Use Cases

[sbillinge]

UC 1 - compute amd from a structure

1. Simon wants to compute the amd from a structure as he is working in diffpy-cmi
2. Simon passes the structure to diffpy.similarity (ds)
3. ds computes the amd and returns it simon

UC 2 - pdd

1. as UC1.-1.3 but Simon wants to compute the pdd

UC 3 - amd-similarity

1. Tina wants to compute the amd similarity (amd-s) between two structures
2. Tina gives both structures to amd_s()
3. amd_s computes the amd-s and returns it to Tina

UC4 - pdd

1. As UC3.1 - 3.3 but Tina wants the pdd-s

UC 5 - to scale

1. As UC3 but a pairwise computation of amd or pdd (or whatever similarity measure we have in the future) over a large set of structures with good performance

[Sparks29032]

Desirable functions:

• Take in structure objects and return x metric (e.g. amd)
• Take in two structure objects and return the similarity given x metric (e.g. amd_s)
  • Can also take n structure objects and return a similarity matrix
    • Background for performance issues (i.e. do not want to recompute n amds)

[sbillinge]

UC6 - paper UC

1. Andrew wants to compare how different metrics perform on a set of structures
2. Andrew makes a query on his database of structures to return a set of structures
3. Andrew's db program returns a set of structures as structure objects
4. Andrew tells ds which similarity metrics he is interested in
5. Andrew gives the set of structures to diffpy.similarity (ds)
6. ds computes the similarity between all structures with each similarity metric
7. ds returns the set of correlation matrices between pairs of similarity metrics

[sbillinge]

UC7 Till wants to figure out which is the best metric for his needs

1. Till has a set of MOFs where he has some feeling about how similar they are
2. Till gives his set of mof structure to ds
3. Till requests a visualization of the similarity map
4. Till requests a set of measures
5. Till requests a reference material to compute distances to
6. ds computes the similarity map by selecting an origin
7. ds returns a the similarity map for each metric
8. Till zooms and scrolls the map
9. Till mouse-overs dots and they tell him what the structure was
10. Till specifies a distance threshold and ds outputs all the structures that are within that distance of the target

[Sparks29032]

Use this structure object: https://wiki.fysik.dtu.dk/ase/index.html

[sbillinge]

Before we get too deep into using ASE structure objects, can we check that they provide a complete description for our needs moving forward. In other words, can we do ase_structure_object(s) <-> diffpy.structure_object reversibly? Or is some key information lost when we go diffpy.structure_object -> ase_structure_object(s) so we can't go back?