Bug in is_hamiltonian: wrong exceptions are caught

[nathanncohen]

sage: g = graphs.CycleGraph(10)
sage: g.allow_loops(True)
sage: g.add_edge(0,0)
sage: g.is_hamiltonian()
False

This happens because in traveling_salesman_problem the call to _scream_is_not_simple raises a ValueError when the graph is not simple, and the same exception is raised when the graph is not hamiltonian.

We can use the EmptySetError to differentiate them, as we cannot optimize anything on an empty set.

Nathann

Component: graph theory

Author: Nathann Cohen

Branch/Commit: 0174081

Reviewer: Vincent Delecroix

Issue created by migration from https://trac.sagemath.org/ticket/16210

[nathanncohen]

Branch: u/ncohen/16210

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

18122c5

trac #16210: Bug in is_hamiltonian: wrong exceptions are caught

[sagetrac-git]

Commit: 18122c5

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

bf44a1c	trac #16210:: Merge into 6.2.rc0
d69303c	trac #16210: Broken doctest

[sagetrac-git]

Changed commit from 18122c5 to d69303c

[videlec]

Reviewer: Vincent Delecroix

[videlec]

comment:4

Hi Nathann,

Why not a NonImplementedError (with a shorter message)?

Vincent

[nathanncohen]

comment:5

Because I am not sure that _scream_if_not_simple should always raise a NotImplementedError in all functions that call it.

What is the problem with long messages ? They give more meaningful information.

Nathann

[videlec]

Changed branch from u/ncohen/16210 to public/16210

[videlec]

comment:6

Let us go for long messages...

I simplified (I hope) the implementation of the function _scream... and correct the doctest in is_hamiltonian. Set it to positive review if you like it...

[sagetrac-git]

Changed commit from d69303c to 0174081

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

0174081

simpler implementation + corrected doctest

[nathanncohen]

comment:8

AHahah. Well, it does more tests than in the first version, but I don't think this can ever become a very important problem :-P

Nathann

[vbraun]

Changed branch from public/16210 to 0174081