I was testing on 2 signatures. both created with different K nonce.
as we all know, it will not lead to the correct private key unless the same nonce is reuse. However, i extended the code to include Babai's Nearest plane algorithm to readjust the wrong private key and get it closer to the correct private key by "guiding it " to the correct pubkey coordinates and it manage to only give me the partial private key from the closest vector.
I think this is interesting and will be exploring further to recover full private key.
result from my test.
root@MSI:/home/krash/test# python3 test.py
INFO:__main__:Using 2 signatures to build the lattice...
INFO:__main__:Short vectors after LLL reduction:
INFO:__main__:Short vector: (166728803922956164814674931204777127007, 0, -269917836034390020152096309044001076683, 0)
INFO:__main__:Short vector: (284880382552528911560114733869411675383, 0, 233299777452147397360970206987837337582, 0)
INFO:__main__:Short vector: (0, 115792089237316195423570985008687907853269984665640564039457584007908834671663, 0, 0)
INFO:__main__:Short vector: (191322878670180847261800180723017804593, 0, -63475476205101254537288001668546129622, 115792089237316195423570985008687907853269984665640564039457584007913129639936)
INFO:__main__:Completed LLL reduction.
INFO:__main__:partial private key from Babai: 000000000000000000000000000000007d6eccb306788626b5f389b33f092c5f
At this point, the experiment is not completed and i only consider something viable if we can recover the full private key. if anyone interested to work with me on how we can refine this experiment further, do DM me.
You need ask questions in crypto.stackexchangr I think.This will be more productive.