Code for the algorithms that recover seeds for the PCG pseudo-random number 
generator.

Written by Julia Sauvage ([email protected]) and
Charles Bouillaguet ([email protected]).

This code is in the public domain.


challenges/ folder: tools to generate challenge inputs and expected output.


Cknown/ folder: case where the default increment is used. 

main.c: Multi-thread program that recovers the seed from 3 consecutive outputs. 
        Takes about 25 CPU-minutes. A simple procedure is repeated 2**37 times.



Cunknown/ folder: case where an unknown increment is used (default in numpy). 

main.c: MPI program that recovers a partial difference between two states from
        10 consecutive outputs. Takes about 12500 CPU-hours. Uses MPI, so is
        parallelizable on a cluster. A simple procedure is repeated 2**51 times.

step2.py: given 64 consecutive outputs and the result from the previous program,
          finds the full difference between any two consecutive states. Very
          fast.

step3.py: given 64 consecutive outputs and the result from the two previous 
          programs, prints the seeds. Very fast.



Other folders: journal article, slides, random junk.