test: add tests for as_adjacency_matrix() in test-conversion

[maelle]

@szhorvat this is to prep for part of #1518, prior to my PR there were no direct tests of as_adjacency_matrix().

[aviator-app[bot]]

Current Aviator status

Aviator will automatically update this comment as the status of the PR changes. Comment /aviator refresh to force Aviator to re-examine your PR (or learn about other /aviator commands).

This PR was merged manually (without Aviator). Merging manually can negatively impact the performance of the queue. Consider using Aviator next time.

---

See the real-time status of this PR on the Aviator webapp.

Use the Aviator Chrome Extension to see the status of your PR within GitHub.

[szhorvat]

You are using randomly generated data, then testing for specific values that depend on the random seed. This is difficult to manage and it's hard to see what goes wrong with failures.

I suggest:

• Create a small specific non-random graph, which has self-loops and multi-edges. 3 or 4 vertices should suffice.
• Create non-random edge weights. I suggest that these should be simple non-integers rounded to a single digit after the point (e.g. 1.2, 0.1, 2.0, etc.), for readability.
• Test the whole matrix for exact equality. Do not just test specific elements. Do not test row or column sums for a specific value. Once the loops argument is sorted out, we can test that row/column sums equal the out/in degree (unweighted) and strength (weighted). That is not true right now, so don't test it yet.
• Ideally, test this both for a directed and undirected instance.

This would make a complete test.

It is important to test self-loops and multi-edges, which is where things go wrong!

[szhorvat]

For example,

dg <- make_graph(c(1,2, 2,1, 2,2, 3,3, 3,3, 3,4, 4,2, 4,2, 4,2), directed=T)
ug <- as.undirected(g, mode="each")

The directed version contains: multi-edges, reciprocal edges, self-loops, multi-self-loops. The undirected version has the same edges, with directions ignored. I chose a 4-vertex graph (instead of a 3-vertex one) to leave room for non-diagonal zeros in the undirected case, and better cover all cases.

Then just add edge weights.

[maelle]

Thank you!!

[maelle]

@szhorvat I've implemented the changes, is this good to go?

[szhorvat]

Based on a quick skim, yes!

[aviator-app[bot]]

This pull request failed to merge: PR cannot be automatically rebased, please rebase manually to continue. After you have resolved the problem, you should remove the blocked pull request label from this PR and then try to re-queue the PR. Note that the pull request will be automatically re-queued if it has the mergequeue label.

Additional debug info: Failed to rebase this PR onto the latest changes from the base branch. You will probably need to rebase this PR manually and resolve conflicts).