Open-ended cryptanalysis of the DES cipher motivated by answering the question:
Suppose one has the output (32 bits), and the subkey (48 bits) for the DES Feistel function.
Can one reconstruct uniquely the 32 bits of input?
des_feistel_backwards.py demonstrates that the answer depends on which output is known and is probabalistic. We found that many outputs have colliding inputs which cause the same output. Can this be exploited?