HdBCS - High-dimensional Bayesian Covariance Selection |
Adrian Dobra (adobra@isds.duke.edu)
Quanli Wang
(quanli@isds.duke.edu) Liang Zhang (lz9@isds.duke.edu) Mike West (mw@isds.duke.edu) |
This site provides C++ code software implementing a an efficient stochastic search algorithm for for exploring spaces of Gaussian Graphical Models. Feel free to download and explore, and let us have your feedback as we update the material. Any identified bugs will be corrected and updated here.
There are two versions of HBCS:
Serial
version of HdBCS.
Parallel
version of HdBCS.
The archives include source code, a README file and examples. The
serial version can be run on a single processor machine and has only
one step. It works fine for datasets in which the number of variables
does not exceed 250. The parallel version has three separate steps: (1) selecting
the most relevant pairs of variables; (2) generating starting models
and (3) improving these starting models until convergence. The first two
steps require a multi-processor computing environment like CSEM, while the third
step is serial. You need the
following publicly available software installed on your system: MPICH, Blitz++ and Lapack.
Research underlying the software presented here was supported in part
the National Science Foundation under Grant No. DMS-0342172, and by
the National Institutes of Health under Grants No. HL-73042 and
CA-112952. Any opinions, findings, conclusions or recommendations
expressed are those of the authors and do not necessarily reflect the views of the National Science Foundation or the National Institutes of Health.