DNest4 is a C++11 implementation of Diffusive Nested Sampling, a Markov Chain Monte Carlo (MCMC) algorithm for Bayesian Inference and Statistical Mechanics. You can use it in a few different ways:
- Implement your model in C++, compile it and have it run super fast.
- Implement trans-dimensional models with the RJObject template class.
- Implement your model by writing just two functions in Python, R, or Julia (this is new and undocumented - email me or take a look in the Templates directory)
- Implement your model as a Python class.
- Write up your model in Python, using a BUGS-style approach see here to learn how (more documentation here).
There is a paper describing DNest4 installation and usage in the Journal of Statistical software. You might also want to read the original paper describing the Diffusive Nested Sampling algorithm itself. If you find this software useful in your research, please cite one or both of these papers. Here are the citations:
Brewer, B., & Foreman-Mackey, D. (2018).
DNest4: Diffusive Nested Sampling in C++ and Python.
Journal of Statistical Software, 86(7), 1 - 33. doi:http://dx.doi.org/10.18637/jss.v086.i07
Brewer, B. J., Pártay, L. B., & Csányi, G. (2011). Diffusive nested sampling.
Statistics and Computing, 21(4), 649-656.
You will need a C++ compiler that supports the C++11 standard, along with Python 3 and the Python packages NumPy, scipy, matplotlib, and Cython.
On some Macs, g++
is an alias for clang
, which is a C compiler. If this
is the case for you, you'll need to edit the first line of the Makefile so
that it uses clang++
, which is a C++ compiler.
You can compile the DNest4 library (libdnest4
) using the Makefile in the
code
directory using:
cd code
make
Along with building the library this will compile all the examples. Then, install the Python package:
python setup.py install
in the root directory of this repository.
However, you can also compile and install the library using SCons. To do this you just need to run:
scons install
By default it will attempt to install the library in /usr/local
(with the library files in
/usr/local/lib
and the headers in /usr/local/include/dnest4
), so the above command
must be run as a user with root access or using sudo
. To install to a different location
you can instead run:
scons install --prefix <install_location>
where <install_location>
is the base path for the install.
To install with GDB enabled during the library's compilation you can add the --debug-mode
flag to the install command.
Currently, the Scons installation does not compile the examples or the Python library. Any additions to this installation process are welcome.
(c) 2015--2018 Brendon J. Brewer and contributors. LICENCE: MIT. See the LICENSE file for details.
This work was supported by a Marsden Fast Start grant from the Royal Society of New Zealand.