Merge branch 'master' into tor/dppl-update

Author: torfjelde

Project.toml

@@ -1,6 +1,6 @@
 name = "Turing"
 uuid = "fce5fe82-541a-59a6-adf8-730c64b5f9a0"
-version = "0.16.2"
+version = "0.17.0"
 
 [deps]
 AbstractMCMC = "80f14c24-f653-4e6a-9b94-39d6b0f70001"
@@ -34,20 +34,20 @@ ZygoteRules = "700de1a5-db45-46bc-99cf-38207098b444"
 
 [compat]
 AbstractMCMC = "3.2"
-AdvancedHMC = "0.2.24"
+AdvancedHMC = "0.3.0"
 AdvancedMH = "0.6"
-AdvancedPS = "0.2"
+AdvancedPS = "0.2.4"
 AdvancedVI = "0.1"
 BangBang = "0.3"
 Bijectors = "0.8, 0.9"
 DataStructures = "0.18"
 Distributions = "0.23.3, 0.24, 0.25"
 DistributionsAD = "0.6"
 DocStringExtensions = "0.8"
-DynamicPPL = "0.12.1"
+DynamicPPL = "0.12.1, 0.13, 0.14"
 EllipticalSliceSampling = "0.4"
 ForwardDiff = "0.10.3"
-Libtask = "= 0.4.0, = 0.4.1, = 0.4.2, = 0.5.0, = 0.5.1"
+Libtask = "0.4, 0.5.3"
 MCMCChains = "4"
 NamedArrays = "0.9"
 Reexport = "0.2, 1"

README.md

@@ -4,6 +4,7 @@
 [![Coverage Status](https://coveralls.io/repos/github/TuringLang/Turing.jl/badge.svg?branch=master)](https://coveralls.io/github/TuringLang/Turing.jl?branch=master)
 [![codecov](https://codecov.io/gh/TuringLang/Turing.jl/branch/master/graph/badge.svg?token=OiUBsnDQqf)](https://codecov.io/gh/TuringLang/Turing.jl)
 [![Documentation](https://img.shields.io/badge/doc-latest-blue.svg)](https://turing.ml/dev/docs/using-turing/)
+[![ColPrac: Contributor's Guide on Collaborative Practices for Community Packages](https://img.shields.io/badge/ColPrac-Contributor's%20Guide-blueviolet)](https://github.com/SciML/ColPrac)
 
 **Turing.jl** is a Julia library for general-purpose [probabilistic programming](https://en.wikipedia.org/wiki/Probabilistic_programming_language). Turing allows the user to write models using standard Julia syntax, and provides a wide range of sampling-based inference methods for solving problems across probabilistic machine learning, Bayesian statistics, and data science. Compared to other probabilistic programming languages, Turing has a special focus on modularity, and decouples the modelling language (i.e. the compiler) and inference methods. This modular design, together with the use of a high-level numerical language Julia, makes Turing particularly extensible: new model families and inference methods can be easily added.
 

docs/pages/index.html

@@ -35,7 +35,7 @@
 
 flux:
   image_path: /tutorials/figures/3_BayesNN_12_1.svg
-  excerpt: "Turing supports Julia's [Flux](http://fluxml.ai/) package for automatic differentiation. Combine Turing and Flux to construct probabalistic variants of traditional machine learning models."
+  excerpt: "Turing supports Julia's [Flux](http://fluxml.ai/) package for automatic differentiation. Combine Turing and Flux to construct probabilistic variants of traditional machine learning models."
   url: "/tutorials/3-bayesnn"
 
 ecosystem:

docs/src/tutorials/index.md

@@ -7,7 +7,7 @@ title: Tutorials
 This section contains tutorials on how to implement common models in Turing. If you prefer to have an interactive Jupyter notebook, please fork or download the [TuringTutorials](https://github.com/TuringLang/TuringTutorials) repository.
 
 
-A list of all the tutorials available can be found to the left. The [introduction]({{site.baseurl}}/tutorials/0-introduction) tutorial contains an introduction to coin flipping with Turing and a brief overview of probabalistic programming.
+A list of all the tutorials available can be found to the left. The [introduction]({{site.baseurl}}/tutorials/0-introduction) tutorial contains an introduction to coin flipping with Turing and a brief overview of probabilistic programming.
 
