I'd like to see a fundamental term in the specification that I'd call AnalysisType for the type of experiment. Not all experiments are time domain simulations: there are steady-state simulations, optimizations, sensitivity analysis, uncertainty quantification etc. It is fine to an AnalysisType TimeDomainSimulation as the first and most important one, but when we deal with UQ for parameters, I would think that most of those are done via steady-state simulations as well.
The following analysis types should be possible, and it should be possible to extend this:
- transientSimulation
- steady-state
- parameterOptimization
- setPointOptimization
- sensitivityAnalysis
- trajectoryOptimization
- linearization
This should be extended and harmonized soon with the output from the OpenScaling project for UQ-type experiments, once they have something published to work against.
I'd like to see a fundamental term in the specification that I'd call AnalysisType for the type of experiment. Not all experiments are time domain simulations: there are steady-state simulations, optimizations, sensitivity analysis, uncertainty quantification etc. It is fine to an AnalysisType TimeDomainSimulation as the first and most important one, but when we deal with UQ for parameters, I would think that most of those are done via steady-state simulations as well.
The following analysis types should be possible, and it should be possible to extend this:
This should be extended and harmonized soon with the output from the OpenScaling project for UQ-type experiments, once they have something published to work against.