Skip to content
The inference code for our Nature paper about globular clusters in M31/Andromeda.
C++ Python Makefile
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
AndromedaMixture
DNest4 @ 9a8c4a1
Data
Documents
.gitignore
.gitmodules
LICENSE
Makefile
OPTIONS
README.md
extract_best_fit.py
main.cpp
rotation_curves.py
showresults.py

README.md

Mixture Model for Andromeda Globular Clusters

(c) 2019 Brendon J. Brewer and Geraint F. Lewis.

This is free software released under the terms of the MIT license. It was used to do the model comparison and parameter estimation calculations for the following paper (Nature, 2019, accepted):

Two major accretion epochs in M31 from two distinct populations of globular clusters

Dougal Mackey, Geraint F. Lewis, Brendon J. Brewer, Annette M. N. Ferguson, Jovan Veljanoski, Avon P. Huxor, Michelle L. M. Collins, Patrick Côté, Rodrigo A. Ibata, Mike J. Irwin, Nicolas Martin, Alan W. McConnachie, Jorge Peñarrubia, Nial Tanvir, Zhen Wan

Instructions to Reproduce Results in the Paper

You will require all the dependencies of DNest4, plus corner.py. First, clone this repo recursively and compile the C++:

git clone --recursive https://github.com/eggplantbren/AndromedaMixture
cd AndromedaMixture
make

Install the DNest4 Python package (unless you already have it), like this (Mac users, see the DNest4 manual for something extra you need to do):

cd DNest4/python
python setup.py install

Now go back to the directory containing the binary, and run DNest4 with ten threads and a random number seed of 1 (this will run model V2, the favoured model):

cd ../..
./main -t 10 -s 1

Once that's finished (it produces 100,000 saved particles), run the postprocessing script:

python showresults.py

This will produce three DNest4 plots and output the marginal likelihood to the screen and posterior samples to posterior_sample.txt (one parameter per column, see the header line for the meaning of each column). After closing those plots, it will produce the corner plot and save it to corner.png.

Changing the rotation model

The default rotation model is V2 from the paper. Modify lines 13 and 14 of main.cpp and re-compile by running make to use a different rotation model. For example, to use model S1, these lines should be modified to

constexpr RotationModel rotation_model = RotationModel::S;
constexpr int num_components = 1;

To exactly reproduce the results from the paper, don't forget to use the correct random number seed from Table 3.

You can’t perform that action at this time.