Added Sidney's improvements in the UNIX bin scripts
100 lines of code changed in 10 files:
More porting of improvements over from 1.4
522 lines of code changed in 10 files:
Bugfix of exp-weighted sampling from gamma-distribution.
This version doesn't get stuck in inefficient rejection sampling.
114 lines of code changed in 1 file:
Added a little document describing why I think equation (3) in Nee is the right one...
109 lines of code changed in 3 files:
Update of the gamma + exponential sampler -- rejection sampling algorithm got too slow
17 lines of code changed in 1 file:
Bayesian skyline Gibbs operator updated to include the exponential Markov prior.
188 lines of code changed in 2 files:
Okay, after a careful read of Sean Nee and some double checking I have found a couple of additional problems with the Yule model implementation. I have fixed these problems. The YuleModel class now implements the likelihood in equation (3) of Nee (2001) "Inferring Speciation Rates from Phylogenies". The previous implementation appeared to follow equation (6) of the same paper -- which is incorrect since the root height is a random variable and the number of tips is conditioned on, rather than vice versa.
135 lines of code changed in 5 files:
The Yule model was incorrect! Marc Suchard was right (thank you Marc, sorry I didn't follow up sooner)... the scaling of lambda by the rootHeight in the denominator was incorrect. *Also*, the root node itself was incorrectly missed out of the calculation (we need to include the root node in the calculation so that every internode interval in the tree contributes to the prior density, otherwise the final interval can have any value it likes and the prior is not proper). I have added a file (testYuleModel.xml) in the examples folder that actually tests the Yule model *by itself*. Who wants to tell the users? It seems like the effect is *usually* small but for data with a lot of taxa and not much sequence variation it could get very very ugly. I suspect this is why a number of our users have been complaining that they get very different results from Yule and coalescent priors on population genetic and phylogeographic data sets... I vote that we patch and ship a BEAST 1.4.3 pronto... AJD
120 lines of code changed in 5 files:
The rest of the previous commit -- parser for the bayesian skyline gibbs sampler,
and additions to the gamma distribution.
500 lines of code changed in 3 files:
Added an almost-Gibbs operator for the generalized Bayesian skyline model.
The operator samples from the posterior distribution of population sizes, conditional on the genealogy,
but not including any external prior. When no prior is provided, the Hastings ratio cancels with the
ratio of likelihoods, and the operator behaves as a Gibbs sampler. Any external prior is taken into
account via the normal Metropolis step.
The operator can be restricted to sample only from a range of population sizes. This is implemented
by rejection sampling. In extreme cases, it may take -very- long to get a decent sample; so when 100
samples have been rejected, a sample at the appropriate extreme end is selected. This is a hack.
The operator can also incorporate (without rejection sampling) a Jeffrey's prior on the geometric
average of population sizes. This however is probably not very useful.
To make all this happen, I've also extended GammaDistribution to sample reliably all valid parameter
values, and some invalid ones (i.e. shape parameter = 0).
427 lines of code changed in 3 files:
Reverted a hack that disabled native library loading.
1 lines of code changed in 1 file:
handle correctly the case where a taxa set is a union of previous sets without any new taxa
29 lines of code changed in 1 file:
remove more unused code
9 lines of code changed in 1 file:
remove unused variables
6 lines of code changed in 1 file:
Ported a few things over from the 1.4.x branch (in BEAUti)
201 lines of code changed in 16 files: