Hi all, Shane Blackett, Poul Nielsen, and I had a meeting today to discuss the simulation metadata specifications.
The following is a summary of the meeting: 1) PCEnv, which is due to be released in the near future, will only support boundIntervals (and properties on the bound interval target) and simulationName, but will not support linearSolver, iterationMethod, or multistepMethod, and so on. 2) tabulationStepSize does not belong in the specification, so it should be removed. Software such as PCEnv should preferably not interpolate onto a grid, but if it does, this should be defined elsewhere, and not in the simulation metadata specification. 3) Most of the discussion related to the properties which specify the solver to be used. Poul suggested that we may be able to use existing algorithm ontologies, such as GAMS or CACM, to specify the solver. It was noted that there are two types of classification, one of the type of mathematical problem to be solved (e.g. terms like stiff and non-stiff), and another describing the type of algorithm to solve the tree. I suggested that the model author could provide an ordered set of terms from either of the ontologies, and terms would be searched in order until the software finds an algorithm matching a term. It was noted that in this case, the model author and the software need to agree on at least some of the terms for this to work (Poul suggested a restricted dictionary). This topic needs further discussion. Best regards, Andrew Miller _______________________________________________ cellml-discussion mailing list [email protected] http://www.cellml.org/mailman/listinfo/cellml-discussion
