GMP division plans

delta trinity deltatrinity at
Sun Apr 2 03:24:40 CEST 2006

With the computer speed increasing, trial-error is becoming faster every 
days.  At this point, wouldn't it be faster to try a number, verify if it's 
the right one, and substract the error (times some factor), to converge 
toward the solution?

Functions could be added for each of those steps, along with details on how 
to use them in the manual.


You got me for a minute... :-)


>From: Torbjorn Granlund <tg at>
>To: gmp-discuss at
>Subject: GMP division plans
>Date: 01 Apr 2006 18:01:29 +0200
>Now that GMP 4.2 is finally finished and releases, the GMP developers are
>concentrating on GMP 5 and GMP 6.
>Every major GMP release has had a new low-level division interface, and the
>division interface of the previous release has been declared obsolete (but
>the old interface is kept for compatibility reasons).
>GMP 5 will be no exception to this tradition; its division interface will
>not look like anything previously seen.
>For GMP 6, we intend to drop division from GMP altogether, urging users to
>use other operations instead to complete their computations.  We believe
>this will be a significant improvement, since truncating division is not
>only slow, it is an algebraically unpleasant operation.
>But before we start removing the files, we would like to solicit the
>feedback from GMP users.  What do you think, should we drop division from
>GMP 6, or should we invent another creative division interface for that
>release?  Do you have any suggestions for how such an interface should
>Since compatibility is important, we will of course keep the deleted
>division functions even for GMP 6 and GMP 7.
>gmp-discuss mailing list
>gmp-discuss at

More information about the gmp-discuss mailing list