Test interpretation of parameterScale

[paulflang]

I am wondering if it would be good to add a test case that checks if an optimizer interprets the (lowerBound, upperBound and) nominalValue of log and log10 scaled parameters on a linear scale.

[dweindl]

Would surely be good to test. To test the correct usage of parameter bounds, we'd have to add actual parameter estimation tests in addition the current plain simulation tests. Should be easy though to construct simple optimization problems to check that.

[paulflang]

I think a test for nominalValue might be sufficient. An optimizer that does the nominal correctly, likely does the bounds correctly anyway. Additionally, if you'd have a lowerBound of 1, and upperBound of 100 and a nomialValue of 9, in a log10 parameterScale, then an optimizer that interprets the nominalValue correctly would probably fail to construct or optimize* the problem (i.e, it cannot make the mistake and assume the bounds are 10^1 and 10^100, cause the nonimalValue would be outside the bounds, which would (hopefully) cause an error).

* Perhaps it should be in the README that the tests only test simulation, and that it is recommended, in addition to the simulation tests, to check if a (very short) optimization run does not error.