Author | Changes | Lines of Code | Lines per Change |
---|---|---|---|
Totals | 60 (100.0%) | 1838 (100.0%) | 30.6 |
msuchard | 15 (25.0%) | 967 (52.6%) | 64.4 |
rambaut | 21 (35.0%) | 548 (29.8%) | 26.0 |
alexei.drummond | 13 (21.7%) | 273 (14.9%) | 21.0 |
jheled | 8 (13.3%) | 46 (2.5%) | 5.7 |
bloomquist | 2 (3.3%) | 3 (0.2%) | 1.5 |
gerton.lunter | 1 (1.7%) | 1 (0.1%) | 1.0 |
As I threatened, I have now deleted dr.app.beastdev (I had one yes and no complaints).
0 lines of code changed in 4 files:
A bit more cleaning up. AbstractCoalescentLikelihood is now called OldAbstractCoalescentLikelihood to indicate to authors of descendent classes that they may wish to try and use the new "AbstractCoalescentLikelihood" which is cleaner and more modular. It also allows for <include> and <exclude> of taxon lists to do subtrees. Basically, all subclasses have to do is implement calculateLogLikelihood() and call getIntervals() and getTree() to get the current intervals and tree objects. See "CoalescentLikelihood" for an example (also the parser for including and excluding taxon lists - otherwise provide nulls).
12 lines of code changed in 2 files:
I have renamed 'NewCoalescentLikelihood' as 'CoalescentLikelihood' to clarify things a bit. NewCoalescentLikelihood was more efficient and clean implementation and it's parser was installed as the default 'coalescentLikelihood' anyway since BEAST v1.4.x.
I have renamed 'CoalescentLikelihood' to AbstractCoalescentLikelihood and removed the parser (made it abstract). This class is used as a base by a number of other classes (i.e., BayesianSkylineLikelihood). I guess that ideally, we see what of AbstractCoalescentLikelihood the descendents actually use and strip out some of it.
1 lines of code changed in 1 file:
Changed remaining skyline's to skyride's to match up with Minin et al. resubmission.
2 lines of code changed in 1 file:
Renamed GMRFSkylineLikelihood to GMRFSkyrideLikelihood to match up with Minin et al. resubmission.
1 lines of code changed in 1 file:
An error in the BeastParser - some parsers were included twice
0 lines of code changed in 1 file:
Updated dr.app.beast.BeastParser to contain all the parsers in dr.app.beastdev.*. Will delete dr.app.beastdev.* soon. Moved BeastMC3 across. Fixed a bug setting BSP group count in Beauti.
219 lines of code changed in 1 file:
add nee 1994 birth/death model. fix likelihood calculation by adding the ignored "tree probability".
3 lines of code changed in 1 file:
Fixed issue 39
5 lines of code changed in 2 files:
variable demographic stuff
15 lines of code changed in 1 file:
Various improvements to BEAUti and the odd bit of cleaning up. Some additions to tracer
239 lines of code changed in 1 file:
Added generalized GTR model with separate root and stationary distributions for irreversible processes. Added a Dirchlet prior distribution for informative root or stationary distributions. NodeTraitLogger just annotates trees with all traits.
4 lines of code changed in 1 file:
Started an InhibitionAssayLikelihood class for combining inhibition assay data with phylogenetic reconstruction
6 lines of code changed in 1 file:
Moved some evolution specific operators from dr.inference.operators to dr.evomodel.operators (all Marc's I think).
5 lines of code changed in 1 file:
Started a MultivariateGammaDistribution for specifying pairwise distance-based rates for the discrete CTMC diffusion models.
1 lines of code changed in 1 file:
Started to add in classes for a multivate OU diffusion process and using Beast for doing (generalized) linear regression problems. May Beast become the new WinBugs!
11 lines of code changed in 1 file:
5 lines of code changed in 1 file:
Add a version of the general time-reversible CTMC that employs Bayesian stochastic variable selection to choose the appropriate number of free parameters. There are still some issues with priors/normalization to be worked out, but otherwise -- LET'S WRITE A PAPER!
BitFlipInSubstitutionModelOperator is only necessary if one wants to mandate an irreducible chain, otherwise BitFlipOperator works fine and almost surely returns irreducible chains.
4 lines of code changed in 1 file:
Moved Marc's stuff into BEAST parser
0 lines of code changed in 1 file:
Map stuff
215 lines of code changed in 1 file:
Started Gibbs operator for fixed effects in the GMRF
1 lines of code changed in 1 file:
still working on genrea dna covarion model -- the parameters seem to be in the wrong cells of the matrix...
3 lines of code changed in 1 file:
implemented general DNA covarion models - untested!
1 lines of code changed in 1 file:
207 lines of code changed in 1 file:
Extended TreeLikelihood to AncestralStateTreeLikelihood. This class provides an ancestral reconstruction of the internal node states when the tree is logged. The class implements NodeAttribute(something) and performs a pre-traversal order simulation of the internal node states. To save on computations, I added a few public accessors to AbstractLikelihoodCore; these should probably be moved elsewhere, as they may conflict with the native likelihoodCores. There's a simple example in testAncestralReconstruction.xml. TreeAnnotator will need some patching to return the modal internal node states.
I apologize if ancestral reconstruction is already handled elsewhere, but in my defense, I could not find it!
BEAST users will probably enjoy doing lots of reconstruction, so let's write some sort of quick paper to inform users.
208 lines of code changed in 1 file:
Removed dependence on colt by using sparse matrix algorithms in mtj, results in several order of magnitude speed increase.
9 lines of code changed in 1 file:
Cleaned up all the expression statistics (mainly porting from 1.4 branch)
10 lines of code changed in 1 file:
tidying
244 lines of code changed in 1 file:
Starting to implement Rue's block updator for the GMRF classes
204 lines of code changed in 1 file:
tidying work towards improving log outputs
0 lines of code changed in 1 file:
Fixed problem logging trees MCMCMC (was giving null taxon labels).
1 lines of code changed in 1 file:
FIrst-pass hack at implementing a weighted random walk on a d-2 Markov lattice. Allows for discretized diffusion on a terrain (topographic) map for use with the "coyote" (errr cougar) data. The MapMaker class currently requires mtj.jar (optimized sparse matrix operations) to construct the eigensystem of the weighted lattice.
2 lines of code changed in 1 file:
Finished a basic implementation of the GMRF coalescent smoother. This implementation currently uses a random walk operator on the population sizes and a Gibbs update of the precision parameter. A (much) better/faster approach (to come shortly) is a joint update using Rue's GMRF library.
6 lines of code changed in 1 file:
Split TreeModel XML parser into separate class because TreeModel is a monolithic class. Fixed a labeling problem introduced into CoalescentLikelihood. However the CoalescentLikelihood now returns weird probabilities after jheled's last commit -- so this still needs looking into.
2 lines of code changed in 1 file:
speed improvements to relaxed clocks (~20% faster to uncorrelated models) - important changes for porting are DiscretizedBranchRates and TreeLikelihood
1 lines of code changed in 1 file:
84 lines of code changed in 1 file:
tidyup and generifying
1 lines of code changed in 1 file:
ongoing work on variable dimentsion skyline
6 lines of code changed in 1 file:
Fixed 2 bugs in patristic distances code
3 lines of code changed in 1 file:
The rest of the previous commit -- parser for the bayesian skyline gibbs sampler,
and additions to the gamma distribution.
1 lines of code changed in 1 file:
(12 more)