BUG FIX: RNG now uses RNGScope to properly get and put the RNG state before calling RNG functions (fixes #8, thanks @bbolker!).
DOCUMENTATION: Added example for compile_reactions()
.
autoplot.ssa()
to plot_ssa()
for compatibility with R 4.0.BUG FIX: Zap small negative propensity and state values.
FEATURE autoplot.ssa()
: Allow plotting the firings.
DEBUG FEATURE ssa()
: Allow returning the GillespieSSA2 instead of running it.
BUG FIX: Use fabs()
instead of abs()
to calculate the absolute value of a floating point value.
BUG FIX: Precompiling returns a list of compiled function pointers, instead of a single function pointer that is secretly an array of function pointers.
GillespieSSA2 is now on CRAN!
MAJOR CHANGE: Split up Rcpp code to make separate parts easier to test.
TESTING: Write unit tests for many of the functions.
MINOR CHANGE: Renamed ssa_direct()
to ssa_exact()
.
MINOR CHANGE: Store firings, buffer and propensity objects only if requested.
BUG FIX: limiting
variable in ssa_btl()
should be an integer, not a double.
MINOR CHANGE: Timer now has millisecond accuracy, instead of second.
MAJOR CHANGE: Remove nu
and propensity_functions
from ssa()
, instead expect a list of reaction()
objects. This function provides a much more natural interface to specifying the effect and propensity of a reaction.
MINOR CHANGE: Apply small allocation optimisations to ssa_btl
, ssa_etl
and ode_em
.
ssa_em()
to ode_em()
.BUG FIX: Fix isinf scope issue for Windows users.
MINOR CHANGE: Move ggplot2 to Suggests.
Complete rewrite of the package:
The main SSA function and all SSA methods have been implemented in Rcpp.
User-defined propensity functions get compiled to Rcpp at runtime.
All SSA methods now assume the state-change matrix nu
to be sparse.
MINOR CHANGE: Added automated testing by travis.
MINOR CHANGE: Fixes to documentation.
MINOR CHANGE: Also output propensities.
Initial beta release of GillespieSSA2:
ode_em()
, an Euler-Marumaya ODE method.