Methods

A solver is described with four indices which specify: time dependence, category, method, and options. The mapping from these fourindices to the name of a command line solver is defined in state/simulationMethods.py. The time dependence indicates whether the solver may simulate time homogeneous or time inhomogeneous models. Time inhomogeneous models have reaction propensity functions that have explicit time dependence. The category indicates what kind of output the solver produces. There are seven possibilities:

Next is the simulation method: direct, next reaction, tau-leaping, etc. Most methods have a number of options. For instance, with tau-leaping you may choose to use adaptive step size or a fixed step size.

Below are the attributes for the method element. The identifier as well as the indices describing the time dependence, category, method, and options and mandatory. The rest of the attributes are optional.

  <method id="Identifier" timeDependence="Integer" category="Integer"
   method="Integer" options="Integer" startTime="Number" equilibrationTime="Number"
   recordingTime="Number" maximumSteps="Integer" numberOfFrames="Integer"
   numberOfBins="Integer" multiplicity="Integer" solverParameter="Number"/>

The starting time is of course when the simulation begins. The equilibration time is how long the simulation is advanced before recording the solution. The recording time is the length of time that the simulation is advanced while recording the solution. If a maximum number of steps is specified, then the solver will abort if this limit is exceeded. For solvers that record data at uniformly-spaced intervals, the number of frames defines the frame times. For solvers that generate histograms one needs to specify the number of bins. The solvers dynamically adjust the bin width to maintain this constant number of bins. For histogram output one may also specify the histogram multiplicity. A higher multiplicity allows one to more accurately assess the error in the empirical probablity distributions for the species populations. The default multiplicity is four. Some solvers require a parameter value. For example tau-leaping with an adaptive step size needs an error tolerance. Which solvers require a parameter is indicated in state/simulationMethods.py.