A* test fails

[gonum]

A* test fails

[UserAB1236872]

Since it's getting the right cost, I wonder if it's simply writing the
predecessors incorrectly.

On Tue, Feb 4, 2014 at 3:41 PM, Sonia Keys [email protected] wrote:

New A* test against Dijkstra shows incorrect result. See commit 0c533c00c533c0a76b7baebe984c59093bd1a9beaaf22deon branch tests.

(Correct) path by Dijkstra is [5 4 2] with path length 21. Path returned
by A* is [5 6 3 4 2] with path length 21.

Note [5 6 3 2] is also length 21. That would be a valid answer showing the
test should be improved, but [5 6 3 4 2] is incorrect.

Reply to this email directly or view it on GitHubhttps://github.com//issues/3
.

[UserAB1236872]

I was doing some weird stuff in one of the steps in A*, I fixed it, but now it gets the wrong right path. ([2,3,6,5] instead of [2,4,5])

[UserAB1236872]

I confirmed that for both the path start=2;goal=5 and the reciprocal path start=5;goal=2 we now get a valid path. I'll leave this open until the test itself is fixed.

[soniakeys]

I changed test data so there are not alternate paths with the same cost. Tests pass nicely now.