 
 Tutorials are under continuous development, but there are some older version available at the TuringTutorials within the [`old-notebooks`](https://github.com/TuringLang/TuringTutorials/tree/master/old-notebooks/) section. Some of these were built using prior versions of Turing and may not function correctly, but they can assist in the syntax used for common models.

docs/src/using-turing/guide.md

@@ -45,7 +45,7 @@ end
 Note: As a sanity check, the expectation of `s` is 49/24 (2.04166666...) and the expectation of `m` is 7/6 (1.16666666...).
 
 
-We can perform inference by using the `sample` function, the first argument of which is our probabalistic program and the second of which is a sampler. More information on each sampler is located in the [API]({{site.baseurl}}/docs/library).
+We can perform inference by using the `sample` function, the first argument of which is our probabilistic program and the second of which is a sampler. More information on each sampler is located in the [API]({{site.baseurl}}/docs/library).
 
 
 ```julia

docs/src/using-turing/performancetips.md

@@ -85,7 +85,7 @@ Alternatively, you could use `filldist` in this example:
 
 ```julia
 @model function tmodel(x, y)
-    params = filldist(truncated(Normal(), 0, Inf), size(x, 2))
+    params ~ filldist(truncated(Normal(), 0, Inf), size(x, 2))
     a = x * params
     y ~ MvNormal(a, 1.0)
 end

docs/src/using-turing/quick-start.md

@@ -1,16 +1,16 @@
 ---
-title: Probablistic Programming in Thirty Seconds
+title: Probabilistic Programming in Thirty Seconds
 ---
 
-# Probablistic Programming in Thirty Seconds
+# Probabilistic Programming in Thirty Seconds
 
 If you are already well-versed in probabilistic programming and just want to take a quick look at how Turing's syntax works or otherwise just want a model to start with, we have provided a Bayesian coin-flipping model to play with.
 
 
 This example can be run on however you have Julia installed (see [Getting Started]({{site.baseurl}}/docs/using-turing/get-started)), but you will need to install the packages `Turing` and `StatsPlots` if you have not done so already.
 
 
-This is an excerpt from a more formal example introducing probabilistic programming which can be found in Jupyter notebook form [here](https://nbviewer.jupyter.org/github/TuringLang/TuringTutorials/blob/master/0_Introduction.ipynb) or as part of the documentation website [here]({{site.baseurl}}/tutorials).
+This is an excerpt from a more formal example introducing probabilistic programming which can be found in Jupyter notebook form [here](https://nbviewer.jupyter.org/github/TuringLang/TuringTutorials/blob/master/notebook/00-introduction/00_introduction.ipynb) or as part of the documentation website [here]({{site.baseurl}}/tutorials).
 
 
 ```julia

src/Turing.jl

@@ -1,13 +1,5 @@
 module Turing
 
-##############
-# Dependency #
-########################################################################
-# NOTE: when using anything from external packages,                    #
-#       let's keep the practice of explictly writing Package.something #
-#       to indicate that's not implemented inside Turing.jl            #
-########################################################################
-
 using Requires, Reexport, ForwardDiff
 using DistributionsAD, Bijectors, StatsFuns, SpecialFunctions
 using Statistics, LinearAlgebra

src/inference/AdvancedSMC.jl

@@ -301,7 +301,7 @@ function AbstractMCMC.step(
     particles = AdvancedPS.ParticleContainer(x)
 
     # Perform a particle sweep.
-    logevidence = AdvancedPS.sweep!(rng, particles, spl.alg.resampler)
+    logevidence = AdvancedPS.sweep!(rng, particles, spl.alg.resampler, reference)
 
     # Pick a particle to be retained.
     Ws = AdvancedPS.getweights(particles)

src/inference/Inference.jl

@@ -325,6 +325,8 @@ function AbstractMCMC.bundle_samples(
     save_state = false,
     stats = missing,
     sort_chain = false,
+    discard_initial = 0,
+    thinning = 1,
     kwargs...
 )
     # Convert transitions to array format.
@@ -363,6 +365,8 @@ function AbstractMCMC.bundle_samples(
         (internals = extra_params,);
         evidence=le,
         info=info,
+        start=discard_initial + 1,
+        thin=thinning,
     )
 
     return sort_chain ? sort(chain) : chain