simudyne.core

Interface Model

All Known Implementing Classes:

AgentBasedModel

public interface Model

The Model is the core abstraction around which the Simudyne engine is designed. A model is considered a stateful unit, which can run some calculation once or more times, with there being optional setup and disposal phases before and after the computational phase.

A traditional regression model could be considered a single-step model, with all data being provided before/during setup, and then a single step phase where the outputs are derived.

In comparison, an iterative model can follow the same procedure, but would take initial conditions before/during setup, and then each step would take the current conditions and move forward a single delta. Iterative models can then behave in non-linear ways, exposing feedback loops as each step is recursively dependent upon the previous.

Paradigms of modeling may be introduced and orchestrated inside step in a modular fashion. Calculate may then represent agents moving forward one time step, the processing of a window of events in a discrete event model, etc.

Method Summary

Modifier and Type	Method and Description
default void	dispose() An optional lifecycle point triggered when a model is restarted or explicitly closed.
default ModelConfig	getConfig()
default void	setup() An optional lifecycle point at which to take configured initial parameters and initialise required internal state.
void	step() Required lifecycle method for implementing the calculation of a model.

Method Detail

setup

default void setup()

An optional lifecycle point at which to take configured initial parameters and initialise required internal state.

dispose

default void dispose()

An optional lifecycle point triggered when a model is restarted or explicitly closed. Can be used to clean up any opened resources from setup.

step

void step()

Required lifecycle method for implementing the calculation of a model.

step will be called once or more times, with the external user or harness able to read outputs, change inputs, etc. in-between calls. Each step represents the moving forward in time, but the explicit notion of progress is left as specific to the implementing model.

getConfig

default ModelConfig getConfig()