Does anyone else think that the fact that the ase vasp calculator just passes the cell vectors to vasp, without reordering them if the triple scalar product is negative, is a bug? VASP certainly refuses to run with such a set of lattice vectors. Noam