by Christoph. Differs from the one by Rudy in BFM since it vectorises the twisted 4D solves in pairs.