Recently I have encountered the following paradox (in my mind at least).
I tried to evaluate the rank of a symbolic matrix in Matlab (not really sure if rank() works as intended in the symbolic case, but anyway), but in order to simplify the coefficients resulting from the calculations, I used vpa() in the matrix.
What I noticed is that, after using vpa command, the matrix suddenly became a full rank matrix.
Has anyone had any previous experience with something similar to what I described? Which one should I trust?