Dev (#62)

* update dependencies

* remove junk files

* remove junk files

* Reorganizing (Good idea Kusti)

* internals.md

* update README

* ignore .vscode directory

* Revert "ignore .vscode directory"

This reverts commit 5753f58.

* Delete settings.json

* split jointdistribution.jl

* parameterize model by module

* make `runtests` throw errors

* Parameterize Model by module

* debugging

* Update for JuliaStaging/GeneralizedGenerated.jl#28

* insert `$sympy`(SymPy) & `$as`(TransformVariables.jl); (#63)

tests passed

* Package upper bounds

* fix demo

* Being less fancy with logpdf

* Move logpdf(m,x,::typeof(codegen)) to codegen.jl

* Fixing stuff I just broke

* working on markovblanket

* working on Markov Blanket

* markov blanket

* add bodyVariables

* export bodyVariables

* make `toposort` us `arguments` instead of `freeVariables`

* up

* ixnay on the ype piracytay

* Type infer for module param (#66)

* apply changes

* fix #65: might need more changes

* Update dependencies + README

* update README

* update README

.travis.yml

@@ -3,7 +3,7 @@ os:
     - linux
     - osx
 julia:
-    - 1.2
+    - 1.3
     - nightly
 env:
     global:

Project.toml

@@ -40,8 +40,37 @@ TransformVariables = "84d833dd-6860-57f9-a1a7-6da5db126cff"
 Zygote = "e88e6eb3-aa80-5325-afca-941959d7151f"
 
 [compat]
-MonteCarloMeasurements = ">=0.3.5"
-julia = "^1.2"
+AdvancedHMC = "0.2"
+Bijectors = "0.4"
+DiffResults = "0.0.4"
+Distributions = "0.21"
+DynamicHMC = "2.1"
+FillArrays = "0.8"
+ForwardDiff = "0.10"
+GeneralizedGenerated = "0.2"
+Graphs = "0.10"
+IRTools = "0.3"
+IterTools = "1.3"
+LazyArrays = "0.14"
+LogDensityProblems = "0.9"
+MLStyle = "0.3"
+MacroTools = "0.5"
+MonteCarloMeasurements = "0.5"
+NamedTupleTools = "0.12"
+Plots = "0.28"
+PyCall = "1.91"
+Reexport = "0.2"
+ResumableFunctions = "0.5"
+ReverseDiff = "0.3"
+SimpleGraphs = "0.3"
+SimplePartitions = "0.2"
+SimplePosets = "0.0"
+StatsFuns = "0.9"
+Stheno = "0.3"
+SymPy = "1.0"
+TransformVariables = "0.3"
+Zygote = "0.4"
+julia = "^1.3"
 
 [extras]
 BenchmarkTools = "6e4b80f9-dd63-53aa-95a3-0cdb28fa8baf"

README.md

@@ -98,7 +98,7 @@ Often these are easier to work with in terms of `particles` (built using [MonteC
 julia> post = dynamicHMC(m(X=truth.X), (y=truth.y,));
 
 julia> particles(post)
-(β = Particles{Float64,1000}[0.558 ± 0.25, 0.768 ± 0.49],)
+(β = Particles{Float64,1000}[0.548 ± 0.24, 0.751 ± 0.5],)
 
 ````
 
@@ -204,11 +204,11 @@ julia> using BenchmarkTools
 
 julia> 
 @btime logpdf($m2(X=X), $truth)
-  802.533 ns (16 allocations: 464 bytes)
+  1.989 μs (47 allocations: 1.05 KiB)
 -15.84854642585797
 
 julia> @btime logpdf($m2(X=X), $truth, $codegen)
-  324.463 ns (5 allocations: 208 bytes)
+  313.860 ns (5 allocations: 208 bytes)
 -15.848546425857968
 
 ````

examples/2019-11-07-demo.jmd

@@ -13,7 +13,7 @@ weave("examples/2019-11-07-demo.jmd"
 
 
 ```julia
-using Revise, Soss, Plots, NamedTupleTools
+using Soss, Plots, NamedTupleTools
 seed = WEAVE_ARGS[:seed]
 import Random; Random.seed!(seed)
 ```