Sierpinski base 4

[jasong]

dat file completed(because I have insomnia)

1,183 KB

How shall we transfer this? Email?

[jasong]

reserved 1-2 billion(just to make sure it works. :) )

Something seems to be wrong with the dat file, I've tried it on two computers, both complain. :(

[geoff]

Originally Posted by jasong

Something seems to be wrong with the dat file, I've tried it on two computers, both complain. :(

If you are using a .dat file with sr2sieve then it must be in base 2, and you must use the -s command line switch to tell sr2sieve that it contains Sierpinski numbers. If it is named anything other than SoB.dat then you must also use the -i switch to specify the file name.

If you want to sieve in base 4 then you can use an ABCD format file. Don't use the -s switch for ABCD files (or anything other than .dat files). I don't think there will be any difference in sieve speed whichever base you use.

The srfile program from the srsieve archive can be use to convert NewPGen files into ABCD format. If your NewPGen files are called 1.txt, 2.txt, etc. then run `srfile -a 1.txt 2.txt' to create an ABCD file.

You can also use srfile to remove factors from the ABCD file. If your factors are in factors.txt and your ABCD file is called sr2data.txt, then run `srfile -k factors.txt -a sr2data.txt' to create a new ABCD file with the factors deleted.

To convert from ABCD file to individual NewPGen files (for PRP testing) run `srfile -g sr2data.txt'.

One thing to remember: sr2sieve only works when the factors are greater than the greatest k value. You need to use srsieve or NewPGen to do the initial sieving. In practice it may be faster to use srsieve to sieve right up to p=2^32 before starting sr2sieve.

[jasong]

/me takes notes. :)

stay tuned.

[jasong]

It complains about the fact that the first n-value of the second file is lower than the last value of the first file. Btw, I used the -a option the second time around, thinking that would correct the problem. I'll continue to use the -a option in subsequent tries, even though I have no idea what it does.

I'm going to do some reading of some files on my computer to see if a "down-and-dirty" approach would work, namely chaining the files together with a few edits. Wish me luck.

-a means make an abcd file.

[jasong]

I'm going to use NewPGen to sieve each file past 2^32. I'm thinking that might've been why sr2sieve complained.

[jasong]

Apparently, I've been wasting everyone's time the past few days, because I have not a clue what has gone wrong. here's the beginning of the file I generated with the NewPGen files:

4
1000000
5000000
k=64494
1000177
+12
+72
+228
+84
+24
+60
+72
+84
+36
+264
+24
+144
+12
+84
+96
+48
+108

For some reason sr2sieve(the intel version) gives me "Invalid or unrecognized file sr2data.txt."

Any ideas?

[Citrix]

The order of the numbers is wrong. The k should be in ascending order.

[jasong]

Okay, I'll try again.

Edit: Result: Invalid or unrecognized file sr2data.txt

Here's my build file:

1000000
5000000
sr2data.txt
S18534.npg
S23451.npg
S60849.npg
S64494.npg

Does anyone know if anything needs to be done to it?

[thommy]

Did you try "srfile -a S18534.npg S23451.npg S60849.npg S64494.npg" to create a ABCD-file? You get it from inside an archive here : http://geocities.com/g_w_reynolds/srsieve/

[jasong]

Originally Posted by thommy

Did you try "srfile -a S18534.npg S23451.npg S60849.npg S64494.npg" to create a ABCD-file? You get it from inside an archive here : http://geocities.com/g_w_reynolds/srsieve/

Finally, SUCCESS!!! So srfile is the key, and those other files were just a waste of time.

Every one who wants the file, PM me your email address, the file's about 1.2-1.3MB, so not THAT big. At the moment, I only have one email address to send to. Also, when reserving, the minimum range to start is 6(meaning 6,000,000,000).

Let's get this show on the road.

Edit: email factor files to jasong at conwaycorp dot net (name them by their starting range when emailing)