string,"theta[$i] ~",stan_string(priors[i]),";")
+      priors_string = string(priors_string,"theta_$i ~ ",stan_string(priors[i]),";")
     end
   end
   priors_string
@@ -34,13 +37,13 @@ function generate_theta(n,priors)
       lower_bound = string("lower=",minimum(priors[i]))
     end
     if lower_bound != "" && upper_bound != ""
-      theta = string(theta,"real","<$lower_bound",",","$upper_bound>"," theta$i",";")
+      theta = string(theta,"real","<$lower_bound",",","$upper_bound>"," theta_$i",";")
     elseif lower_bound != ""
-      theta = string(theta,"real","<$lower_bound",">"," theta$i",";")
+      theta = string(theta,"real","<$lower_bound",">"," theta_$i",";")
     elseif upper_bound != ""
-      theta = string(theta,"real","<","$upper_bound>"," theta$i",";")
+      theta = string(theta,"real","<","$upper_bound>"," theta_$i",";")
     else
-      theta = string(theta,"real"," theta$i",";")
+      theta = string(theta,"real"," theta_$i",";")
     end
   end
   return theta
@@ -50,9 +53,10 @@ function stan_inference(prob::DiffEqBase.DEProblem,t,data,priors = nothing,
                             stanmodel = nothing;alg=:rk45,
                             num_samples=1000, num_warmup=1000, reltol=1e-3,
                             abstol=1e-6, maxiter=Int(1e5),likelihood=Normal,
-                            vars=(StanODEData(),InverseGamma(3,3)),nchains=1,
-                            sample_u0 = false, save_idxs = nothing, diffeq_string = nothing, printsummary = true)
-
+                            vars=(StanODEData(),InverseGamma(3,3)),nchains=[1],
+                            sample_u0 = false, save_idxs = nothing, diffeq_string = nothing, 
+                            printsummary = true, output_format = :mcmcchains)
+
   save_idxs !== nothing && length(save_idxs) == 1 ? save_idxs = save_idxs[1] : save_idxs = save_idxs
   length_of_y = length(prob.u0)
   save_idxs = something(save_idxs, 1:length_of_y)
@@ -63,24 +67,26 @@ function stan_inference(prob::DiffEqBase.DEProblem,t,data,priors = nothing,
   else
     length_of_parameter = length(prob.p) + sample_u0 * length(save_idxs)
   end
-  
+
   if stanmodel === nothing
     if alg ==:adams
-      algorithm = "integrate_ode_adams"
+      algorithm = "ode_adams_tol"
     elseif alg ==:rk45
-      algorithm = "integrate_ode_rk45"
+      algorithm = "ode_rk45_tol"
     elseif alg == :bdf
-      algorithm = "integrate_ode_bdf"
+      algorithm = "ode_bdf_tol"
     else
       error("The choices for alg are :adams, :rk45, or :bdf")
     end
     hyper_params = ""
     tuple_hyper_params = ""
     setup_params = ""
     thetas = ""
+    theta_names = ""
     theta_string = generate_theta(length_of_parameter,priors)
     for i in 1:length_of_parameter
-      thetas = string(thetas,"theta[$i] = theta$i",";")
+      thetas = string(thetas,"real theta_$i",";")
+      theta_names = string(theta_names,"theta_$i",",")
     end
     for i in 1:length_of_params
       if isa(vars[i],StanODEData)
@@ -97,18 +103,18 @@ function stan_inference(prob::DiffEqBase.DEProblem,t,data,priors = nothing,
     stan_likelihood = stan_string(likelihood)
     if sample_u0
       nu = length(save_idxs)
+      dv_names_ind = findfirst("$nu", theta_names)[1]
       if nu < length(prob.u0)
-        u0 = "{"
+        u0 = ""
         for u_ in prob.u0[nu+1:length(prob.u0)]
           u0 = u0*string(u_)
         end
-        u0 = u0*"}"
-        integral_string = "u_hat = $algorithm(sho, append_array(theta[1:$nu],$u0), t0, ts, theta[$(nu+1):$length_of_parameter], x_r, x_i, $reltol, $abstol, $maxiter);"
