Solving models in a Python loop vs adding a parameter sweep and solving that instead

[oarcelus]

That is basically the question. Is there any significant overhead related to solving a given model in a loop for different parameters that might be saved in a python list vs adding a paramater sweep in the model using the Java API with the mph library and adding all the parameters there instead?

Additionally, in the case wanting to parallelize the sweep using the 'Distributed Parametric Sweep' box in the Parameter Sweep node (for example 3 parameters with 2 cores for each parameter), will initializing a single client with cores=6 and evaluating it effectivelly run the distributed parameter sweep?