ADR Suggestion, Generic Theory Class

[rozyczko]

Another discussion easyscience/EasyDiffractionLib#97 goes into design details of the Phase class, used to manipulate the atomistic object, with properties like atoms, cell, background etc.
This is a good picture for diffraction and spectroscopy (and Bragg Edge), where the underlying model is a system composed of atoms.
However, for techniques like reflectometry, this isn't the case. Here, the system is a set of layers and surfaces.
For QENS this is even more simple, with the model being a set of simple functions.

This means the base class for this descriptor should be a bit simpler than the Phase we discussed.
For starters, it can't be called Phase. How about calling this a Theory? Or specifically BaseTheory, considering those are models we use to generate the theory curve to compare/fit to the experiment.

Diffraction would use it as a base class for Phase, refl for Layer etc.

[AndrewSazonov]

Why not just use model? Model and BaseModel seem more natural to me. And, regardless of the technique, there is always a physical model, which can be a phase, crystal, layer, etc.

[rozyczko]

I guess because model is already a very loaded term.
We use "model" to describe multiple objects - Qt models, data models for calculators, fit models etc.

Adding another concept of model would make it more confusing, I think.

[AndrewSazonov]

I agree that there are already a number things called model in our projects, and it can be confusing. But if I saw an object called theory, it would confuse me even more. And at the moment I personally don't see a better name than model. Maybe there are other ideas besides model and theory?

[rozyczko]

But it is a theory, right? We have experimental data to which we fit our theoretical description of the system.
The profile we get from the atomistic model is a theoretical profile. So just like for experimental data and parameters we have Experiment, for theoretical data and parameters we should have Theory.
Again, this is about comparing theory to experiment.

[rozyczko]

TheoreticalModel

Grudgingly accepted consensus.

For technique-specific implementations we would derive from TheoreticalModel to create say
Sample or Phase for diffraction
Layer for reflectometry
etc

[rozyczko]

... and also covered in more detail and depth in https://github.com/orgs/easyscience/discussions/29

Therefore, closing this discussion.