mersenneforum.org  

Go Back   mersenneforum.org > Extra Stuff > Miscellaneous Math

Reply
 
Thread Tools
Old 2011-07-12, 16:01   #1
JohnFullspeed
 
May 2011
France

101000012 Posts
Default New Computation

for the english versioon I use Google Traductioon

I'm not sure that is a good idea English is less good than mine( It's ppossible!)

I join the Frrench version perhas somr one translate.
Thanks for those wicht do it..
Merci a ceux qui le font

Traduction (français > anglais)
Method John Fullspeed
The method of the mask


The principle is to calculate a mask that sulfide move the selected area

Creating the mask

Information is a list with all the NP
2 are removed multiple 5,7,11,13,17,23

The mask
put the mask from 0 or a multiple of the size of the mask
PRTI on you so determined enleber the muliple Odd ComenC in a 29
After you get the NP on a correspondanr interval to the size of the mask


Information for this method is very much faster than the best possible version of Erastothene for two reasons:


1-It removes only the odd multiples so we calculate half

2 Erasthothene working on 100% of the values of interval I only l!
30%



Quote:

Méthode John Fullspeed
La methode du masque

Le principe est de calculer un masque,qu'il sulfura de déplacer sur la zone choisie

Création du masque
1 On renseigne une liste avec tous les NP
2 On retire les multiple de 5,7,11,13,17,23

Utilisation du masque
poser le masque depuis 0 ou un multiple de la taille du masque
Sur la partie ainsi déterminée vous enleber les multiples impairs en comencant a 29
A l'issue vous obtenez les NP sur un interval correspondanr à la taille du masque

Pour information cette méthode est beaucoup plu rapide que le meilleure version possible d' Erastothene et cela pour deux raisons:


1-On ne supprime que les multiples impairs donc on calcule deux fois moins

2 Erasthothene travaille sur 100% des valeurs de l!interval moi seulement sur
30%
JohnFullspeed is offline   Reply With Quote
Old 2011-07-12, 17:25   #2
firejuggler
 
firejuggler's Avatar
 
Apr 2010
Over the rainbow

247710 Posts
Default

Quote:
Mask's method

The method is to calculate a mask , wich you can apply to a number range.

1)One populate the mask with all the prime
2) One remove the multiple of 5,7,11,13,17,23 ( note : he forgot 3 and 19)

Use of the mask
Put the mask at 0 to a multiple of the size of the mask.
On that range, One remove odd multiple of the mask population , starting with 29.
Once you are done, you have all the prime numberon that interval.

FYI, this method is far quicker than the most efficient implementation of Erasthomene sieve for 2 reason :

-One remove only the odd multiple so, one compute 2 time less
-While erastomene work on all the interval value, my method work on 1 third of the value
Ok so :
first flaw :you say the first step is to 'populate' the mask with all the prime number .... but how? since your objective is to determie them?

second : you forgot 3 and 19 in your listed prime

third :it can only work up to 292, after that you may misssome, unless you continue to 'populate' your mask after that

4 : you will remove more ' candidate' if you start with the lowest prime


and I think this method is called the ' wheel'

------------------------
premier probleme: La premiere etape de ta methode consiste a remplir le 'masque' avec TOUT les nombres premier... comment, puisque le but de ton implementation est de les determiner?

second probleme : tu as oublié 3, et 19 dans ta liste de NP

Troisieme probleme : a moi de continuer a nourrir ton 'masque' avec de nouveau nombre, ça ne marchera que jusqu'a 292

Quatrieme chose : Tu retirera plus de candidat si tu commence par le plus petit des nombres premier.


Et je crois que cette methode existe deja, sous le nom de ' roue'
firejuggler is offline   Reply With Quote
Old 2011-07-12, 18:02   #3
ldesnogu
 
ldesnogu's Avatar
 
Jan 2008
France

3×179 Posts
Default

You are reinventing wheel sieving

Faster sieves go one step further: you don't need to store multiples of 2 (which you already know), 3, 5 and so on. For instance, if you remove multiples of 2 and 3, you only need to take care of numbers that have the form 6n+1 or 6n+5. This means that in a single byte you can store 6x4 candidates.

Last fiddled with by ldesnogu on 2011-07-12 at 18:29
ldesnogu is offline   Reply With Quote
Old 2011-07-12, 18:17   #4
firejuggler
 
firejuggler's Avatar
 
Apr 2010
Over the rainbow

2,477 Posts
Default

6n+2? it will be always even...
5n+2 or 7n+2 might be more interesting ;p

Last fiddled with by firejuggler on 2011-07-12 at 18:17
firejuggler is offline   Reply With Quote
Old 2011-07-12, 18:30   #5
ldesnogu
 
ldesnogu's Avatar
 
Jan 2008
France

3·179 Posts
Default

Oops, fixed it
ldesnogu is offline   Reply With Quote
Old 2011-07-12, 19:37   #6
JohnFullspeed
 
