Merge pull request #290 from elfi-dev/dev

Release v. 0.7.3

.dockerignore

@@ -104,3 +104,6 @@ ENV
 *.swo
 *.swp
 
+# Compiled stuff
+bdm
+elfi/examples/cpp/bdm

.travis.yml

@@ -11,11 +11,11 @@ matrix:
     before_install:
     - mkdir -p /Users/travis/.matplotlib
     - "echo 'backend: TkAgg' > /Users/travis/.matplotlib/matplotlibrc"
-    - brew update
-    - brew upgrade python
-    - pip3 install virtualenv
-    - virtualenv py3env -p python3
-    - source py3env/bin/activate
+    - travis_retry wget "http://repo.continuum.io/miniconda/Miniconda3-latest-MacOSX-x86_64.sh" -O miniconda.sh
+    - "bash miniconda.sh -b -p $HOME/miniconda"
+    - "export PATH=$HOME/miniconda/bin:$PATH"
+    - "conda create -y -n elfi python=3.6 scipy"
+    - "source activate elfi"
 
 cache: pip
 

CHANGELOG.rst

@@ -1,6 +1,16 @@
 Changelog
 =========
 
+0.7.3 (2018-08-30)
+------------------
+- Fix bug in plot_pairs which crashes in case of 1 parameter
+- Fix bug in plot_marginals which outputs empty plots in case where we have parameter more than 5
+- Fix crashing summary and plots for samples with multivariate priors
+- Add progress bar for inference methods
+- Add method save to Sample objects
+- Add support for giving seed to `generate`
+- Implement elfi.plot_params_vs_node for plotting parameters vs. node output
+
 0.7.2 (2018-06-20)
 ------------------
 - Added support for kwargs in elfi.set_client

Dockerfile

@@ -1,16 +1,10 @@
-FROM python:3.6
+FROM ubuntu:bionic
 
-RUN apt-get update && apt-get install -y graphviz
+RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y graphviz make python3-scipy jupyter python3-matplotlib
+RUN pip3 install graphviz
 
 WORKDIR /elfi
 ADD . /elfi
 
-RUN pip install numpy graphviz
-RUN pip install -e .
-RUN pip install -r requirements-dev.txt
-
-# matplotlib unable to show figures
-RUN mkdir -p /root/.config/matplotlib
-RUN echo "backend : Agg" >> /root/.config/matplotlib/matplotlibrc
-
-CMD ["/bin/bash"]
+RUN pip3 install -e .
+RUN pip3 install -r requirements-dev.txt

README.md

@@ -1,4 +1,4 @@
-**Version 0.7.2 released!** See the CHANGELOG and [notebooks](https://github.com/elfi-dev/notebooks).
+**Version 0.7.3 released!** See the CHANGELOG and [notebooks](https://github.com/elfi-dev/notebooks).
 
 **NOTE:** For the time being NetworkX 2 is incompatible with ELFI.
 
@@ -13,10 +13,10 @@ ELFI - Engine for Likelihood-Free Inference
 
 <img src="https://cloud.githubusercontent.com/assets/1233418/20178983/6e22ee44-a75c-11e6-8345-5934b55b9dc6.png" width="15%" align="right"></img>
 
-ELFI is a statistical software package written in Python for likelihood-free inference (LFI) such as Approximate 
-Bayesian Computation ([ABC](https://en.wikipedia.org/wiki/Approximate_Bayesian_computation)). 
-The term LFI refers to a family of inference methods that replace the use of the likelihood function with a data 
-generating simulator function. ELFI features an easy to use generative modeling syntax and supports parallelized 
+ELFI is a statistical software package written in Python for likelihood-free inference (LFI) such as Approximate
+Bayesian Computation ([ABC](https://en.wikipedia.org/wiki/Approximate_Bayesian_computation)).
+The term LFI refers to a family of inference methods that replace the use of the likelihood function with a data
+generating simulator function. ELFI features an easy to use generative modeling syntax and supports parallelized
 inference out of the box.
 
