Released: 2016-04-10
- Fixes a different bug that could occur when using the "-eI" option. The contemporaries buffer was always cleared correctly between repetitions, and could cause scrm to throw an error (#104). Thanks again to @jackkamm for reporting this bug.
Released: 2016-03-23
- Fixes that bug that could lead to false results when the "-eI" option was used to include non-contemporary samples. In that case, scrm could skip the time interval between the MRCA of the contemporary samples and the time of the "-eI" event (if any), and neglect all events during that time. We recommend to repeat all simulations which used "-eI" with the updated version (#102). Thanks to @jackkamm for reporting this bug.
Released: 2016-02-07
- New command line option "--transpose-segsites" that prints the segregating sites matrix with rows repesenting mutations instead of individuals. This also adds the time at which the mutation occurred as additional information (#85).
- scrm now throws an error when mutation or recombination rate changes have invalid sequence positions (#82).
- The length of segments between recombinations is now always reported in non-scientific notation (#81).
- Improved the error message when lines could not coalescence because of missing migration or negative population growth (#87).
Forest.coalescence_finished
has not set totrue
after a pairwise coalescence event (#89). This had no effect on the command line version of scrm.- Fixes scrm's tests suite on 32 bit systems (#98).
Released: 2015-07-09
- scrm had extensive memory consumption when simulating trees in large models (#79). This is now fixed, it is however no longer possible to use "-O" and "-T" at the same time.
Released: 2015-06-04
- It is now possible to specify the approximation exact window in number of recombination events (#73).
- scrm now uses a conservative approximation by default (#75).
- Critical: Position based rates changes (
-sr
and-st
) were only applied to the first locus. All other loci are simulated with the rates from the end of the first one (#74). - In large models, the Newick trees could get larger than the character limit of a C++ string. In this cases, only incomplete trees were printed (#76).
Released: 2015-05-18
- Bug fix: Fixed a cache invalidation error that could lead to runs being aborted after 100k recombinations (#70). Thanks to Jerome Kelleher for report this.
- Minor: Small updates to the documentation and citation information.
Released: 2015-04-07
- New feature: Added flag "--print-model" which prints a textual representation of the demographic model for verification and debugging (#60).
- Bug fix: When multiple population splits and/or merges occurred at the same time, only one of them affected each line while the others were ignored (#61).
- New feature: Added support for partial population admixtures (
-eps
, #62). - In addition to using "-seed" scrm now also supports "-seeds", as ms also supports both.
Released: 2015-04-04
- Bug fix: Wrong population size where calculated when migration rates changes or population splits and merges occurred in an growth period (#56)
- Bug fix: When -es was used at time 0, scrm ran into an endless loop (#53)
- Improvement: Added an error message when a population size is set to 0 (#52)
Released: 2015-03-29
- Improved memory management (#36)
- Added option '-p' to set number of significant digits in output (#47)
- Switched to std::mt19937_64 as default random generator (#49)
- Support arguments in scientific notation (#50, #51)
Released: 2014-12-23
- Bug fix: Fix reproducibility problem when the large sample optimization was active (#42).
Released: 2014-10-23
- Bug fix: Implement missing '-m' and '-em' arguments (#32).
Released: 2014-10-21
- Improved the autotools configuration to support more compilers and operating systems (#10, #27).
- Changed the Oriented Forest summary statistic (#25).
- Various minor cleanups in the code base to simplify creation of an R package containing scrm (#29).
Released: 2014-09-10
-
New
oriented forest
summary statistic as suggested byJ. Kelleher, A.M. Etheridge, N.H. Barton (2014) Coalescent simulation in continuous space: Algorithms for large neighbourhood size, Theoretical Population Biology, Volume 95, August2014, Pages 13-23, ISSN 0040-5809, http://dx.doi.org/10.1016/j.tpb.2014.05.001.
- Optimized the generation of newick trees (#22) and use a buffer (#23)
Released: 2014-08-04
- Improved the handling & storage of contemporary nodes. This gives a huge performance boost if scrm is used with large sample sizes (>1000) (#20).
- Optimized scrm for use with many populations (#20).
- Added more automatic tests of the produced distribution of trees (#20).
Released: 2014-07-09
- Fixed an access to unmapped memory
Released: 2014-06-18
- Fix file permissions
- Remove clang warning suppression
- Added a man page
Released: 2014-06-02
- Option '-es' is now ms-compatible (#16).
- It is now possible to use 3 seeds (as in ms).
- Added help and version information.
- Small performance tweaks.
- Added variable recombination & mutation rates
- Very first node in the tree was assigned to wrong population
- The input time of "-eI" option was not scaled
- Fixed the scaling of growth rates
Algorithm passes all tests now, starting explicit versioning.