Mathworld, and ReverseSmarandache Primes

[axn]

Originally Posted by ericw

Indeed. Another question for Paul or anyone else (I am having trouble rejoining primeform on Yahoo groups to ask there.) I see PrimeGrid has been busily finding new Thabit ibn Kurrah primes 3*2^n - 1. I found some references to that search in various places:
<snip>
However, I did not manage to find a page listing the current status/search limits. Any know if there is one?

http://www.primegrid.com/stats_321_llr.php
This is the combined stats for + & - sides, but the active range is narrow enough that it should be good enough for your purpose.

Originally Posted by ericw

Also, may I infer from the above that Paul's 321 search

http://www.mersenneforum.org/321sear...%20search.html

is no longer active? Thanks.

Yes. With Paul U's blessings, Primegrid has been continuing that effort.

[paulunderwood]

For the list of Thabit ibn Kurrah primes see the wiki page

[ericw]

Originally Posted by pinhodecarlos

Please check this address:

http://www.primegrid.com/server_status_subprojects.php

Perfect. Thank you.

[Batalov]

Originally Posted by ericw

Yes, still searching. As of today the search has reached 47889 47888 ... 3 2 1 without finding any more PRPs.

I could rewrite the Sm sieve into a Smr sieve. For Sm, I've sieved as far as 3.6e11; for the 5-digit Smr's sieving would be faster and could definitely be done as deep as Sm and probably deeper. the good thing about Smr series is that it is significantly thicker (they all end with 1 which is already a huge difference from Sm; Sm has only 4/30 candidates surviving primes<=5, while Smr has 1/3)

[Batalov]

I had a bit of free time and rewrote the Sm() sieve for Smr().
I am sieving to p<2E10 for starters (for 47889<n<99999) and could PRP test the sieved candidates as well.

Can we divide the range? May I suggest that I
1) post the sieve file for (for 47889<n<60000)
2) test above 60000 ?

[Batalov]

There are no Smr(n) primes in 50000<=n<80000. Finishing with 80000<=n<99999 now.

[ericw]

Originally Posted by Batalov

There are no Smr(n) primes in 50000<=n<80000. Finishing with 80000<=n<99999 now.

Nice. I will update MathWorld. Do you care to polish off 48250 to 49999 while you're at it (since clearly your search will finish way before mine gets there).

[Batalov]

I can. There are only 53 candidates there, over which I jumped to save a margin. (There are 1452 candidates in 50k<=n<100k.)

I will sieve in 100000<=n<200000, too, and start testing sometime later (maybe via the PRPNet, say, port 1201 on my little toy PRPserver).

[rogue]

Don't forget to grab the latest pfgw source and PRPNet source. The biggest change is to eliminate the client spitting out the long decimal representation of the number being tested, which only happens on a generic server. The client right now generates the decimal representation so that it can compute the length of the number and sends it back to the server, which then updates the database. It isn't optimal, but it works. The alternative is to write server specific code for the Sm and Smr searches.

[Batalov]

Ok, thanks!

[Batalov]

Here are Smr (reverse concatenated sequence) results for this 5 decimal digit range.

6 decimal digit range will follow sometime later (need to sieve first). Perhaps 100000<=n<200000 for starters.