 Currently implemented LFI methods:
@@ -32,8 +32,8 @@ ELFI also integrates tools for visualization, model comparison, diagnostics and
 
 See examples under [notebooks](https://github.com/elfi-dev/notebooks) to get started. Full
 documentation can be found at http://elfi.readthedocs.io/. Limited user-support may be
-asked from elfi-support.at.hiit.fi, but the 
-[Gitter chat](https://gitter.im/elfi-dev/elfi?utm_source=share-link&utm_medium=link&utm_campaign=share-link) 
+asked from elfi-support.at.hiit.fi, but the
+[Gitter chat](https://gitter.im/elfi-dev/elfi?utm_source=share-link&utm_medium=link&utm_campaign=share-link)
 is preferable.
 
 
@@ -48,9 +48,9 @@ pip install elfi
 or on some platforms using Python 3 specific syntax:
 ```
 pip3 install elfi
-``` 
+```
 
-Note that in some environments you may need to first install `numpy` with 
+Note that in some environments you may need to first install `numpy` with
 `pip install numpy`. This is due to our dependency to `GPy` that uses `numpy` in its installation.
 
 ### Optional dependencies
@@ -77,7 +77,7 @@ pip install elfi
 
 ### Docker container
 
-A simple Dockerfile for command-line interface is also provided. Please see [Docker documentation](https://docs.docker.com/).
+A simple Dockerfile for command-line interface is also provided. This is especially suitable for running tests. Please see [Docker documentation](https://docs.docker.com/) for details.
 
 ```
 git clone --depth 1 https://github.com/elfi-dev/elfi.git
@@ -88,11 +88,11 @@ docker run -it elfi
 
 ### Potential problems with installation
 
-ELFI depends on several other Python packages, which have their own dependencies. 
+ELFI depends on several other Python packages, which have their own dependencies.
 Resolving these may sometimes go wrong:
 - If you receive an error about missing `numpy`, please install it first.
 - If you receive an error about `yaml.load`, install `pyyaml`.
-- On OS X with Anaconda virtual environment say `conda install python.app` and then use 
+- On OS X with Anaconda virtual environment say `conda install python.app` and then use
 `pythonw` instead of `python`.
 - Note that ELFI requires Python 3.5 or greater so try `pip3 install elfi`.
 - Make sure your Python installation meets the versions listed in `requirements.txt`.

docs/api.rst

@@ -38,6 +38,7 @@ Below is the API for creating generative models.
 
 .. autosummary::
    elfi.draw
+   elfi.plot_params_vs_node
 
 Inference API
 -------------
@@ -177,6 +178,8 @@ Modelling API classes
 
 .. autofunction:: elfi.draw
 
+.. autofunction:: elfi.plot_params_vs_node
+
 .. This would show undocumented members :undoc-members:
 
 

docs/installation.rst

@@ -92,7 +92,8 @@ means the current folder.
 Docker container
 ----------------
 
-A simple Dockerfile for command-line interface is also provided. Please see `Docker documentation`_.
+A simple Dockerfile for command-line interface is also provided. This is especially suitable 
+for running tests. Please see `Docker documentation`_ for details.
 
 .. _Docker documentation: https://docs.docker.com/
 

docs/quickstart.rst

@@ -28,7 +28,7 @@ for the standard deviation ``sigma``. All distributions from
 For likelihood-free models we typically need to define a simulator and
 summary statistics for the data. As an example, lets define the
 simulator as 30 draws from a Gaussian distribution with a given mean and
-standard deviation. Let's use mean and variance as our summaries:
+standard deviation. Let’s use mean and variance as our summaries:
 
 .. code:: ipython3
 
@@ -125,7 +125,7 @@ posterior using threshold value 0.5:
     
 
 
-Let's plot also the marginal distributions for the parameters:
+Let’s plot also the marginal distributions for the parameters:
 
 .. code:: ipython3
 

docs/usage/BOLFI.rst

@@ -101,7 +101,7 @@ initialize the GP model with a dictionary of previous results by giving
 
 The BOLFI class can now try to ``fit`` the surrogate model (the GP) to
 the relationship between parameter values and the resulting
-discrepancies. We'll request only 100 evidence points (including the
+discrepancies. We’ll request only 100 evidence points (including the
 ``initial_evidence`` defined above).
 
