sprps

Collection of strong pseudoprimes to small prime bases

This packet contains the strong pseudoprimes to, at the time of writing, the prime bases 2 to 2293 that are smaller than 2^32 (4,294,967,296).

The raw data is in the folder raw with one strong pseudoprime per line with a Unix line-end (ASCII \n 0x0a).

The factored data is in the folder factored with one strong pseudoprime per line with a Unix line-end (ASCII \n 0x0a), followed by a colon (ASCII : 0x3a) followed by one space (ASCII 0x20) followed by the space separated factors.

There are some tables with strong pseudoprimes to the composite bases 4 to 100 in the folder raw_composite and their factors in factored_composite.

There are Fermat pseudoprimes to base 3 (three) in the folder misc up to 2^40 (1,099,511,627,776) together with their factoring. Because it is quite easy to get the strong pseudoprimes to base 3 (three) from that data the strong pseudoprimes to base 3 (three) together with their factoring is also in the folder misc.

There are several files with strong pseudoprimes to several bases ((((((2,3,5),7),11),13),17),19) and (2, 3, 5, 7, 61) up to 2^64 in the folder misc.

There is a script compute_3psp_parallel.sh to run the 3-psp search in parallel in the folder scripts. Currently base 3 is hardcoded but can be changed manually in the script.

What is not in the repository?

Large files:

• A file with the modular order of three for all primes below 2^32 is in an Google drive
• Combinations of three of the current (may or may not been updated) bases (gzipped). Google drive
• Combinations of three of the current (may or may not been updated) bases (gzipped), prime-bases only. Google drive

How to get the whole repository as a compressed packet

wget https://github.com/czurnieden/sprps/archive/master.tar.gz

For the branch master. For the other branches yet to come replace master.tar.gz with other_branch.tar.gz.

Why?

It took over 150 CPU hours to generate that data and I hate to waste that.

License

The license here is "unlicense" but the raw data itself, being just data, is Public Domain, of course. Yes, there is a difference.

Is there more to come?

Probably.

There are currently beavering away:

• PRP of base 3 up to 2^64 (which means SPRP to base 3, too)
• SPRP to more prime bases (planned is up to base 10007)
• SPRP to more composite bases (planned is up to base 10008)

Planned:

• extend 2-SPRP >2^64 (first one >2^64 is, believe it or not: 1 + 2^64)
• extend 7-SPRP and 61-SPRP >2^32 but only a bit to check distribution

Help needed?

If you have some CPU cycles to spare or already computed some tables/data not in this collection: you are heartily welcome!

Where are the sources?

All tables have been generated with Dana Jacobsen's ntheory. The relevant snippets are listed at "Pseudoprime Statistics, Tables, and Data" at the end of the page.

For values above 2^64 the relevant functions from the big integer library LibTomMath have been used. Check directory src for some code.

The factors were generated with factor(1) by Paul Rubin, Torbjorn Granlund, and Niels Moller which can be found in the GNU coreutils.

How long do you think MS will accept such large amount of non-code?

Get it while it's hot!