v0.1.7

Added

• it's possible to set explicit time dependence of rates now when starting a simulation

Fixed

• changed references to deprecated classes in examples and docstrings
• better handling of detecting explicit time-dependence in functional rates.

v0.1.6

Changed

• renamed a parameter in StochasticEpiModel.simulate to stop_simulation_on_vanishing_total_event_rate, because it describes the mechanism better (adjusted TemporalNetwork) accordingly

Fixed

• Disallowing resizing a visualization app on Linux OSes to avoid a recursion error
• metadata such as epipack.__version__ is now correctly given out

Added

• a SymbolicSEIRModel
• a routine to generate random geometric graphs
• the possiblity to add a callback function in StochasticEpiModel.simulate such that event tracking is possible

v0.1.5

Added

• methods to the IntegratorMixin class that allow integration until a stop condition is reached

Fixed

• specified for MatrixEpiModel that the leading eigenvalue of a Jacobian should be measured by largest real part
• specified for MatrixEpiModel that the leading eigenvalue of the next generation matrix should be measured by largest magnitude (because R is the matrix's spectral radius)
• in StochasticEpiModel, convert rate values to float by default

v0.1.4

Added

• A new small-world network styling method based on 1d lattice distance (in epipack.networks)
• methods to compute Jacobian and next generation matrices (NGMs) in MatrixEpiModel, as well as R0 from said NGMs (TODO: add docs)
• tests for these methods
• epipack.distributions module, which deals with fitting empirical distributions to sums of exponentially distributed random variables (still in dev mode, also TODO: add docs)
• tests for this module
• methods to EpiModel that save events that have been set. This will be used to generate model flowcharts with graphviz at some point
• the possibility to pass a function to StochasticEpiModel.simulate that checks for a custom stop condition

v0.1.3

• fixed dependency issues with pyglet, apparently the "shapes" module did not appear until lately. Defined a range of versions for pyglet
• fixed bug in example code in README.md

v0.1.2

1. paper review at JOSS done (openjournals/joss-reviews#3097)
2. fixed bug where the reset_events flag was ignored in EpiModel.set_processes()

v0.1.1

Added

• GeneralInteractiveWidget was added to allow interactive display of general functions
• a very basic SDE integrator was added (no diffusion coefficent matrix, and no system-dependent diffusion coefficients)

Changed

• InteractiveIntegrator can now plot derivatives
• Range and LogRange classes will behave like floats whenever necessary

Fixed

• behavior of the SamplableSet class
• a bug where the reaction rate of nodes in weighted networks is scaled by the node's degree and not by its strength

v0.1.0

First public version. Added temporal networks, an interactive jupyter widget and more symbolic methods.

Visualizations

Added a complete visualization framework for stochastic simulations.

Catch situations with endless loops

It might happen that the maximum total event rate is non-zero but the true total event is zero. In this case, a mechanism now counts the number of unsuccessful events. If this number is above a threshold, the true event rate is evaluated and checked.