Re: Surrogate factoring, reasons for my concerns

From: Marcel Martin (mm_at_ellipsa.no.sp.am.net)
Date: 07/15/04

  • Next message: cryptokid: "Re: Required properties for encryption algorithms." 
    Date: Thu, 15 Jul 2004 12:11:09 +0200

    Jeroen Boschma a écrit :
    >
    > James Harris wrote:
    > > > function [k_num, k_denom] = get_k(j, N)
    > > >
    > > > k_num = (-j*N + N^2*sqrt(j^2 - N^2 + 1));
    > > > k_denom = (j^2 - N^2);
    > > >
    > >
    > > You forgot that it's "+/-" not just "+".
    > >
    > > You need to check both possibilities wherever there's a square root
    > > taken.
    > >
    > > In a way it's a minor mistake, but I just kind of find it freaking
    > > annoying.
    >
    > There's possibly a good explanation why Claude only left the plus-sign (I had his mail on the screen
    > when I typed in the code). I did not search for it, but a small code-change is quickly made:
    >
    > * f_1 f_2 = (T^2-1)/4
    > * plus and minus sign in k-calculation
    >
    > There were no result changes (0 percent factored for primes up to 1e4).

    Yes, gcd(x - yN, N) = gcd(x + yN, N) but since this is not obvious 'for
    the eye', JSH believes this is different. 

    -- 
mm
http://www.ellipsa.net/
mm@ellipsa.no.sp.am.net  ( suppress no.sp.am. )
  • Next message: cryptokid: "Re: Required properties for encryption algorithms."

    Relevant Pages
    • Quantcast