 .. code:: ipython3
@@ -263,7 +263,7 @@ is a sophisticated algorithm, and in some cases one may get warnings
 about diverged proposals, which are signs that `something may be wrong
 and should be
 investigated <http://mc-stan.org/misc/warnings.html#divergent-transitions-after-warmup>`__.
-It is good to understand the cause of these warnings although they don't
+It is good to understand the cause of these warnings although they don’t
 automatically mean that the results are unreliable. You could try
 rerunning the ``sample`` method with a higher target probability
 ``target_prob`` during adaptation, as its default 0.6 may be inadequate

docs/usage/external.rst

@@ -10,11 +10,11 @@ If your simulator or other operations are implemented in a programming
 language other than Python, you can still use ELFI. This notebook
 briefly demonstrates how to do this in three common scenarios:
 
--  External executable (written e.g. in C++ or a shell script)
+-  External executable (written e.g. in C++ or a shell script)
 -  R function
 -  MATLAB function
 
-Let's begin by importing some libraries that we will be using:
+Let’s begin by importing some libraries that we will be using:
 
 .. code:: ipython3
 
@@ -43,7 +43,7 @@ introduced in this tutorial.
 
 We demonstrate here how to wrap executables as ELFI nodes. We will first
 use ``elfi.tools.external_operation`` tool to wrap executables as a
-Python callables (function). Let's first investigate how it works with a
+Python callables (function). Let’s first investigate how it works with a
 simple shell ``echo`` command:
 
 .. code:: ipython3
@@ -64,7 +64,7 @@ simple shell ``echo`` command:
 
 
 
-The placeholders for arguments in the command string are just Python's
+The placeholders for arguments in the command string are just Python’s
 ```format strings`` <https://docs.python.org/3/library/string.html#formatstrings>`__.
 
 Currently ``echo_sim`` only accepts scalar arguments. In order to work
@@ -103,16 +103,16 @@ Complex external operations :math:`-` case BDM
 ----------------------------------------------
 
 To provide a more realistic example of external operations, we will
-consider the Birth-Death-Mutation (BDM) model used in `*Lintusaari at al
-2016* <https://doi.org/10.1093/sysbio/syw077>`__ *[1]*.
+consider the Birth-Death-Mutation (BDM) model used in `Lintusaari at al
+2016 <https://doi.org/10.1093/sysbio/syw077>`__ *[1]*.
 
 Birth-Death-Mutation process
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 We will consider here the Birth-Death-Mutation process simulator
 introduced in *Tanaka et al 2006 [2]* for the spread of Tuberculosis.
 The simulator outputs a count vector where each of its elements
-represents a "mutation" of the disease and the count describes how many
+represents a “mutation” of the disease and the count describes how many
 are currently infected by that mutation. There are three rates and the
 population size:
 
@@ -128,8 +128,8 @@ population size:
 It is assumed that the susceptible population is infinite, the hosts
 carry only one mutation of the disease and transmit that mutation
 onward. A more accurate description of the model can be found from the
-original paper or e.g. `*Lintusaari at al
-2016* <https://doi.org/10.1093/sysbio/syw077>`__ *[1]*.
+original paper or e.g. `Lintusaari at al
+2016 <https://doi.org/10.1093/sysbio/syw077>`__ *[1]*.
 
 .. For documentation
 
@@ -140,8 +140,8 @@ original paper or e.g. `*Lintusaari at al
 
 This simulator cannot be implemented effectively with vectorized
 operations so we have implemented it with C++ that handles loops
-efficiently. We will now reproduce Figure 6(a) in `*Lintusaari at al
-2016* <https://doi.org/10.1093/sysbio/syw077>`__ *[2]* with ELFI. Let's
+efficiently. We will now reproduce Figure 6(a) in `Lintusaari at al
+2016 <https://doi.org/10.1093/sysbio/syw077>`__ *[2]* with ELFI. Let’s
 start by defining some constants:
 
 .. code:: ipython3
@@ -154,7 +154,7 @@ start by defining some constants:
     # The zeros are to make the observed population vector have length N
     y_obs = np.array([6, 3, 2, 2, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0], dtype='int16')
 
-Let's build the beginning of a new model for the birth rate
+Let’s build the beginning of a new model for the birth rate
 :math:`\alpha` as the only unknown
 
 .. code:: ipython3
@@ -214,8 +214,8 @@ thousands of simulations. We will also here redirect the output to a
 file and then read the file into a numpy array.
 
 This is just one possibility among the many to implement this. The most
-efficient would be to write a native Python module with C++ but it's
-beyond the scope of this article. So let's work through files which is a
+efficient would be to write a native Python module with C++ but it’s
+beyond the scope of this article. So let’s work through files which is a
 fairly common situation especially with existing software.
 
 .. code:: ipython3
@@ -265,7 +265,7 @@ fairly common situation especially with existing software.
                                         process_result=process_result, 
                                         stdout=False)
 
-Now let's replace the echo simulator with this. To create unique but
+Now let’s replace the echo simulator with this. To create unique but
 informative filenames, we ask ELFI to provide the operation some meta
 information. That will be available under the ``meta`` keyword (see the
 ``prepare_inputs`` function above):
@@ -306,9 +306,9 @@ Completing the BDM model
 ~~~~~~~~~~~~~~~~~~~~~~~~
 
