PNC k shortest simple path (for directed graphs) #40284

kappybar · 2025-06-22T13:48:17Z

Implement PNC k shortest simple path for directed graphs.
I implemented for only directed graphs for simplicity first.

📝 Checklist

The title is concise and informative.
The description explains in detail what this PR is about.
I have linked a relevant issue or discussion.
I have created tests covering the changes.
I have updated the documentation and checked the documentation preview.

⌛ Dependencies

#40248
#40217

dcoudert

Why is it necessary to define PathCandidate in a .h file ? can't it be done in file path_enumeration.pxd ?

src/sage/graphs/path_enumeration.pyx

kappybar · 2025-06-23T11:03:35Z

Why is it necessary to define PathCandidate in a .h file ? can't it be done in file path_enumeration.pxd ?

Because I want to use priority_queue of C++ standard library, the type of PathCandidate must be struct of C++. Thus, I think I need definition of PathCandidate in .h file. However, There may be another way of doing so without .h file. I am looking for it.

dcoudert · 2025-06-23T12:21:13Z

isn't it possible to use a tuple instead of PathCandidate ? or may be cdef priority_queue[pair[pair[double,bint], pair[int, int]]] heap_sorted_paths.
Priority should go to the path with smallest length and in case several paths have the same length, we want first the simple paths.

src/sage/graphs/path_enumeration.pyx

dcoudert

also, the key to the paper should be ACN2023 (Al Zoobi, Coudert, Nisse)

dcoudert · 2025-06-24T12:29:06Z

something goes wrong in the algorithm. Paths are not reported in increasing order of weight

sage: g = DiGraph(graphs.Grid2dGraph(2, 6).relabel(inplace=False))
sage: for u, v in g.edge_iterator(labels=False):
....:     g.set_edge_label(u, v, 1)
sage: %time X = [w for w, P in pnc_k_shortest_simple_paths(g, 5, 1, by_weight=True, report_weight=True, labels=True, report_edges=True)]
CPU times: user 1.39 ms, sys: 0 ns, total: 1.39 ms
Wall time: 1.4 ms
sage: print(X)
[4.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 8.0, 8.0, 8.0, 8.0, 8.0, 8.0, 8.0, 8.0, 8.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 10.0, 10.0, 10.0, 10.0, 8.0]

src/sage/graphs/path_enumeration.pyx

dcoudert · 2025-06-27T14:25:47Z

You should rebase the branch on the last beta.

If you do another commit, you can add the doi to the paper :doi:`10.1145/3626567`.

dcoudert · 2025-06-27T15:39:34Z

It's not fully fixed

sage: from sage.graphs.path_enumeration import yen_k_shortest_simple_paths as yen
sage: from sage.graphs.path_enumeration import feng_k_shortest_simple_paths as feng
sage: from sage.graphs.path_enumeration import pnc_k_shortest_simple_paths as pnc
sage: D = graphs.Grid2dGraph(5, 5).relabel(inplace=False).to_directed()
sage: %time len(list(yen(D, 0, 24)))
CPU times: user 14 s, sys: 408 µs, total: 14 s
Wall time: 14.1 s
8512
sage: %time len(list(feng(D, 0, 24)))
CPU times: user 1.23 s, sys: 988 µs, total: 1.23 s
Wall time: 1.24 s
8512
sage: %time len(list(pnc(D, 0, 24)))
CPU times: user 427 ms, sys: 995 µs, total: 428 ms
Wall time: 428 ms
8530

dcoudert

Something goes wrong. The method can yield non simple paths.

sage: from sage.graphs.path_enumeration import pnc_k_shortest_simple_paths as pnc
sage: D = graphs.Grid2dGraph(5, 5).relabel(inplace=False).to_directed()
sage: for P in pnc(D, 0, 24):
....:     if len(P) != len(set(P)):
....:         print(P)
....:         break
....: 
[0, 5, 10, 15, 16, 17, 18, 13, 12, 11, 6, 1, 2, 7, 8, 13, 14, 19, 24]

vertex 13 is repeated.

