Propose Shape and Geometry Base Classes

[aalbino2]

With the new geometry base classes from @mkuehbach we can now tackle also the issue #7, too

The NXcg_hexahedron_set.yaml seems ready to be used in an NXsample base class.

Most of the use cases encountered just need this base class.

We may want to include also NXcg_cylinder_set.yaml, or other shapes in the future.

Concerning this, there is a point we touched in one of the last meetings: how to include the choice among multiple geometry base classes in the NXsample? @mkuehbach was mentioning a container class for all the cg shapes, another way is to list all of them as optional, or to use some enumeration. We may discuss it next time

[mkuehbach]

Great to hear Andrea that this is useful. I agree, let's discuss the last point next time

[aalbino2]

In today's meeting the need of a connection class arose.

After defining a geometric locus, a complementing but detached information is needed describing the connections among the set of points, surfaces, polihedra, or whatever composes our geo description.

This is again a challenging task if we want a common class for area A, B, C.

The main needs are:

• Area A: a rather simple description of layer stacking or sample portions description (spatial localization may here be needed, too)

• Area B: a complete and articulated description of grains, boundaries, meshes, that overlap, intersects, stack into a macroscopic sample (https://doi.org/10.48550/arXiv.2205.13510)

• Area C: connections among points to instantiate molecules or moieties, connection among molecules or moieties to instanciate higher order aggregates (like amino acids making proteins) (https://doi.org/10.1007/s11047-020-09801-7). A cg_point_set instance containing the connection class can be an entity that can again be connected to other entities thanks to a new instance of connection class.

Further discussions (feel free to modify and add bullet points):

• [ ] How to tackle the problem of spatial localization, together to topological connection?
• [ ] Geometric description of electron density or spin density from .cube files (or similar formats)
• [ ] inclusion of approaches to describe set of atoms via fitting functions e.g. https://www.sciencedirect.com/science/article/pii/S0969212605000584 or https://doi.org/10.1063/1.5111045

[mkuehbach]

Very nice summary Andrea, it should be noted here that the example here from Area C specifically shows what is usually considered as the secondary structure of proteins, alas there are further hierarchies: https://en.wikipedia.org/wiki/Protein_structure, it would be nice to capture in this case all four levels of the hierarchy. What Luca Ghiringhelli also mentioned were the inclusion of approaches to describe set of atoms via fitting functions e.g. https://www.sciencedirect.com/science/article/pii/S0969212605000584 or https://doi.org/10.1063/1.5111045

[aalbino2]

Thanks Markus for your comment, yes actually this connection class should accept simple or complex entities to build up to the desired level of hierarchy needed, shouldn't it? The last point you mention, I will include in the bullet point of the above comment

[tomio13]

A nice summary Andrea! I would believe that Area A will go beyond films and layer stacks and reach the point that in some cases descriptions equivalent to that of Area B are needed. Which actually simplifies the situation a bit I hope.