Modeling Sampling Error

[researcher]

@junpenglao

I would say it is reasonable, as scaling the input change the measure in the parameter space: remember the volume stays the same under different parameterization, which means the uncertainty might scale differently.

Using less samples didn’t change the outcome.