This release adds the rather technical command line option --lcs (or -b), which internally switches to Booth’s LCS algorithm when establishing wether a sequence is its own minimal rotation. This does not change the actual output of the program in any way.
This seems to be slower or equivalent to the default (naive) algorithm for any number of vertices (up to 255) on our test machine, even though we only tested it by generating a small subset of functional digraphs with a “large” number of vertices, of course. We include this option anyway because:
- it might actually work faster on your machine (let us know if it is the case!);
- this implements the actual
$O(n^3)$ -delay algorithm described in the paper, rather than a$O(n^4)$ one, which might be useful for research-related reasons.