[j08ny]

Thanks, a paper is being prepared with the full details and an improved method. The sensitivity of the method to noise (one bad inequality in the lattice can make it not find the key) is really worth looking at, as that would improve the number of signatures necessary for the attack severely.