Search for BKZ 2.0 lattice-reduction strategies using fplll and fplll.

The strategizer uses git submodules to fix specific revisions of fplll and fpylll to use. Hence, make sure to clone this repository recursively by calling

$ git clone --recursive git@github.com:fplll/strategizer.git

If you have previously cloned without the --recursive switch, run git submodule init and git submodule update.

We assume you are using a virtualenv for isolating Python build environments. Then, to install fplll and fpylll run

$ (strategizer) ./setup.sh

Then run

$ (strategizer) export LD_LIBRARY_PATH="$VIRTUAL_ENV/lib"

to allow Python to find fplll.

We need to estimate how many nodes can be enumerated in a second, for this run:

$ (strategizer) python ./set_mdc.py

To run the strategy search, first check the help

$ (strategizer) python ./strategize.py --help

and then run e.g.

$ (strategizer) python ./strategize.py --max-block-size 40 --threads 2 --samples 128

To compare two different strategies, try:

$ (strategizer) python ./compare.py A.json B.json

This software is written by:

• Martin R. Albrecht
• Léo Ducas
• Marc Stevens

and licensed under the GPLv2+.

The best way to contact the authors is by contacting the fplll-devel mailinglist.