Add 2nd order polynomial supermirror efficiency function

[SimonHeybrock]

Fixes #65

I slightly extended the Zoom notebook with a usage example.

[jokasimr]

(minor) second degree polynomial?

[SimonHeybrock]

Thanks, fixed!

[jokasimr]

Maybe it's better like this, not sure. But alternatively we could keep the parameters dimensionless and deal with the units in the __call__ method. The advantage is that we / the users don't have to care about the parameter units.

[SimonHeybrock]

Maybe it's better like this, not sure.

The core philosophy of Scipp is to avoid values without units wherever possible. It avoids really bad and hard to track-down bugs, as well as being self-documenting. Fewer bugs + documentation is what we chose over minor inconvenience.

[jokasimr]

Oh but I didn't suggest not having units. I suggested to define the parameters as dimensionless. We would still do unit checks in the __call__ function.

[jokasimr]

But sure, it's probably better like this.

[SimonHeybrock]

Oh but I didn't suggest not having units. I suggested to define the parameters as dimensionless.

Don't the values of the parameters depend on their units? That is, if the user has in mind "b is 123/nm" and sets b=123 we get nonsense?

[astellhorn]

But you are also testing wrong powers, e.g. a, b, c all in 1/angstrom (first part of the test) - and that does not give an error?

[SimonHeybrock]

You mean code like this?

    with pytest.raises(sc.UnitError, match=" to `dimensionless` is not valid"):
        eff = pol.SecondDegreePolynomialEfficiency(
            a=sc.scalar(1.0, unit='1/angstrom'),
            b=sc.scalar(1.0, unit='1/angstrom'),
            c=sc.scalar(1.0),
        )
        eff(wavelength=wav)

Note the with pytest.raises(sc.UnitError, match=" to dimensionless is not valid"): --- this checks that the code raises the expected exception.

[astellhorn]

Ah ok, and the following code which in principle is correct from dimensions of a,b,c checks that it does not comply with some wavelength/time , i.e., for comparing a,b,c in their respective angstrom units with a "wavelength" in angstrom/s ?

with pytest.raises(sc.UnitError, match=" to `dimensionless` is not valid"):
eff = pol.SecondDegreePolynomialEfficiency(
            a=sc.scalar(1.0, unit='1/angstrom**2'),
            b=sc.scalar(1.0, unit='1/angstrom'),
            c=sc.scalar(1.0),
        )
        eff(wavelength=wav / sc.scalar(1.0, unit='s'))

[SimonHeybrock]

Correct!

[astellhorn]

Ah, perfect, understood! Will be tested on the zoom data this week :)

[jokasimr]

Looks good to me. Left some minor comments.