[seqfan] Re: Purely algorithmic number sequence identification

[Philipp Emanuel Weidmann]

In the past two weeks, Sequencer has seen some big upgrades. It now has
a parallel search engine which automatically makes use of all available
CPU cores and a "smart" formula generator that avoids building redundant
expressions like "Abs(Abs(x))" etc., as well as some other performance
improvements. As a result, search is now around 30 times faster on a
typical quadcore machine, with depth 6 searches completing in seconds
(and depth 6 in particular now being available on the web interface
http://www.sequenceboss.org/).

Also, number theoretic and additional combinatorial primitives are now
available: Prime numbers, values of Euler's totient function, Bernoulli
and Catalan numbers and double factorials can now be found.

I believe the system is now ready to be included in Superseeker. To go
with Charles' suggestion of just using the program as a "black box", the
command line

  java -jar sequencer.jar -o 1 2 3 4 5 > tempfile

will result in a minimal (progress messages suppressed) tempfile that
could just be appended to Superseeker's output. This is probably the
easiest way to go about integrating the two systems for now.

Best regards
Philipp




On Tue, 2015-02-24 at 08:17 -0500, Neil Sloane wrote:
> What do you think of adding your program to Superseeker?  It sounds like
> this would definitely be worth doing.
> 
> By the way, can you do anything with A122536?  We have 200 terms, but no
> formula or recurrence!
> 
> Best regards
> Neil
> 
> Neil J. A. Sloane, President, OEIS Foundation.
> 11 South Adelaide Avenue, Highland Park, NJ 08904, USA.
> Also Visiting Scientist, Math. Dept., Rutgers University, Piscataway, NJ.
> Phone: 732 828 6098; home page: http://NeilSloane.com
> Email: njasloane at gmail.com
>