Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.



Re: Help with a factorisation, Please [Important]
Posted:
Jan 24, 2014 10:50 AM


On 01/18/2014 04:13 PM, Port563 wrote: > "David Bernier" <david250@videotron.ca> wrote in message > news:lbem7l$2nk$1@dontemail.me... >> On 01/18/2014 01:27 PM, Port563 wrote: >>> Hlauk wrote: > >>> The fact remains that "your" composites, if entered into the Alpertron >>> applet (after their consitutent prime pair has been multiplied out, so >>> what I enter into the applet does _not_ have a "*" in it) are very easy >>> to factorise, while the original composite chosen by David Bernier is >>> not. >>> >>> Try it for yourself, please. In this order! >>> >>> Choose one of your pairs yourself. >>> >>> Multiply them out using something other than the Alpertron applet. >>> >>> Put the product into http://www.alpertron.com.ar/ECM.HTM and, having >>> switched >>> the number of threads to the maximum for your CPU(s), factorise it. >>> >>> Depending on the power of your PC, it will do it taking from under 1 >>> second >>> to about 15 seconds. >>> >>> Now take the 130 digit composite supplied by David. Use the applet to >>> factorise >>> it. >>> >>> Go and make some coffee. >>> >>> Have a walk. >>> >>> Do some gardening. >>> >>> Then even bake a cake? If your PC is old, you will have time. >>> >>> I think you will then see my point. >>> >>> If they appear to be of a different type from the POV of the applet, then >>> they _are_ of a different type. >>> >>> It follows that either there is something unusual about your prime pairs, >>> or >>> about David's composite. >> >> I have a large collection of pseudorandom bytes, going back >> to a video capture (on a few occasions) of "snow" or >> "white noise" on a CRT tube. I mixed the bytes using recommended >> procedures/heuristics. A 130 digits modulus was chosen, >> as I had previously factored RSA129 about two years >> ago with CADONFS 1.0 . >> >> Following Pubkeybreaker's ECM and quadratic sieve threat >> discussion, I decided on two 65digit primes. >> I extracted digits from a base64 encoding of "random bytes" >> by hand, and finally obtained two large enough 65digit >> pseudorandom numbers m1 and m2 >> (meaning that the product was 130 digits). >> >> Using PARI/gp, I tested for primality the next few thousand >> odd numbers after the pseudorandom 'm1' and 'm2'. >> >> This gave the primes p > m1 , and q > m2, with >> p*q having 130 digits. >> >> So, p and q are "nothingupmysleeve" 65digit primes. >> I guess they were selected through partly noisedependent >> procedures, anyway I didn't interfere to cause bias. >> >> There are, from what I recall, >> weak rsa moduli (unrelated to NSA work on promoting >> "fake" rng generators). I didn't test for >> "weak rsa moduli" in the original, preNSArevelations, >> sense. > > > > I was sure there was nothing peculiar about your 130 bit "nearly prime", > David. > > The time it takes to be broken is what I would expect for a nearly prime of > that size whose prime factors are of approximately the same magnitude. > > It is Dan's pairs that are peculiar! > > > > > > > > >
My CADONFS 2.0 job with Perl script broke down with an error. So, now I come clean with the truth:
69176818182079702962583905305944091851396378426614231238000704281*68643130470496608225965397098921704892910199186048409746525386899 47485133560063190433548259679204355907477810307482663990525310023918\ 21388435849831808696657092215642611599144927466443608010614619
quit
dave
 http://www.bibliotecapleyades.net/sociopolitica/last_circle/1.htm



