Sample from new distributions of initial conditions

[jonrkarr]

• DistributionInitConcentration.mean, DistributionInitConcentration.std
• Compartment.mean_init_volume, Compartment.std_init_volume

[jonrkarr]

Done:

• Volume: Volume sampling from normal distribution implemented in DynamicCompartment.__init__
• Species counts: Species count sampling from normal distribution implemented in ModelUtilities.concentration_to_molecules

Todo:

• The species counts must be initialized conditioned on the sampled compartment volumes. This requires refactoring the order in MultialgorithmSimulation.__init__ in which the compartment volumes and species counts are initialized
• It would best to refactor the initialization of simulation state so that it is separate from the construction of simulation objects. This will allow the establishment of the object graph in the construction of the objects and then initialization of simulation state using the object graph.
• It would also be good to rename some of these methods to indicate their role in the initialization of simulations state. E.g. ModelUtilities.concentration_to_molecules doesn't clearly indicate its role in the initialization of species counts conditioned on initial compartment volumes.

[artgoldberg]

initialization done refactoring and renaming to be considered

[artgoldberg]

WRT "It would best to refactor the initialization of simulation state so that it is separate from the construction of simulation objects. This will allow the establishment of the object graph in the construction of the objects and then initialization of simulation state using the object graph", I see your point.

However, in my implementation I addressed this issue by partially initializing some dynamical objects, like DynamicCompartments and the species populations accessed by DynamicSpeciess, before creating the remaining dynamic objects and preparing them for simulation. Thus this refactoring isn't urgent now and I'm moving it to a new issue.