Back trace (slightly cleaned up):
#0 0x0000000000403407 in internal::real_default_impl<double, false>::run (x=@0x0)
#1 0x000000000040222a in numext::real<double> (x=@0x0)
#2 0x000000000040160b in internal::ldlt_inplace<1>::unblocked<MatrixXd, Transpositions<-1, -1, int>, Matrix<double, -1, 1, 0, -1, 1> > (mat=..., transpositions=..., temp=..., sign=@0x7fffffffd818)
#3 0x00000000004012db in LDLT<MatrixXd, 1>::compute (this=0x7fffffffd7e0, a=...)
#4 0x0000000000400fcb in LDLT<MatrixXd, 1>::LDLT (this=0x7fffffffd7e0, matrix=...)
#5 0x00000000004009de in main () at forum.cc:9
Fix for this was very simple. However, I had to change the value of rcond for 0x0 matrices (from 1 to inf) to avoid introducing a lot of extra logic into the unit test. At least octave agrees on that (I don't have a running Matlab version at the moment). I hope this does not break anything.
Pushed to devel and 3.3:
-- GitLab Migration Automatic Message --
This bug has been migrated to gitlab.com's GitLab instance and has been closed from further activity.
You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.com/libeigen/eigen/issues/785.