search

SimulationEverywhere / devstone

Implementation of the devstone benchmark on multiple simulators
BSD 2-Clause "Simplified" License
1 stars 3 forks source link

Error in Conceptual structure of LI and HI models #5

Open juanlanuzag opened 5 years ago

juanlanuzag commented 5 years ago

For the LI model, this paper says " In LI models of depth d and width w, we have d coupled components with w-1 atomic components each (except for the innermost one, with only one atomic component)."

In cadmium-ref-LI.cpp it says it's W=3, D=3 but its actually W=4, D=3 (one of the 4 models from W=4 is the coupled one). From DEVS Diagrammer: image

This change has to be made in the cadmium-devstone.cpp too.

For the HI model, cadmium-ref-HI.cpp is also a W=4 D=3 instead of a W=3 D=3. Also, the links between atomic models are wrong: image

There should be only one link between the output of _atomic_L20 to the input of _atomic_L21 and the is a link missing between the output of _atomic_L21 and the input of _atomic_L22

sdavtaker commented 5 years ago

It is possible you find more errors like that, the whole project was a draft I was working on. :) About the W/D, I read a few different papers and I remember they were not super consistent. I will double check with other papers and come with comments. Nice catch on the HI connections. Validating those models the way I was doing is super error-prone, but I didn't have access to the DEVSdiagramer yet. It will be definitely easier to do the test based in DEVSDiagramer, if you can prioritize that, it will save us a lot of time in the future.

juanlanuzag commented 5 years ago

Sure, when I have some time I'll work on making the tests based on the Devs diagrammer.

sdavtaker commented 5 years ago

Checked the ref, it is as you mentioned, lets make it that way and keep that paper as the ref.