Welcome to the EPIC user’s guide!¶
Easy Parameter Inference in Cosmology (EPIC) is my implementation in Python of a MCMC code for Bayesian inference of parameters of cosmological models and model comparison via the computation of Bayesian evidences.
Details¶
Author: | Rafael J. F. Marcondes |
---|---|
Contact: | rafaelmarcondes@alumni.usp.br |
Repository: | https://bitbucket.org/rmarcondes/epic |
License: | BSD License 2.0 with an added clause that if you use it in your work
you must cite this user’s guide published in the arXiv repository
as: Marcondes R. J. F., “EPIC - Easy Parameter inference in
Cosmology: The user’s guide to the MCMC sampler”. arXiv:1712.00263
[astro-ph.IM]. See the LICENSE.txt file in the root directory of
the source code for more details. |
Contents¶
About the author¶
I’m a Brazilian engineer and physicist graduated from the Federal University of Sao Carlos (Engineering Physics, 2009), the University of Campinas (M.Sc. in Physics, 2012) and the University of Sao Paulo (Ph.D. in Physics, 2016). Besides developing EPIC, I have worked mainly with tests of interacting dark energy models using growth of structure and galaxy clusters data.
See the inSPIRE-HEP website to access my publications.
Changelog¶
Version 1.4.2¶
New in this version (December 2019)
- Multiprocessing performance improvement for KDE in
analyze
. - GUI now also shows the dark energy equation of state for the selected model.
- “Load” button becomes a menu button and now also gives an option to start a
new simulation from a previously configured
.ini
file. - Fix a bug with the command
monitor
. - Evolution of convergence is plotted after a MCMC simulation is finished.
- Other small changes and bug fixes.
Version 1.4¶
New in this version (September 2018):
- New EoS parametrization Jassal-Bagla-Padmanabhan (JBP).
- New datasets included: updated CMB distance priors from Planck 2018.
- New command
burst
to generate a triangle plot of the chains separately. - You can now resume simulations from the GUI using the “Load” button.
- New button to empty list of solved models for distance comparison in the GUI.
Changed in this version:
- Renamed fast-varying EoS models to Linder-Huterer (LH) and Felice-Nesseris-Tsujikawa (FNT) models.
- Simulation output now displays total number of steps when resuming simulations (rather than the number of steps since last run).
Version 1.3.2¶
New in this version (September 2018):
- Fix logic regarding when background solutions should be recalculated in models that need numerical solution.
- New option
--save-rejected
to also write rejected states to disk. - New option
--use-chain
to plot results of specific chains. - Option
--group-name
is now--plot-prefix
and is always applied withplot
command, even when only one simulation is plotted. - Other bug fixes.
Version 1.3¶
New in this version (July 2018):
- You can now run MCMC from the graphical interface!
- EPIC creates a home folder for the
*.ini
configuration files, user modifications (new models, species, datasets, etc) at the users’s home directory or other location set through the environment variable$EPIC_USER_PATH
.
Changed in this version:
- Simplified installation and setup process. When installing from PyPi, only
the
epic.py
script and the relevant.ini
files are exposed to the user in EPIC’s home folder. Depending on your system configuration you will be able to runepic.py
from any directory. - Other minor improvements.
Version 1.2.1¶
Changed in this version:
- Use gif images for button labels in GUI when TkVersion is inferior to 8.6.
Version 1.2¶
New in this version (July 2018):
- All the chains start at the same point now by default (use
--multi-start
to make it behave as previously). - Option to calculate (when plotting results with the command
plot
) how many sigmas of detection for a given parameter when--kde
is given (need toanalyze
with--kde
first. If convergence is obtained within the required tolerance then--kde
is included automatically). - EPIC now brings a graphical user interface (GUI) for its cosmological
calculator! Try it with the
gui
command. Check the new section in this user guide for instructions. - Bug fixes and other minor improvements.
Version 1.1¶
New in this version (May 2018):
- A new module for Cosmology calculations written from scratch, following an intensively object-oriented approach. This update facilitates the use of this code for Cosmology calculations separated from a MCMC simulation. Try it interactively with jupyter notebook!
- Observational data sets and likelihood calculations have also been reworked and improved. Choice of observations and models and new model set up process should be more transparent now.
- Parallel Tempering algorithm has been removed in this version and may return as a new implementation in a future release.
Version 1.0.4¶
- Uses astropy.io.fits instead of pyfits for loading JLA (v4) files.
Version 1.0.2¶
New in this version:
- Included instructions in documentation on how to show two or more results in the same triangle plot.
- Added section “About the author” to documentation.
- This changelog.
- New background in html documentation favicon, uses readthedocs’ color.
- Included arXiv eprint number of the PDF version of this documentation in license information.
- Slightly reduced mathjax fontsize in html documentation.
- Other minor changes to documentation.
Version 1.0.1¶
- First release on PyPi.