Nicholas Landry
Nicholas Landry
This could be useful in standardizing our documentation: https://github.com/jshwi/docsig
The current data structure is inherited from NetworkX and is key:val pairs of node ID:[x, y]. In the drawing functions, it can be inefficient to draw with a for-loop and...
We should look into how we can potentially use the same functions in XGI for different classes.
We should have `Hypergraph` inherit from `DiHypergraph` or vice-versa as well as `IDView`/`DiIDView` to avoid overly duplicated code.
A natural input type to support would be a dictionary, e.g., ``` H = xgi.Hypergraph() H.add_nodes_from({1, {"name": "a"}, 2: {"name": "b"}, 3: {"name": "c"}, 4: {"name": "d"}}) ``` This input...
For example, `H.edges.members` doesn't have an `asdict()` or `aslist` method.
Some generators have the "hypergraph" in the name, some have acronyms, etc. It would be good to have a more consistent naming scheme. See #286.
Currently, some XGI methods support empty edges and some do not. Ones that do: * `dual()`: the dual of a hypergraph with isolated nodes must have empty edges. * `add_edges_from()`...
It would be nice to have the ability to specify weights for node-edge pairs so that we could, among other things, implement something like this: https://dl.acm.org/doi/abs/10.1145/3340531.3412034
According to https://github.com/conda/conda, to do this, we need to * make a recipe: https://docs.conda.io/projects/conda-build/en/stable/concepts/recipe.html * have an account on anaconda.org and have anaconda-client downloaded * build the recipe * upload...