dcoudert · 2025-06-28T09:42:50Z

src/sage/graphs/path_enumeration.pyx

+      a path is returned. Otherwise a tuple of path length and path is
+      returned.
+
+    OUTPUT: iterator


this line is not useful. It can be removed.

dcoudert · 2025-06-28T09:44:28Z

src/sage/graphs/path_enumeration.pyx

+
+    ALGORITHM:
+
+    This algorithm is based on the `feng_k_shortest_simple_paths` algorithm


use the form the ``feng_k_shortest_simple_paths`` . A single ` is for math mode.

dcoudert · 2025-06-28T09:45:05Z

src/sage/graphs/path_enumeration.pyx

+        [4.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 6.0, 8.0, 8.0,
+         8.0, 8.0, 8.0, 8.0, 8.0, 8.0, 8.0, 10.0, 10.0, 10.0, 10.0]
+
+    Same tests as yen_k_shortest_simple_paths::


Here as well, use ``yen_k_shortest_simple_paths``::

dcoudert · 2025-06-28T09:45:42Z

src/sage/graphs/path_enumeration.pyx

+
+    Same tests as yen_k_shortest_simple_paths::
+
+        sage: g = DiGraph([(1, 2, 20), (1, 3, 10), (1, 4, 30), (2, 5, 20), (3, 5, 10), (4, 5, 30)])


this first graph is not used.

dcoudert

Let's try to make the doc build. It's currently failing.

dcoudert · 2025-06-28T15:29:19Z

src/sage/graphs/path_enumeration.pyx

+    deviations occur. See Postponed Node Classification algorithm in [ACN2023]_
+    for the algorithm description.
+
+    EXAMPLES:


for the EXAMPLES and TESTS blocks, use EXAMPLES:: and TESTS::, unless it is followed by sub blocks, which is not the case here.

dcoudert · 2025-06-28T15:30:04Z

src/sage/graphs/path_enumeration.pyx

+    G.delete_edges(G.incoming_edges(source, labels=False))
+    G.delete_edges(G.outgoing_edges(target, labels=False))
+
+    by_weight, weight_function = self._get_weight_function(by_weight=by_weight,


Here you can use G. _get_weight_function.

dcoudert · 2025-06-28T15:32:02Z

src/sage/graphs/path_enumeration.pyx

+                # target vertex is not reachable from v
+                return -1
+            if v in ancestor_idx_dict:
+                if ancestor_idx_dict[v] <= t or ancestor_idx_dict[v] == len(path) - 1:


len(path) - 1 could be a parameter of the method. I know it's a minor optimization.

dcoudert · 2025-06-28T15:35:37Z

src/sage/graphs/path_enumeration.pyx

+        if is_simple:
+            # output
+            if report_edges and labels:
+                P = [edge_labels[e] for e in zip(path[:-1], path[1:])]


FYI, you could simply use zip(path, path[1:]). the zip iterator ends as soon as one of the list is exhausted. No need to change.

github-actions · 2025-06-29T02:49:20Z

Documentation preview for this PR (built with commit 211ae0c; changes) is ready! 🎉
This preview will update shortly after each push to this PR.

dcoudert

The following improvements can be done. Let me know if you agree.

vertices in unnecessary_vertices could be removed from G, and so not considered when calling shortest_path_func
method ancestor_idx_func could be defined outside the while loop. It suffices to add it parameter ancestor_idx_dict
dictionary ancestor_idx_dict should be initialized only if is_simple is True.

Finally, this new method can be exposed in shortest_simple_paths and become the default for directed graphs. I'm not sure it's faster for small digraphs than Yen, but for large ones, it's clearly faster.

