|
|
Meeting --- 08/06/05
- NESTED TIME:
- During the previous meeting we discussed sub-experiments,
as an alternative to the more general Extended Time considered before.
A sub-experiment is a DEVS model specification that is embedded in an
atomic-DEVS variant (the Wrapper). The sub-experiment is performed
when a transition occurs in the Wrapper, thus actually creating a time base
orthogonal to the upper level's.
- This setup has been implemented in PyDEVS, with the sub-experiment
linked to the wrapper's external-transition.
- During the present meeting, we concluded that it would be preferable to
ling a sub-experiment to the wrapper's internal transition, so that a
stand-alone wrapper (i.e., not coupled to some generator) would be
meaningful.
- Considerations on data access led us to the notion of Experimental
Frame, that includes:
- a Collector, that processes the sub-experiment's trace to
produce a value usable by the higher level (the wrapper);
- a Monitor, that determines when a sub-experiments terminates,
- an Initializer, that initializes the sub-experiment.
- MULTI-TIME DEVS:
- Observing that the time-base of a DEVS is not constrained to the real
numbers, we came up with a nice idea of a time-base to be tested: DD/MM/YY (isomorphic
to the natural numbers).
- From a reusability point-of-view, this is important as the time scale of
an experiment can be seconds, while the time scale of another can be days.
We considered how such experiments could be coupled.
- VARIABLE STRUCTURES:
- Short discussion on Variable Structure DEVS ("it's in the eye of
the beholder!...")
- OVERVIEW:
- Finally, we reviewed/refreshed the project overview. A detailed outline
will follow.
TODO:
- Write a detailed outline of the project;
- Implement nested time using internal transition, and formalize the idea;
- Implement the time base example.
|
