My two cents: if `dseupd_` gives random output for the same input then it is most likely a bug in ARPACK, although it would be good to have an isolated test case that does not involve igraph. @jgmbenoit have you managed to produce such a test case, or do you need a hand with creating one? If you have submitted a bug report to ARPACK, can you add the link here so I can keep an eye on it?
Re the `--without-scg` option, @szhorvat noted that the bug does not arise in the spectral coarse graining routines (which could easily be made optional) but in eigenvector calculations (which are pretty central to igraph) so there is probably not much point in adding `--without-scg`. -- You are receiving this because you were mentioned. Reply to this email directly or view it on GitHub: https://github.com/igraph/igraph/issues/1469#issuecomment-770267692
