View Single Post
Old 2007-12-16, 14:34   #7
monst's Avatar
Mar 2007

179 Posts

I took another approach that minimizes data entry into MultiSieve. I wrote a program that generates single-base GCW (Generalized Cullen/Woodall) starting files for MultiSieve. I make sure n and b are not both odd (since 2 is a factor for these numbers) and instruct MultiSieve to begin sieving at 3 via the header line...
ABC $a*$b^$a$c // CW Sieved to: 3

This allows me to launch MultiSieve with the "Continue Sieveing" selection. The only parameter I need to change from run to run is the "Save File Name". I do not change the log file name.

These files start off very large. After a few seconds MultiSieve eliminates alot of the candidates, reducing the file size a lot, but creates (or appends to) a large log file. I'm using a version of MultiSieve that has a "SieveUntil" feature implemented through the ini file. I set this just high enough to sieve past the Max k (max n in gcwsieve) so the file can then be run with the more efficient gcwsieve. I need to keep deleting the log file after every few MultiSieve runs since it gets very large. (I don't think I can suppress the logging in MultiSieve.)

So bottom line is this...
I am past the interim file manipulation request. I have better control with the method I'm using so I'm all set here. (Actually I had started writing a parser as you suggested when I decided on this other method.)

What would be most beneficial is having command line control of the entire process. This would require one of the following:

1.) Command line control of MultiSieve -- actually this would only be needed for "Continue Sieving" with "SieveUntil" control.


2.) Being able to start the sieve in gcwsieve

Last fiddled with by monst on 2007-12-16 at 14:35
monst is offline   Reply With Quote