Closed lhq1 closed 4 months ago
1) There is no reference for Tungsten. It is experimental, use at your own risk, I make no promises that its secure. Maybe at some point I will be able to prove something but so far I have not been successful.
2) Here are some references https://eprint.iacr.org/2022/1014 https://eprint.iacr.org/2022/712
Thanks, ill take a look at the config
fixed
Some codes are used to accelerate matrix multiplication in dual LPN assumption, such as QC code, EA code, ExConv code et al. I have several questions:
For Tungsten code, I still don't know its definition and construction after searching for papers about PCG. Could you please provide some references? I also wonder about its features compared to ExConv code.
For the weight of noise vectors, I note there exists a lower bound: when length N is less than 512, it is 64; otherwise, it is 40. In [BCGI18], there doesn't exist such a lower bound. Could you please explain more? BCGI18: Compressing vector OLE.
By the way, I found a bug: In syndromeDecodingConfigure, there should be
break
after line 133, otherwise, all configure of EA code will fail.