Closed maurges closed 1 year ago
I'm reviewing other proofs as well. I found that in $\Pi^{\text{mod}}$ you deviate from the CGGMP protocol by computing $xi = \sqrt[4]{y\{a_i, b_i}}$ instead of $xi = \sqrt[4]{y\{a_i, b_i}} \mod N$. See my comments in the specs.
Why not using BigNumber::inverse
? I.e.
let n_inverse = n.inverse(&phi).ok_or(Error::NotInversible)?;
Regarding sqrt: you are wrong, it is a modulo square root. But it seems I do deviate from the paper in another place, which you pointed out.
Ah yes, github code wrongly pointed out to definition of sqrt, it showed crate::common::sqrt. I'll use IDE instead :)
But it seems I do deviate from the paper in another place, which you pointed out.
Where?
where
The one with y_{a,b}. Actually since we then immediately do a modulo square root of that, I think this change I made is correct. Let me think for some time and I'll update the writeup
Yep, it seems to me I made a correct change with calculating y'. Updated the writeup.
Why not using BigNumber::inverse
Fixed. Why was that not a comment in code? =)
Fixed. Why was that not a comment in code? =)
I can't comment code that hasn't been modified :(
I've reviewed mod, and fac proofs (with specs), looks great! Just one suggestion: can we add a comment where untrivial algorithms like jacobi
and blum_sqrt
were taken from? Either in the code or in spec
I've reviewed mod, and fac proofs (with specs), looks great! Just one suggestion: can we add a comment where untrivial algorithms like jacobi and blum_sqrt were taken from? Either in the code or in spec
I'll make a issue out of that, PR can be merged without it
Created #20
Ok, I was just being slow =)
Changes Пfac and Пdec
Also updated documentation on other proofs to no longer mention l+1 ranges