-      else 
-        integral_string = "u_hat = $algorithm(sho, theta[1:$nu], t0, ts, theta[$(nu+1):$length_of_parameter], x_r, x_i, $reltol, $abstol, $maxiter);"
+        integral_string = "u_hat = $algorithm(sho, [$(theta_names[1:dv_names_ind]),$u0]', t0, ts, $reltol, $abstol, $maxiter, $(rstrip(theta_names[dv_names_ind+2:end],',')));"
+      else
+        integral_string = "u_hat = $algorithm(sho, [$(theta_names[1:dv_names_ind])]', t0, ts, $reltol, $abstol, $maxiter, $(rstrip(theta_names[dv_names_ind+2:end],',')));"
       end
     else
-      integral_string = "u_hat = $algorithm(sho, u0, t0, ts, theta, x_r, x_i, $reltol, $abstol, $maxiter);"
+      integral_string = "u_hat = $algorithm(sho, u0, t0, ts, $reltol, $abstol, $maxiter, $(rstrip(theta_names,',')));"
     end
     binsearch_string = """
       int bin_search(real x, int min_val, int max_val){
@@ -120,8 +126,8 @@ function stan_inference(prob::DiffEqBase.DEProblem,t,data,priors = nothing,
                 out = mid_pt;
                 range = 0;
             } else {
-                range = (range + 1) / 2; 
-                mid_pt = x > mid_pt ? mid_pt + range: mid_pt - range; 
+                range = (range + 1) / 2;
+                mid_pt = x > mid_pt ? mid_pt + range: mid_pt - range;
             }
         }
         return out;
@@ -141,26 +147,18 @@ function stan_inference(prob::DiffEqBase.DEProblem,t,data,priors = nothing,
       $diffeq_string
     }
     data {
-      real u0[$length_of_y];
+      vector[$length_of_y] u0;
       int<lower=1> T;
       real internal_var___u[T,$(length(save_idxs))];
       real t0;
       real ts[T];
     }
-    transformed data {
-      real x_r[0];
-      int x_i[0];
-    }
     parameters {
       $setup_params
       $theta_string
     }
-    transformed parameters{
-      real theta[$length_of_parameter];
-      $thetas
-    }
     model{
-      real u_hat[T,$length_of_y];
+      vector[$length_of_y] u_hat[T];
       $hyper_params
       $priors_string
       $integral_string
@@ -169,9 +167,13 @@ function stan_inference(prob::DiffEqBase.DEProblem,t,data,priors = nothing,
         }
     }
     "
-    stanmodel = CmdStan.Stanmodel(num_samples=num_samples, num_warmup=num_warmup, name="parameter_estimation_model", model=parameter_estimation_model, nchains=nchains, printsummary = printsummary)
+    stanmodel = StanSample.SampleModel("parameter_estimation_model", parameter_estimation_model, nchains; printsummary = printsummary, method = StanSample.Sample(;num_samples = num_samples, num_warmup = num_warmup))
   end
   parameter_estimation_data = Dict("u0"=>prob.u0, "T" => length(t), "internal_var___u" => view(data, :, 1:length(t))', "t0" => prob.tspan[1], "ts" => t)
-  return_code, chains, cnames = CmdStan.stan(stanmodel, [parameter_estimation_data])
-  return StanModel(stanmodel, return_code, chains, cnames)
+  rc = stan_sample(stanmodel; data = parameter_estimation_data)
+  if success(rc)
+    return StanResult(stanmodel, rc, read_samples(stanmodel; output_format=output_format))
+  else
+     rc.err
+  end
 end
diff --git a/test/runtests.jl b/test/runtests.jl
@@ -8,12 +8,10 @@ const GROUP = get(ENV, "GROUP", "All")
 if GROUP == "All" || GROUP == "Core"
     @time @safetestset "DynamicHMC" begin include("dynamicHMC.jl") end
     @time @safetestset "Turing" begin include("turing.jl") end
-    @time @safetestset "ABC" begin include("abc.jl") end
+    # @time @safetestset "ABC" begin include("abc.jl") end
 end
 
