Add class CircuitsMatroid #37148
Add class CircuitsMatroid #37148
Conversation
gmou3
force-pushed
the
circuits_matroid
branch
from
41a7d54 to
ed0f386
Compare
gmou3
force-pushed
the
circuits_matroid
branch
from
ed0f386 to
b17a2e3
Compare
1-line outputs and iterable to SetSystem
gmou3
force-pushed
the
circuits_matroid
branch
from
4ae07e2 to
bed97d8
Compare
src/sage/matroids/unpickling.pyx
Outdated
| from sage.matroids.basis_matroid cimport BasisMatroid | ||
| from sage.matroids.linear_matroid cimport LinearMatroid, RegularMatroid, BinaryMatroid, TernaryMatroid, QuaternaryMatroid | ||
| from sage.matroids.circuit_closures_matroid cimport CircuitClosuresMatroid |
There was a problem hiding this comment.
Let's not needlessly change the order of things, especially to something just as random.
Co-authored-by: Travis Scrimshaw <clfrngrown@aol.com>
|
Thanks for the detailed feedback. I was planning to further enrich the examples after the merging of this, and after editing some database matroids to be circuits matroids (the plan is to make the doctests much faster via a few strategic recasting of matroids into |
gmou3
force-pushed
the
circuits_matroid
branch
from
2179c8a to
f38dce7
Compare
Change 'order' to 'ordering'; use 'ordering' for consistency with matroid.p*. Add nonspanning_circuits_iterator() and improve docstrings.
gmou3
force-pushed
the
circuits_matroid
branch
from
f38dce7 to
3767c65
Compare
More examples and other mostly minor improvements Docstring improvements: 1-line references and outputs, 95 column limit in examples
gmou3
force-pushed
the
circuits_matroid
branch
from
48e45e5 to
d564669
Compare
|
I think I found a solution for which we can both be happy (easily readable and avoids checks and indents). As for the references, by having them in one line we save a lot of whitespace. But I don't mind much either way; let me know if you want me to revert this. |
|
A few blanklines isn’t really a big deal, and the previous way was more readable with how it is formatted IMO. |
There was a problem hiding this comment.
Thank you. The only thing left is the order change in unpickling.pyx. I think you wanted alphabetical, but it is not. I also am necessarily not in favor of these input sorting-type changes; yet, I won't reject it here (provided it is resorted by some reasonable way).
|
Documentation preview for this PR (built with commit e3b6119; changes) is ready! 🎉 |
I strategically recast a few matroids in `database_matroids.py` to `CircuitsMatroid`s (given sagemath#37148). On my old laptop, this results in a `x5` speedup in the testing of the file, and in a `x2` speedup in the testing of the whole of the `matroids` module (saving `~1m`). URL: sagemath#37338 Reported by: gmou3 Reviewer(s): Travis Scrimshaw
This matroid subclass allows the definition and internal representation of a matroid based on its flats (closed sets). This representation can be advantageous for some algorithms. This completes the list of the most standard matroid definitions (bases, circuits, flats, rank function). This PR is similar to sagemath#37148. - [x] The title is concise, informative, and self-explanatory. - [x] The description explains in detail what this PR is about. - [x] I have linked a relevant issue or discussion. - [x] I have created tests covering the changes. - [x] I have updated the documentation accordingly. URL: sagemath#37340 Reported by: gmou3 Reviewer(s): gmou3, Matthias Köppe
Added a subclass
CircuitsMatroid(of the mainMatroidclass). This enables the storage of a matroid based on its list of circuits, the verification using the circuit axioms, etc. The circuits description can be very succinct and it can facilitate efficient implementations of algorithms (one such example is the aforementioned verification).Up till now, when a matroid was defined via the keyword
circuits-Matroid(circuits=[...])- it was converted to aBasisMatroid. This required the computation of the bases upon definition which could be very slow.For some relevant discussion, have a look at #36820 and #36956. This PR is my proposed resolution of #36820 (Note that this provides a workaround and not a direct implementation of the issue's request. In fact, I deem that a direct implementation is undesirable due to memory considerations; see #36956.)
📝 Checklist