Add reading and writing for graphs in DIMACS graph format #4591
Conversation
* Simple randomized maxcut heuristic * Remove approximation guarantee test Test is not expected to pass all the time due to randomness. * Preserve random state for cut heuristic * Add one exchange max cut heuristic * Fixes && Refactoring && Default deterministic output * maxcut/simple.py: Add docstring and make initial_cut optional. * maxcut/simple.py: Add shuffle again. Make initial set a set. Co-authored-by: Jonas Charfreitag <jcharfreitag@cs.uni-bonn.de>
* Fix parse_edgelist behavior with multiple attributes * fixed test case Co-authored-by: chris <chris@orchid>
* updated draw_networkx + added test * added newline * skip test if numpy is not installed * change skip if numpy is not available * switch elif to if
* add code for tree isomorphism, with tests * fix typo in comment * one more typo in comments * fix some PEP8 formatting, that flake8 didn't care about * rename files as tree_isomorphism * run code through black formatter * incorporate feedback from dschult in PR4067 * fix missing import for not_implemented_for decorator * swap edge order randomly in testing routine positive_single_tree * run black on test_tree_isomorphism.py * spacing tweak to allow CI test of docs Co-authored-by: Dan Schult <dschult@colgate.edu>
There was a problem hiding this comment.
Thanks for this!
I haven't looked very carefully yet but some quick thoughts/tips.
- connect the new module's docs to the document reference in
/doc/reference/readwrite/index.rstas well as a new filedimacs.rstto mimic the other readwrite sections. - The doc_string of
generate_dimacsdoesn't match the function signature... others? - The tests only use the node attribute "value". What about edge attributes and/or other attribute names?
- I don't understand the meaning of a line like
p 3 3in the dimacs file. I could/will look it up, but is there an easy way to describe what it does? Would it be worth a 1 paragraph description ofp # #,n # #,e # #? Or are the subtleties that make it worth having users read the format doc from the start?
Thanks for your feedback. |
There was a problem hiding this comment.
I only did a quick pass but a few things jumped out. I'm not familiar with the DIMACS format and though some material is referenced in the module docstring, I think it would be an improvement to have (if possible) a brief description of the format there as well.
| Format | ||
| ------ | ||
| The DIMACS graph format is a human readable text format. See | ||
| http://archive.dimacs.rutgers.edu/pub/challenge/graph/doc/ccformat.dvi |
There was a problem hiding this comment.
Is there a link to a hosted page instead of a downloadable .dvi file? If not NBD
There was a problem hiding this comment.
This is the most comprehensive one I could find.
networkx/readwrite/dimacs.py
Outdated
| assert ( | ||
| expected_num_nodes == g.number_of_nodes() | ||
| ), f"Expected to read {expected_num_nodes} nodes but found {g.number_of_nodes()}" | ||
| assert ( | ||
| expected_num_edges == g.number_of_edges() | ||
| ), f"Expected to read {expected_num_edges} edges but found {g.number_of_edges()}" |
There was a problem hiding this comment.
I think asserts should be avoided code, due to differences in behavior under certain conditions. This does seem a more specialized case than a ValueError though, since it's a final validation check before returning.
In this case, I think I would define a new exception for this module (say class DimacsValidationError(Exception)) and raise that on failure. Maybe other reviewers would have better suggestions though... please chime in!
There was a problem hiding this comment.
I added an excpetion.
use nx.tempty_graph Co-authored-by: Ross Barnowski <rossbar@berkeley.edu>
# Conflicts: # networkx/readwrite/__init__.py
Closes #4590
See here for details about the format.