sci different implementations of the floating-point LLL reduction algorithm