 We are now ready to finish up the BDM model. To reproduce Figure 6(a) in
-`*Lintusaari at al 2016* <https://doi.org/10.1093/sysbio/syw077>`__
-*[2]*, let's add different summaries and discrepancies to the model and
-run the inference for each of them:
+`Lintusaari at al 2016 <https://doi.org/10.1093/sysbio/syw077>`__ *[2]*,
+let’s add different summaries and discrepancies to the model and run the
+inference for each of them:
 
 .. code:: ipython3
 
@@ -421,7 +421,7 @@ the functionality of R from within Python. You can install it with
 ``pip install rpy2``.
 
 Here we demonstrate how to calculate the summary statistics used in the
-ELFI tutorial (autocovariances) using R's ``acf`` function for the MA2
+ELFI tutorial (autocovariances) using R’s ``acf`` function for the MA2
 model.
 
 .. code:: ipython3
@@ -436,7 +436,7 @@ model.
 
 .. _issue: https://github.com/ContinuumIO/anaconda-issues/issues/152
 
-Let's create a Python function that wraps the R commands (please see the
+Let’s create a Python function that wraps the R commands (please see the
 documentation of `rpy2 <http://rpy2.readthedocs.io>`__ for details):
 
 .. code:: ipython3
@@ -600,7 +600,7 @@ Replace the summaries S1 and S2 with our R autocovariance function.
 
 
 
-Finally, don't forget to quit the MATLAB session:
+Finally, don’t forget to quit the MATLAB session:
 
 .. code:: ipython3
 
@@ -625,6 +625,6 @@ References
    Kaski, Jukka Corander; Fundamentals and Recent Developments in
    Approximate Bayesian Computation. Syst Biol 2017; 66 (1): e66-e82.
    doi: 10.1093/sysbio/syw077
--  [2] Tanaka, Mark M., et al. "Using approximate Bayesian computation
-   to estimate tuberculosis transmission parameters from genotype data."
+-  [2] Tanaka, Mark M., et al. “Using approximate Bayesian computation
+   to estimate tuberculosis transmission parameters from genotype data.”
    Genetics 173.3 (2006): 1511-1520.