May 2011
France

7·23 Posts
Default

Quote:
Originally Posted by firejuggler View Post
Ok so :
first flaw :you say the first step is to 'populate' the mask with all the prime number .... but how? since your objective is to determie them?

second : you forgot 3 and 19 in your listed prime

third :it can only work up to 292, after that you may misssome, unless you continue to 'populate' your mask after that

4 : you will remove more ' candidate' if you start with the lowest prime


and I think this method is called the ' wheel'

------------------------
premier probleme: La premiere etape de ta methode consiste a remplir le 'masque' avec TOUT les nombres premier... comment, puisque le but de ton implementation est de les determiner?

second probleme : tu as oublié 3, et 19 dans ta liste de NP

Troisieme probleme : a moi de continuer a nourrir ton 'masque' avec de nouveau nombre, ça ne marchera que jusqu'a 292

Quatrieme chose : Tu retirera plus de candidat si tu commence par le plus petit des nombres premier.


Et je crois que cette methode existe deja, sous le nom de ' roue'


Merci de repondre en français After to try Google traduction

I make the answer in englis mais merci beaucoup

Il n'est pas difficile de faire la liste des nombres premiers: se qui est difficile cc'est de faire une liste avec SEULEMENT les nombres premier

Personeally I use the fact that every prime number is 6k+1
OU 6k-1 So making the liste ot these value you have All the primes

Sample

i:=5
j:=7
repat
Ptime(i):=1;
Prime(j):=1;
I:=I+6
J:=J+6
Until J> Wanted


second probleme : tu as oublié 3, et 19 dans ta liste de NP
true for 19 but 3 is not use:
Using 6 for reason in my list I remove directly 2 and 3 multiple (2*3)


Troisieme probleme : a moi de continuer a nourrir ton 'masque' avec de nouveau nombre, ça ne marchera que jusqu'a 292

No limit for the mask
I use a mask of 223,092,870
you must begin at a multiple of the sizeof the mask

Sample
223092870*10^2000 or more
Just using the mask you eliminate 180,000,000
on the, 223092870 of the mask



Quatrieme chose : Tu retirera plus de candidat si tu commence par le plus petit des nombres premier.

They are not in my list because 6 power


This method is not 'the weel'
The goal of the weel is to fond value primes with a
base (the weel)


a weel 2-19 take n and try to divide n by 2,3,5,7,11,13,17,19
It don't use a mask; The weel is not a tools to make thelist of primes it's a tools to factored...
Wiky don't know weel sieve. Were find more information
The weel is not a sieve!!!!!!

The princip of the weel is not to make a mask and to
make the Prime list:Show us!!!
No trace on the forum. It's speeder tan Erastho?

What I think important is that I bud a set of value smaller then
the complet interval and I cn use a mask not o n the primes but on the composites You take the interval you want and with no code you reduce the
candidats

NO CODE SO NO TIME
NO LIMITS FOR PRIMES 100000 DIGITS I YOU WANT
the mask is universal!!!! you computte it 1 time and you can use it
allyour live

For the twins for example you can reduce the candidat to
900,000 for each set of 223 000 000and ypu have just todee if prime(i)+2 is primes

If we can use PrimeGrid we can compare with other search
JohnFullspeed is offline   Reply With Quote
Old 2011-07-13, 02:09   #7
Christenson
 
Christenson's Avatar
 
Dec 2010
Monticello

6B016 Posts
Default

Cher John:
Oubliez les traducions google ou votre traducions...le methode de firejuggler est mieux. Ecrivez seulement en Francais.
Christenson is offline   Reply With Quote
Old 2011-07-13, 05:33   #8
JohnFullspeed
 
May 2011
France

101000012 Posts
Default

Quote:
Originally Posted by Christenson View Post
Cher John:
Oubliez les traducions google ou votre traducions...le methode de firejuggler est mieux. Ecrivez seulement en Francais.
Quelle methode?
Il a juste dit


Quote:
6n+2? it will be always even...
5n+2 or 7n+2 might be more interesting ;p
Si vous vouez je peux vous fournnir une version plus rapide de la cretion du masque et une autre methode de filtrage sans les multiples(mais les;puissances)

Quells remarques sur la methdoe 6 k+-1?

John
JohnFullspeed is offline   Reply With Quote
Old 2011-07-13, 10:54   #9
Christenson
 
Christenson's Avatar
 
Dec 2010
Monticello

24×107 Posts
Default

Le methode de Firejuggler est a pense clairement sur les mots.....ca fait bien...
Christenson is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
A computation worthy of the name fivemack Factoring 121 2010-02-20 18:32
New Pi Computation Record ldesnogu Lounge 11 2010-01-07 14:42
Value of computation fivemack Lounge 0 2008-09-05 20:23
Saving computation in ECM dave_dm Factoring 8 2004-06-12 14:18
quantum particles (and computation)? ixfd64 Lounge 5 2003-10-17 21:18

All times are UTC. The time now is 03:29.

Tue Dec 1 03:29:29 UTC 2020 up 82 days, 40 mins, 1 user, load averages: 1.23, 1.50, 1.66

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.