 if GROUP == "Stan" || GROUP == "All"
-    using Pkg
-    Pkg.add("CmdStan")
     @time @safetestset "Stan_String" begin include("stan_string.jl") end
     @time @safetestset "Stan" begin include("stan.jl") end
 end
diff --git a/test/stan.jl b/test/stan.jl
@@ -1,4 +1,4 @@
-using CmdStan, DiffEqBayes, OrdinaryDiffEq, ParameterizedFunctions,
+using DiffEqBayes, OrdinaryDiffEq, ParameterizedFunctions,
       RecursiveArrayTools, Distributions, Test
 
 println("One parameter case")
@@ -19,24 +19,21 @@ priors = [truncated(Normal(1.5,0.1),1.0,1.8)]
 bayesian_result = stan_inference(prob1,t,data,priors;num_samples=300,
                                  num_warmup=500,likelihood=Normal)
 
-sdf  = CmdStan.read_summary(bayesian_result.model)
-@test sdf[sdf.parameters .== :theta1, :mean][1] ≈ 1.5 atol=3e-1
+@test mean(get(bayesian_result.chains,:theta_1)[1]) ≈ 1.5 atol=3e-1
 
 # Test norecompile
 bayesian_result2 = stan_inference(prob1,t,data,priors,bayesian_result.model;
                                   num_samples=300,num_warmup=500,likelihood=Normal)
 
-sdf  = CmdStan.read_summary(bayesian_result.model)
-@test sdf[sdf.parameters .== :theta1, :mean][1] ≈ 1.5 atol=3e-1
+@test mean(get(bayesian_result2.chains,:theta_1)[1]) ≈ 1.5 atol=3e-1
 
 priors = [truncated(Normal(1.,0.01),0.5,2.0),truncated(Normal(1.,0.01),0.5,2.0),truncated(Normal(1.5,0.01),1.0,2.0)]
 bayesian_result = stan_inference(prob1,t,data,priors;num_samples=300,
                                  num_warmup=500,likelihood=Normal,sample_u0=true)
 
-sdf  = CmdStan.read_summary(bayesian_result.model)
-@test sdf[sdf.parameters .== :theta1, :mean][1] ≈ 1. atol=3e-1
-@test sdf[sdf.parameters .== :theta2, :mean][1] ≈ 1. atol=3e-1
-@test sdf[sdf.parameters .== :theta3, :mean][1] ≈ 1.5 atol=3e-1
+@test mean(get(bayesian_result.chains,:theta_1)[1]) ≈ 1. atol=3e-1
+@test mean(get(bayesian_result.chains,:theta_2)[1]) ≈ 1. atol=3e-1
+@test mean(get(bayesian_result.chains,:theta_3)[1]) ≈ 1.5 atol=3e-1
 
 sol = solve(prob1,Tsit5(),save_idxs=[1])
 randomized = VectorOfArray([(sol(t[i]) + .01 * randn(1)) for i in 1:length(t)])
@@ -45,17 +42,15 @@ priors = [truncated(Normal(1.5,0.1),0.5,2)]
 bayesian_result = stan_inference(prob1,t,data,priors;num_samples=300,
                                  num_warmup=500,likelihood=Normal,save_idxs=[1])
 
-sdf  = CmdStan.read_summary(bayesian_result.model)
-@test sdf[sdf.parameters .== :theta1, :mean][1] ≈ 1.5 atol=3e-1
+@test mean(get(bayesian_result.chains,:theta_1)[1]) ≈ 1.5 atol=3e-1
 
 
 priors = [truncated(Normal(1.,0.01),0.5,2),truncated(Normal(1.5,0.01),0.5,2)]
 bayesian_result = stan_inference(prob1,t,data,priors;num_samples=300,
                                  num_warmup=500,likelihood=Normal,save_idxs=[1],sample_u0=true)
 
-sdf  = CmdStan.read_summary(bayesian_result.model)
-@test sdf[sdf.parameters .== :theta1, :mean][1] ≈ 1. atol=3e-1
-@test sdf[sdf.parameters .== :theta2, :mean][1] ≈ 1.5 atol=3e-1
+@test mean(get(bayesian_result.chains,:theta_1)[1]) ≈ 1. atol=3e-1
+@test mean(get(bayesian_result.chains,:theta_2)[1]) ≈ 1.5 atol=3e-1
 
 println("Four parameter case")
 f1 = @ode_def begin
@@ -74,8 +69,7 @@ priors = [truncated(Normal(1.5,0.01),0.5,2),truncated(Normal(1.0,0.01),0.5,1.5),
           truncated(Normal(3.0,0.01),0.5,4),truncated(Normal(1.0,0.01),0.5,2)]
 
 bayesian_result = stan_inference(prob1,t,data,priors;num_samples=100,num_warmup=500,vars =(DiffEqBayes.StanODEData(),InverseGamma(4,1)))
-sdf  = CmdStan.read_summary(bayesian_result.model)
-@test sdf[sdf.parameters .== :theta1, :mean][1] ≈ 1.5 atol=1e-1
-@test sdf[sdf.parameters .== :theta2, :mean][1] ≈ 1.0 atol=1e-1
-@test sdf[sdf.parameters .== :theta3, :mean][1] ≈ 3.0 atol=1e-1
-@test sdf[sdf.parameters .== :theta4, :mean][1] ≈ 1.0 atol=1e-1
+@test mean(get(bayesian_result.chains,:theta_1)[1]) ≈ 1.5 atol=1e-1
+@test mean(get(bayesian_result.chains,:theta_2)[1]) ≈ 1.0 atol=1e-1
+@test mean(get(bayesian_result.chains,:theta_3)[1]) ≈ 3.0 atol=1e-1
+@test mean(get(bayesian_result.chains,:theta_4)[1]) ≈ 1.0 atol=1e-1