ProgressMeter

src-old/MadsEmcee.jl

@@ -23,8 +23,10 @@ Example:
 Mads.emcee(llhood, numwalkers=10, numsamples_perwalker=100, thinning=1)
 ```
 """
-function emcee(llhood::Function, numwalkers::Integer, x0::AbstractArray, numsamples_perwalker::Integer, thinning::Integer, a::Number=2.)
-	@assert length(size(x0)) == 2
+function sample(llhood::Function, numwalkers::Integer, x0::AbstractMatrix, numsamples_perwalker::Integer, thinning::Integer, a::Number=2.; filename::AbstractString="", load::Bool=true, save::Bool=true, rng::Random.AbstractRNG=Random.GLOBAL_RNG)
+	if filename != "" && isfile(filename) && load
+		chain, llhoodvals = JLD2.load(filename, "chain", "llhoods")
+	end
 	x = copy(x0)
 	chain = Array{Float64}(undef, size(x0, 1), numwalkers, div(numsamples_perwalker, thinning))
 	lastllhoodvals = RobustPmap.rpmap(llhood, map(i->x[:, i], 1:size(x, 2)))
@@ -34,18 +36,18 @@ function emcee(llhood::Function, numwalkers::Integer, x0::AbstractArray, numsamp
 	batch1 = 1:div(numwalkers, 2)
 	batch2 = div(numwalkers, 2) + 1:numwalkers
 	divisions = [(batch1, batch2), (batch2, batch1)]
-	@ProgressMeter.showprogress 4 for i = 1:numsamples_perwalker
+	@ProgressMeter.showprogress 1 for i = 1:numsamples_perwalker
 		for ensembles in divisions
 			active, inactive = ensembles
-			zs = map(u->((a - 1) * u + 1)^2 / a, rand(Mads.rng, length(active)))
-			proposals = map(i->zs[i] * x[:, active[i]] + (1 - zs[i]) * x[:, rand(Mads.rng, inactive)], 1:length(active))
+			zs = map(u->((a - 1) * u + 1)^2 / a, rand(rng, length(active)))
+			proposals = map(i->zs[i] * x[:, active[i]] + (1 - zs[i]) * x[:, rand(rng, inactive)], 1:length(active))
 			newllhoods = RobustPmap.rpmap(llhood, proposals)
 			for (j, walkernum) in enumerate(active)
 				z = zs[j]
 				newllhood = newllhoods[j]
 				proposal = proposals[j]
 				logratio = (size(x, 1) - 1) * log(z) + newllhood - lastllhoodvals[walkernum]
-				if log(rand(Mads.rng)) < logratio
+				if log(rand(rng)) < logratio
 					lastllhoodvals[walkernum] = newllhood
 					x[:, walkernum] = proposal
 				end
@@ -56,6 +58,9 @@ function emcee(llhood::Function, numwalkers::Integer, x0::AbstractArray, numsamp
 			end
 		end
 	end
+	if save && filename != ""
+		JLD2.save(filename, "chain", chain, "llhoods", llhoodvals)
+	end
 	return chain, llhoodvals
 end
 

src/MadsMonteCarlo.jl

@@ -4,6 +4,7 @@ import JSON
 import AffineInvariantMCMC
 import DocumentFunction
 import BlackBoxOptim
+import ProgressMeter
 import Random
 
 function p10_p50_p90(x::AbstractMatrix)
@@ -113,11 +114,13 @@ function emceesampling(madsdata::AbstractDict, p0::AbstractMatrix; filename::Abs
 		chain, llhoods, observations = loadecmeeresults(madsdata, filename)
 		if isnothing(chain)
 			if execute
-				@info("AffineInvariantMCMC will be executed!")
+				@info("AffineInvariantMCMC will be executed! No preexisting data to load ...")
 			else
+				@warn("No preexisting data to load! AffineInvariantMCMC will be not executed!")
 				return nothing, nothing, nothing
 			end
 		else
+			@info("AffineInvariantMCMC preexisting data loaded!")
 			return chain, llhoods, observations
 		end
 	end
@@ -132,11 +135,13 @@ function emceesampling(madsdata::AbstractDict, p0::AbstractMatrix; filename::Abs
 	if newnsteps < 2
 		newnsteps = 2
 	end
+	@info("AffineInvariantMCMC burning stage ...")
 	burninchain, _ = AffineInvariantMCMC.sample(arrayloglikelihood, numwalkers, p0, newnsteps, 1; filename="", rng=Mads.rng, save=false)
 	newnsteps = div(nsteps, numwalkers)
 	if newnsteps < 2
 		newnsteps = 2
 	end
+	@info("AffineInvariantMCMC exploration stage ...")
 	chain, llhoods = AffineInvariantMCMC.sample(arrayloglikelihood, numwalkers, burninchain[:, :, end], newnsteps, thinning; filename="", rng=Mads.rng, save=false)
 	chain, llhoods =  AffineInvariantMCMC.flattenmcmcarray(chain, llhoods)
 	observations = Mads.forward(madsdata, permutedims(chain))

src/MadsSensitivityAnalysis.jl

@@ -505,7 +505,7 @@ function saltellibrute(madsdata::AbstractDict; N::Integer=1000, seed::Integer=-1
 	for i = eachindex(paramkeys)
 		madsinfo("Parameter : $(paramkeys[i])")
 		cond_means = Array{OrderedCollections.OrderedDict}(undef, numoneparamsamples)
-		@ProgressMeter.showprogress 1 "Computing ... "  for j = 1:numoneparamsamples
+		@ProgressMeter.showprogress 1 "Computing ... " for j = 1:numoneparamsamples
 			cond_means[j] = OrderedCollections.OrderedDict()
 			for k = eachindex(obskeys)
 				cond_means[j][obskeys[k]] = 0.