Calibrate entry resistance of LeakyLineDoublets

[MattBrst]

I came across a situation where I wanted to calibrate the entry resistance of a Leaky Line Doublet.
Am I correct that this is not possible with the cal.fit() of TTim currently?

This would be my conclusion, since supported parameters are only 'kaq', 'Saq', 'c', 'Sll' and 'kzoverkh'.

We might put this wish somewhere on the nice-to-have list.

[dbrakenhoff]

This is possible (with some limitations), but in a "slightly" more cryptic way:

There is a method called set_parameter_by_reference() in the Calibrate class that allows you to set calibration parameters based on an array. Unfortunately, the resistance res of a LeakyLineDoublet is not an array, but the derived computation variable resfac is (thickness aquifer divided by resistance).

So presumably something along these lines should work.

cal.set_parameter_by_reference("ld_res", parameter=ld.resfac, initial=5.0, pmin=1.0, pmax=10)

Note that this doesn't work so well when your line doublet spans multiple layers with different thicknesses, since it will set the parameter to be equal in each layer.

For an explanation of set_parameter_by_reference() see this notebook.

[mbakker7]

Please pay close attention to the way the parameter is set by reference in the example notebook. It must be a window into an array. So in the above example, it should be parameter=ld.resfac[0:1], assuming resfac is an array.

[dbrakenhoff]

Please pay close attention to the way the parameter is set by reference in the example notebook. It must be a window into an array. So in the above example, it should be parameter=ld.resfac[0:1], assuming resfac is an array.

Oh so you can calibrate a single layer using slices! Neat :)

[MattBrst]

Good to know!

Maybe we should add information in the cal.set_parameter() which makes it easier to find the cal.set_parameter_by_reference() function? For example:

1. Just add a pointer in the docstring
2. Check for parameters and raise an error message

[mbakker7]

Adding a pointer -> very good idea.
What did you have in mind with the second suggestion. We already raise an error when a parameter name is supplied that has not been implemented: "Supported parameters are 'kaq', 'Saq', 'c', 'Sll' or 'kzoverkh'."

[MattBrst]

My thoughts with the second suggestion: Change the error message error when a parameter name is supplied that has not been implemented to:

"Supported parameters are 'kaq', 'Saq', 'c', 'Sll' or 'kzoverkh'. Note that other parameters may be calibrated through the set_parameter_by_reference() function."