unum-cloud
Feature: Add DiGraph, MultiGraph and MultiDiGraph graph types in Python bindings
Describe what you are looking for
I am looking for DiGraph, MultiGraph and MultiDiGraph graph types in Python bindings
Can you contribute to the implementation?
- [X] I can contribute
Is your feature request specific to a certain interface?
Official Python bindings
Contact Details
Is there an existing issue for this?
- [X] I have searched the existing issues
Code of Conduct
- [X] I agree to follow this project's Code of Conduct
@davvard, can you please share the table describing expected behavior for edge ID generation here?
| Graph (node_ids, edge_ids) | Edge Ids |
|---|---|
| Graph (false, false) | None |
| Graph (true, false) | None |
| Graph (false, true) | Auto Generated |
| Graph (true, true) | Auto Generated |
| DiGraph (false, false) | None |
| DiGraph (true, false) | None |
| DiGraph (false, true) | Auto Generated |
| DiGraph (true, true) | Auto Generated |
| MultiGraph (false, false) | User Defined (Auto Generated) |
| MultiGraph (true, false) | User Defined (Auto Generated) |
| MultiGraph (false, true) | User Defined (Auto Generated) |
| MultiGraph (true, true) | User Defined (Auto Generated) |
| MultiDiGraph (false, false) | User Defined (Auto Generated) |
| MultiDiGraph (true, false) | User Defined (Auto Generated) |
| MultiDiGraph (false, true) | User Defined (Auto Generated) |
| MultiDiGraph (true, true) | User Defined (Auto Generated) |
So node_ids means "labeled nodes" and edge_ids means "labeled edges"? And Edge Ids means "how to generate edge IDs"?
Gotcha, @davvard . Can you please patch it into a more readable form with 4 columns, so that its easier to reason about it.
Is it a good idea to have both auto-generation and user-defined IDs in the same class? In that case the user may end up accidentally overwriting and edge we have created... isn't that the case? Is this how NetworkX works?
@ashvardanian Yes I think it is good idea...if you create edges with attributes with auto-generation and after it you add edge with ID that already exists the program will merge edge attributes...This is how NetworkX works. Example
db = ustore.DataBase()
graph = ustore.MultiGraph(db, 'multigraph', relations='edges')
graph.add_edge(1, 2, name='edge1', weight = 2)
graph.add_edge(1, 2, key=0, weight=3)
print(list(graph.edges(data=True))) # [(1, 2, {'name': 'edge1', 'weight': 3})]
Gotcha! Full compatibility is great! And how do they generate the IDs, @davvard ?
| Class Name | Labeled Edges | How to generate IDs? |
|---|---|---|
| Graph | ❌ | None |
| Graph | ✅ | Auto Generated |
| DiGraph | ❌ | None |
| DiGraph | ✅ | Auto Generated |
| MultiGraph | ❌ | User Defined (Auto Generated) |
| MultiGraph | ✅ | User Defined (Auto Generated) |
| MultiDiGraph | ❌ | User Defined (Auto Generated) |
| MultiDiGraph | ✅ | User Defined (Auto Generated) |
@davvard That sounds ugly... as we would either need to store a counter for every pair... or retrieve the list of all the neighbors every time. Which IDs will NetworkX generate for the edge between nodes A and B ?
- add A-B, add A-B, remove last, add A-B
- add A-B, add A-B, remove first, add A-B