[root]/src/dr/inference/markovjumps
Author | Changes | Lines of Code | Lines per Change |
---|---|---|---|
Totals | 55 (100.0%) | 1380 (100.0%) | 25.0 |
msuchard | 52 (94.5%) | 1366 (99.0%) | 26.2 |
rambaut | 3 (5.5%) | 14 (1.0%) | 4.6 |
Consistently clearing complete histories within CodonPartitionedRobustCounting
8 lines of code changed in 2 files:
Cleaned up the format of the complete history logging. Now has {{SITE,TIME,FROM_STATE,TO_STATE},...}. I am aware this will mess up people's processing scripts but it was a mess before (unnecessary nesting of parentheses and '-1's all over the place).
3 lines of code changed in 1 file:
Cleaned up the format of the complete history logging. Now has {SITE,{TIME,FROM_STATE,TO_STATE},...}. I am aware this will mess up people's processing scripts but it was a mess before (unnecessary nesting of parentheses and '-1's all over the place).
8 lines of code changed in 1 file:
Fixing inconsistency when reporting a filtered state-change history
56 lines of code changed in 3 files:
Standardizing format for change events; these should always be an array of arrays because the number of events is variable.
3 lines of code changed in 1 file:
Possible new string format for change events
38 lines of code changed in 1 file:
Logging of complete histories to trees has a more consistent format (a set of nested arrays).
3 lines of code changed in 1 file:
Fixing reporting of site- and branch-specific S and N counts
20 lines of code changed in 1 file:
Improved user-friendliness of robust counting errors
2 lines of code changed in 2 files:
Added flexibility to check numerical stability of BSSVS substitution models
11 lines of code changed in 1 file:
Playing with complete history logging; honeymoons are meant for bug-fixing
47 lines of code changed in 2 files:
Re-implemented logging of random local clock indicators in tree log files; this ability was lost when we converted over to tree traits.
1 lines of code changed in 1 file:
Fixed incorrect calculations when using a BranchSubstitutionModel with MarkovJumps
2 lines of code changed in 1 file:
Uniformized simulation now falls back on rejection sampling when numerically instable.
22 lines of code changed in 3 files:
Added logging ability for complete substitution history under uniformization; there is a high likelihood that I broke some of the Markov jumps/reward mechanisms in this update.
26 lines of code changed in 3 files:
Tracking down numerical instability for very long and very short simulation times under Hobolth's algorithm
11 lines of code changed in 1 file:
Modifying CodonPartitionedRobustCounting to differentiate between internal and external branches; minor optimization of uniformized state history simulation
25 lines of code changed in 2 files:
Fixed bug in rewards; JUnitTests are awesome
1 lines of code changed in 1 file:
Fixed store/restore of subordinated process for Uniformization; also fixed reward process for Uniformization
34 lines of code changed in 2 files:
Better immediate calculation caching for MarkovJumps
47 lines of code changed in 2 files:
MarkovJumpsBeagleTreeLikelihood now supports all CTMC that can be uniformized (I believe this is all finite-state models).
9 lines of code changed in 2 files:
Completed UnitTest for simulation via Uniformization; added UniformizedSubstitutionModel to connect MarkovJumpsTreeLikelihood with the Uniformization method
1 lines of code changed in 2 files:
Added methods to simulate end-conditioned CTMC using the Uniformization method; this is the most optimal, simulation algorithm available and should make Stochastic Mapping and Coalescent-with-Migration models much, much faster (when real eigen decompositions are not available ... if the model is real diagonalizable MarkovJumps is still the way to go)
341 lines of code changed in 4 files:
Can now simulate (and report) the complete history of any model in BEAST, most importantly codon models (with branch rate variation) and report the true number of synonymous and nonsynonymous mutations in a synthetic dataset.
33 lines of code changed in 2 files:
Further work for RobustCounting
220 lines of code changed in 2 files:
Trimming code from MarkovJumpCore
49 lines of code changed in 2 files:
Added JUnitTest for counting synonymous jumps; also refactored a function in making codon models to avoid duplicating code
0 lines of code changed in 1 file:
Intermediate steps in implementing robust counting of S and N using nucleotide models
7 lines of code changed in 1 file:
Refactoring MarkovJumpsCore after Asger Hobolth reminded me that one wants to compute more than just the mean process
34 lines of code changed in 2 files:
(1) Small fix to trait rate Gibbs operator and (2) initial classes for robust counting on codon partitioned models
35 lines of code changed in 1 file:
MarkovJumps now supports Markov reward processes. Also a typo fix in BeastMain and some unnecessary casting in AMTL.
44 lines of code changed in 2 files:
Interfacing markovjumps to a TreeLikelihood for jump reconstruction
6 lines of code changed in 1 file:
Refactoring markovjumps into a core and a wrapper
233 lines of code changed in 1 file: