Removed <oldCoalescentLikelihood> parser and created a testBSP.xml file to check backwards compatibility with BEAST 1.4 (should add <expect> values from a long 1.4 run).
481 lines of code changed in 3 files:
Fix a bug with the scaleAllIndependently option. The Hastings ratio was not computed properly. I can't figure out how bad it was... Please note that I rely on the existing formulla and I am not sure if this one is correct !
236 lines of code changed in 1 file:
Created a likelihood core that uses BigDecimal to act as a comparison for investigations into loss of precision and underflow on large trees. ExperimentalTreeLikelihood is one that does the comparisons.
1039 lines of code changed in 2 files:
Added the ability to force the use of a Java nucleotide core (rather than native) in order to assist debugging.
779 lines of code changed in 8 files:
Working towards correct version with auxiliary variables.
75 lines of code changed in 1 file:
Marc changes in the GMRFBlockUpdater require the scaleFactor to be greater than 1.0. Changed the error message in the parser to reflect this.
1 lines of code changed in 1 file:
Added configurable optimization schedules and got the JointOperator working. Optimization schedule can proceed as 1/N, 1/sqrt(N) and 1/log(N). One specifies the option in a new "optimizationSchedule" attribute in the <operators> constructor. Choices = "default" (1/N), "sqrt" and "log". Also messed around a bit with operator performance output in MCMC to handle joint operators with multiple coercible transition kernels.
1055 lines of code changed in 6 files:
Fixed annoying logging problem in GMRFSkyrideBlockUpdater.
1 lines of code changed in 1 file:
Modified MH-Ratio ratio in the GMRFBlockUpdater so that the Joint Operator works correctly.
25 lines of code changed in 2 files:
Unrolled the loops in two of functions. Results in slightly better performance.
1959 lines of code changed in 2 files:
Better error message
1 lines of code changed in 1 file:
Changed coercible parameter so that a value ~ 0 means very small changes to precision and so that Rue's scalar is >= 1 per his book.
369 lines of code changed in 1 file:
These are the Java backend files for the native AminoAcidLikelihoodCore.
132 lines of code changed in 2 files:
Created a native AminoAcidLikelihoodCore to speed up computations when using protein data. Also edited makefiles so that it compiles. I know that the makefile works fine for Linux, but I haven't tested it for MacOSx or Win32.
The native core drops runtime by 45%.
551 lines of code changed in 6 files:
Added to the comment for the logNormalPrior in DistributionLikelihood class so that the parameters mu, stdev, and meanInRealSpace are a little easier to understand.
11 lines of code changed in 1 file:
add JRI library. remove beastdev
5 lines of code changed in 1 file:
remove empty line
0 lines of code changed in 1 file:
Updated version build number primarily as a way of checking SVN committing.
2 lines of code changed in 2 files:
Had to revert some changes made in 944 to TreeAnnotator so that it works again. Added AncestralSequenceAnnotator which will resample internal node states and provide posterior summarizes.
39 lines of code changed in 2 files:
Ooops forgot to include this file in last update (JointOperator). Sorry for the additonal email.
1 lines of code changed in 1 file:
Enables multiple proposals before accept/reject. Helps to deal with high autocorrelation between parameter that use different operators.
84 lines of code changed in 1 file:
Fix something that looks like a bug and some intelliJ warnings
8 lines of code changed in 1 file:
Set an emphiracally better threshold for count until changing a weight
1 lines of code changed in 1 file:
Prints the seed out into the log-file header. Will make more easier recreating errors for users who refuse to manually set the seed.
2 lines of code changed in 1 file:
Java and native libraries for Java-to-R communication. Ideally, the 2D HPDs are not hard to implement directly in Java. This should be done and the dependency on R removed!
0 lines of code changed in 2 files:
Made some substantial mods to TreeAnnotator. If the native JRI libraries exist, TA will call R to calculate 2D HPDs. These are necessary for lat/long posterior summaries. If the natives do not exist, then TA should behave normally (I hope). Please don't be too piss if I broke something!
2346 lines of code changed in 10 files:
reduce all sequences to one nuc
104 lines of code changed in 1 file:
Add Model restore Listener. Currently a Likelihood which is not a model is not notifed when an operator fails and all models are restored. This creates a quite nasty bug. Now such an object can get notified via the new callback
41 lines of code changed in 5 files:
Change method name to something I understand
87 lines of code changed in 27 files:
example file
1476 lines of code changed in 1 file:
factor common code
15 lines of code changed in 1 file:
use generic get-attribute
14 lines of code changed in 1 file:
fix comment
2 lines of code changed in 1 file:
fix issue 53
3 lines of code changed in 1 file:
Added a 'Trait through time' plot. Give the name of any trait stored in the tree (including 'rate').
638 lines of code changed in 5 files:
Added a credit for Joseph to the Tracer about box.
1 lines of code changed in 1 file:
Big improvement to Tracer: You can now select trace files that have different traces and Tracer will show you the common set (irrespective of order). This allows you to compare (say) rootHeights from two different models even if the stats are in different orders.
Also fixed a bug where an exception was being thrown when selecting two trace files of different lengths.
188 lines of code changed in 10 files:
Comments. factor out some functionality in NumberColumn
27 lines of code changed in 2 files:
Add a constructor so that VDdemographicFunction can be used as a generic demographic
23 lines of code changed in 1 file:
Allow any tree. Previously only a TreeModel was accepted
5 lines of code changed in 1 file:
Allow an implicit one off offset.
5 lines of code changed in 1 file:
another fix for parser rules
7 lines of code changed in 1 file:
Added an inspectParameterValues, a less overhead method when only reading the parameter is required. Unfortuanetly there is no way to enfore this in Java (in C++ one returns a const reference or array)
42 lines of code changed in 2 files:
use an array of correct size in toArray()
5 lines of code changed in 1 file:
remove unused globals
3 lines of code changed in 1 file:
Added a replaceChild method.
Made beginEdit return edit state
fix bug in setRoot
49 lines of code changed in 5 files:
Fixed defualt attribut conversion. now it really works for integers
Added format attribute to column - currently percent and boolean formats implemented. Those are for display only (screen log) as Tracer parses only numbers.
158 lines of code changed in 4 files:
Added a logging mechanism to the GMRFSkyrideBlockUpdateOperator that keeps track of the Newton-Raphson algorithm.
54 lines of code changed in 1 file:
Added an MRCA trait statistic
130 lines of code changed in 3 files:
(partial) fix to parser rules
11 lines of code changed in 1 file:
fix parser rules
12 lines of code changed in 1 file:
Reverted a debugging edit committed by mistake
13 lines of code changed in 2 files:
Added an ExternalLengthStatistic.
145 lines of code changed in 4 files:
Some improvements for mixing with relaxed clock models
1 lines of code changed in 1 file:
May 2008 »