Merge branch 'master' of https://github.com/madsjulia/BIGUQ.jl

Conflicts:
	src/BIGDT.jl

.travis.yml

@@ -4,6 +4,7 @@ os:
   - linux
 julia:
   - 0.4
+  - 0.5
 notifications:
   email: false
 script:

COPYING

@@ -1,8 +1,7 @@
 MADS: Model Analysis & Decision Support in Julia (Mads.jl v1.0) 2016
 
-http://mads.lanl.gov
-http://madsjulia.lanl.gov
-http://gitlab.com/mads/Mads.jl
+https://mads.lanl.gov
+https://github.com/madsjulia
 
 Licensing: GPLv3: http://www.gnu.org/licenses/gpl-3.0.html
 

REQUIRE

@@ -1,8 +1,8 @@
 julia 0.4
-DataFrames
+RobustPmap
+AffineInvariantMCMC
 Klara 0.6
+DataFrames
 ForwardDiff
 BlackBoxOptim
-Distributions
-RobustPmap
-Mads
+Distributions

Readme.md

@@ -3,11 +3,15 @@ BIGUQ (Bayesian Information Gap Decision Theory for Uncertainty Quantification,
 
 [![BIGUQ](http://pkg.julialang.org/badges/BIGUQ_0.4.svg)](http://pkg.julialang.org/?pkg=BIGUQ&ver=0.4)
 
+[![BIGUQ](http://pkg.julialang.org/badges/BIGUQ_0.5.svg)](http://pkg.julialang.org/?pkg=BIGUQ&ver=0.5)
+
 [![Build Status](https://travis-ci.org/madsjulia/BIGUQ.jl.svg?branch=master)](https://travis-ci.org/madsjulia/BIGUQ.jl)
 
 [![Coverage Status](https://coveralls.io/repos/madsjulia/BIGUQ.jl/badge.svg?branch=master)](https://coveralls.io/r/madsjulia/BIGUQ.jl?branch=master)
 
-BIGUQ is a module of MADS. MADS (Model Analysis & Decision Support) is an open-source [Julia](http://julialang.org) code designed as an integrated high-performance computational framework performing a wide range of model-based analyses:
+BIGUQ is a module of MADS.
+
+MADS (Model Analysis & Decision Support) is an open-source [Julia](http://julialang.org) code designed as an integrated high-performance computational framework performing a wide range of model-based analyses:
 
 * Sensitivity Analysis
 * Parameter Estimation
@@ -30,7 +34,7 @@ Installation
 After starting Julia, execute:
 
 ```
-Pkg.add("Anasol")
+Pkg.add("BIGUQ")
 ```
 
 Installation behind a firewall

src/BIGDT.jl

@@ -1,5 +1,5 @@
 # @everywhere using ArrayViews
-
+"BigOED type"
 type BigDT
 	makeloglikelihood::Function # we give it a set of likelihood parameters, and it gives us a conditional likelihood function. That is, it gives us a function of the parameters that returns the likelihood of the data given the parameters
 	logprior::Function # the function encoding our prior beliefs
@@ -30,9 +30,9 @@ function getmcmcchain(bigdt::BigDT, likelihoodparams; steps=10 ^ 2, burnin=10, n
 			return l1 + conditionalloglikelihood(params)
 		end
 	end
-	burninchain, burninllhoodvals = Mads.emcee(loglikelihood, numwalkers, broadcast(+, bigdt.nominalparams, 1e-6 * randn(length(bigdt.nominalparams), numwalkers)), burnin, 1)
-	chain, llhoodvals = Mads.emcee(loglikelihood, numwalkers, broadcast(+, bigdt.nominalparams, 1e-6 * randn(length(bigdt.nominalparams), numwalkers)), steps, thinning)
-	return Mads.flattenmcmcarray(chain, llhoodvals)
+	burninchain, burninllhoodvals = AffineInvariantMCMC.sample(loglikelihood, numwalkers, broadcast(+, bigdt.nominalparams, 1e-6 * randn(length(bigdt.nominalparams), numwalkers)), burnin, 1)
+	chain, llhoodvals = AffineInvariantMCMC.sample(loglikelihood, numwalkers, broadcast(+, bigdt.nominalparams, 1e-6 * randn(length(bigdt.nominalparams), numwalkers)), steps, 1)
+	return AffineInvariantMCMC.flattenmcmcarray(chain, llhoodvals)
 end
 
 function get_min_index_of_horizon_with_failure(bigdt::BigDT, sample::Vector, horizons::Vector) # called in getfailureprobabilities

src/BIGOED.jl

@@ -135,10 +135,6 @@ function dobigoed(bigoed::BigOED, hakunamatata::Real, numlikelihoods::Int, numho
 	return dobigoed(bigoed, hakunamatata, numlikelihoods, numhorizons, numobsrealizations, acceptableprobabilityoffailure, makegetfailureprobabilities_mc(modelparams))
 end
 
-function dobigoed(bigoed::BigOED, hakunamatata::Real, numlikelihoods::Int, numhorizons::Int, numobsrealizations::Int, acceptableprobabilityoffailure::Real)
-	return dobigoed(bigoed, hakunamatata, numlikelihoods, numhorizons, numobsrealizations, acceptableprobabilityoffailure, getfailureprobabilities)
-end
-
 function makedecisionforproposedobs(proposedobsarray, i, bigoed, numhorizons, getfailureprobfnct)
 	local bigdts = makebigdts(bigoed, i, proposedobsarray)
 	local maxfailureprobsarray = Array(Array{Float64, 1}, length(bigdts))

src/BIGUQ.jl

@@ -2,10 +2,10 @@
 
 """
 MADS: Model Analysis & Decision Support in Julia (Mads.jl v1.0) 2016
+module BIGUQ
 
-http://mads.lanl.gov
-http://madsjulia.lanl.gov
-http://gitlab.com/mads/Mads.jl
+https://mads.lanl.gov
+https://github.com/madsjulia
 
 Licensing: GPLv3: http://www.gnu.org/licenses/gpl-3.0.html
 
@@ -37,9 +37,9 @@ import ForwardDiff
 import BlackBoxOptim
 import Distributions
 import RobustPmap
-import Mads
+import AffineInvariantMCMC
 
 include("BIGDT.jl")
 include("BIGOED.jl")
 
-end
+end