What is the difference between shortest_simple_paths and all_shortest_paths?

[TOMOFUMI-KONDO]

I searched the method to find k the shortest paths in the networkx graph, and found functions named 'shortest_simple_paths' and 'all_shortest_paths'. It seems that Both of them are used to find all the shortest paths between specified source and target nodes. What is the difference of those methods? How I decide which to use.

[dschult]

The function shortest_simple_paths yields all the simple paths in order of length. So most of the resulting paths are not "shortest paths". To explain the name, think of "I want the 6 shortest simple paths".

The function all_shortest_paths yields all of the possible shortest paths. All of these paths will have the same length. The use for this function is when there are multiple ways to go that are the same distance. Instead of just giving one example of a shortest path, this function provides them all.

Perhaps the docs should include a comparison of these functions... Anyone know if a paragraph be included in a Sphinx "See Also" section?

[TOMOFUMI-KONDO]

Thank you. I understood the difference of two functions.
I agree that docs should include a comparison of them.

[rossbar]

Perhaps the docs should include a comparison of these functions... Anyone know if a paragraph be included in a Sphinx "See Also" section?

A text description can be added to entries in the See Also section, though I think the intent is to support short descriptions rather than paragraphs. That said, more documentation on how the shortest path functions related to one another would certainly be useful. Another potential place to beef up documentation would be the shortest_path module/package docstring. IMO it'd also be a great candidate for an nx-guide to provide a more in-depth, executable example to highlight what each of the functions does.