Ali Akhavi (GREYC, Caen)

Lattice reduction algorithms such as LLL and its floating-point variants have a very wide range of applications in computational mathematics and in computer science: polynomial factorization, cryptology, integer linear programming, etc. It can occur that the lattice to be reduced has a dimension which is small with respect to the dimension of the space in which it lies. This happens within LLL itself. We describe a randomized algorithm specifically designed for such rectangular matrices. It computes bases satisfying, with very high probability, properties similar to those returned by LLL. It significantly decreases the complexity dependence in the dimension of the embedding space. Our technique mainly consists in randomly projecting the lattice on a lower dimensional space, by using two different distributions of random matrices.