Approach to percolation with networkx graphs

[SCiarella]

Gemdat collects infromation about the transition into the jump class. It is then possible to

• [x] create a graph from jump
• [x] estimate the activation energy from the MD trajectory using an empirical approach: $$ A_ab = -k_B T log( #jumps_ab / t_occupation) $$ and this activation energy can become the edges of the graph

Associated to this graph there will be two types of labels:

• edge labels: corresponding to the phase (?)
• node labels: corresponding to the environment (e.g. tetra/octa) the labels (in particular the node labels) should consists of lookup tables provided by the user, corresponding to the specific material structure.

If the above is implemented, it would be possible to identify percolating paths of this graph using standard graph techniques (like colouring the periodic images), while excluding/including specific parts of the graph.

---

• [ ] #328

[stefsmeets]

For the node labels we now have the site label, and for the edges we have the activation energy.