Re: SF: Back to theory
From: Tim Peters (tim.one_at_comcast.net)
Date: 02/28/05
- Next message: examachine_at_gmail.com: "Re: Existence of mathematical entities (Re: Successor Axiom: on what grounds TF?)"
- Previous message: John Wilkins: "Re: an true information theory"
- Maybe in reply to: David C. Ullrich: "Re: SF: Back to theory"
- Next in thread: Tim Peters: "Re: SF: Back to theory"
- Reply: Tim Peters: "Re: SF: Back to theory"
- Messages sorted by: [ date ] [ thread ]
Date: Mon, 28 Feb 2005 18:00:49 -0500
[Tim Peters]
[...]
> f and g are generated exactly the same way in the algorithms that try
>
> gcd(f-g, M)
>
> yet those do better than random when splitting T^6. For some still-
> unknown reason, this way of generating f and g is better than random at
> finding f and g in the same equivalence class mod p or mod q, and despite
> that (as above) using f or g on its own is a provably miserable way to
> look for a factor.
Ah, I suspect this is going to be funny in the end <wink>. More later,
after a long test run ends. Suffice it for now to say that this no longer
looks surprising, it doesn't even look interesting -- and that there are
more effective ways to generate winning f and g without bothering to factor
anything.
- Next message: examachine_at_gmail.com: "Re: Existence of mathematical entities (Re: Successor Axiom: on what grounds TF?)"
- Previous message: John Wilkins: "Re: an true information theory"
- Maybe in reply to: David C. Ullrich: "Re: SF: Back to theory"
- Next in thread: Tim Peters: "Re: SF: Back to theory"
- Reply: Tim Peters: "Re: SF: Back to theory"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