If you prefer, we can do these changes in a follow up PR.

kappybar · 2025-06-29T13:29:47Z

Yes I agree with you. This implementation can be improved in these ways or more. I will make another follow up PR.
Thank you for your valuable comments to this PR!

dcoudert

LGTM.

sagemathgh-40364: Improve PNC algorithm       Improve the implementation of PNC algorithm. See comments in sagemath#40284. ### 📝 Checklist  - [x] The title is concise and informative. - [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 and checked the documentation preview. ### ⌛ Dependencies    sagemath#40284 URL: sagemath#40364 Reported by: Yuta Inoue Reviewer(s): David Coudert

github-actions bot added the s: needs review label Jun 22, 2025

dcoudert added c: graph theory gsoc: 2025 Tag for GSoC2025 issues/PRs labels Jun 23, 2025

dcoudert reviewed Jun 23, 2025

View reviewed changes

src/sage/graphs/path_enumeration.pyx Outdated Show resolved Hide resolved

src/sage/graphs/path_enumeration.pyx Outdated Show resolved Hide resolved

src/sage/graphs/path_enumeration.pyx Outdated Show resolved Hide resolved

src/sage/graphs/path_enumeration.pyx Outdated Show resolved Hide resolved

dcoudert suggested changes Jun 24, 2025

View reviewed changes

src/sage/graphs/path_enumeration.pyx Outdated Show resolved Hide resolved

src/sage/graphs/path_enumeration.pyx Outdated Show resolved Hide resolved

dcoudert reviewed Jun 24, 2025

View reviewed changes

dcoudert reviewed Jun 27, 2025

View reviewed changes

src/sage/graphs/path_enumeration.pyx Outdated Show resolved Hide resolved

kappybar added 6 commits June 28, 2025 17:04

pnc k shortest simple path for directed graphs

795d9b2

add path_enumeration.pxd in meson.build

5649d6c

fix styles of pnc_k_shortest_simple_paths

704a10d

remove path_enumeration.{pxd, h}

62eb197

fix case that no path in G - path and add tests

230a94e

fix styles

974d049

kappybar force-pushed the pnc_algorithm branch from 4f66628 to 974d049 Compare June 28, 2025 08:12

add doi

20141b7

dcoudert suggested changes Jun 28, 2025

View reviewed changes

fix bug of getting non-simple path in pnc algorithm

52228c9

dcoudert suggested changes Jun 28, 2025

View reviewed changes

minor fix

ec9bc14

add function to csv-table

211ae0c

dcoudert reviewed Jun 29, 2025

View reviewed changes

dcoudert approved these changes Jun 29, 2025

View reviewed changes

dcoudert removed the s: needs review label Jun 29, 2025

dcoudert added the s: positive review label Jun 29, 2025

kappybar mentioned this pull request Jul 2, 2025

Improve PNC algorithm #40364

Merged

5 tasks

vbraun merged commit 1a1d0c4 into sagemath:develop Jul 6, 2025
26 checks passed

github-actions bot removed the s: positive review label Jul 6, 2025


		ALGORITHM:

		This algorithm is based on the `feng_k_shortest_simple_paths` algorithm


		Same tests as yen_k_shortest_simple_paths::

		sage: g = DiGraph([(1, 2, 20), (1, 3, 10), (1, 4, 30), (2, 5, 20), (3, 5, 10), (4, 5, 30)])

Uh oh!

PNC k shortest simple path (for directed graphs) #40284

PNC k shortest simple path (for directed graphs) #40284

Uh oh!

Conversation

kappybar commented Jun 22, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

📝 Checklist

⌛ Dependencies

Uh oh!

dcoudert left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

kappybar commented Jun 23, 2025

Uh oh!

dcoudert commented Jun 23, 2025

Uh oh!

Uh oh!

Uh oh!

dcoudert left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dcoudert commented Jun 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

dcoudert commented Jun 27, 2025

Uh oh!

dcoudert commented Jun 27, 2025

Uh oh!

dcoudert left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dcoudert left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

github-actions bot commented Jun 29, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

dcoudert left a comment

Choose a reason for hiding this comment

Uh oh!

kappybar commented Jun 29, 2025

Uh oh!

dcoudert left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

kappybar commented Jun 22, 2025 •

edited

Loading

dcoudert left a comment •

edited

Loading

dcoudert commented Jun 24, 2025 •

edited

Loading

github-actions bot commented Jun 29, 2025 •

edited

Loading