Wrong definition of cofactor

The implementation of the cofactor, in ufl/algorithms/expand_compounds.py, line 125, is misleading, since the returned value is not the cofactor matrix, but the adjoint matrix, which is the transpose of the former.

The issue can be easily addressed: for instance, the code

return as_matrix([[A[1,1], -A[0,1]], [-A[1,0], A[0,0]]])

becomes

return as_matrix([[A[1,1], -A[1,0]], [-A[0,1], A[0,0]]])

and so on for n=3 and n=4.

Then, also the inverse matrix must be fixed accordingly (line 168):

return self.cofactor(None, A) / self.determinant(None, A)

to

return self.transposed(None, self.cofactor(None, A)) / self.determinant(None, A)

An alternative could be to introduce a new operator ``adjoint'', but the name might be too generic in dolfin.

Comments (3)