v1.0.0

Changelog

• 716da62 Add --max-numerator to allow finding impure arctan terms
• 511b7b3 Add -calculate-norms to calculate norms on the fly
• 28b38d1 Add -evolve to evolve prime groups
• 0c6d27b Add -group and -groups-file flags
• c7f9e2f Add -group-extra-primes to add primes to a group
• a72aae1 Add -iteration-type flag with normal, random, weighted
• 4d15a38 Add -max-no-new flag to stop PSLQ if we haven't had a result for a while
• 7a1bf36 Add -norms-cache-dir to use prebuild cache of norms bitmaps
• 616cb9c Add -search-x flag to do a linear search over possible x values
• 0edbd60 Add -skip-relations flag to skip zeros and pi relations already found
• 7729489 Add -skip-zeros flag
• 18381b7 Add -sort-max-x flag to sort by max X value in evolution step
• 61c1449 Add -trim-start flag
• f2ef9a7 Add VCS version
• e981db4 Add factoring part to normscache
• a24dd48 Add missing -max-no-new to parameters output
• 788bda0 Add more debug info about PSLQ failures and trimming
• 1ce2d42 Add range flag type
• 664bedd Add regression tests
• 3bffbf3 Add score_arctan.py to process logs
• f835c44 Add start of README
• 71f85c8 Add top results for reference
• 5ee3729 Cache groups to speed up by a factor of 1.5x
• 882cc0d Convert to use 64 bit X values. Primes are still 32 bit.
• f327c73 Don't reject groups bigger than maxprime when using -calculate-norms
• 48f8dc3 Exit if no groups after filtering
• cb67fa3 Factor prime generation into its own package
• e3ec8b9 Find arctan formulae for Pi using factors of Gaussian norms.
• d24b216 Find the norms using the minimum of memory
• 99db1d7 Fix candidate cache misses
• 044936c Fix norm trimming as it was rejecting norms it shouldn't have done
• 43b7aef Fix stopping before all workers had finished
• 3e78041 Fix typo in log statement
• 3fd03af Flesh out README and add LICENSE file
• 0245c80 Lint and remove unused coded
• 59431ba Make candidate norms be plain 128 bit values to reduce memory usage hugely
• 08d709e Make isquare test configurable and tweak for speed
• b403bd2 Make sure norms are sorted after calculation before saving
• 6954779 Micro optimize findNorms for 3% gain
• 00c771d Optimise isSquare to use less memory to improve speed
• 9d5ee9b Optimise isqrt128 to double its speed
• 49f1562 Optimize memory usage of norm candidates
• 0b71c47 Put some stats in
• b0b1e40 Re-arrange searching to be more efficient with norm calculation
• 339c9d6 Record the prime set in the ArctanFormula and print it out
• 929dcd1 Reduce maxPrime to 100M to see if bitmaps are small enough
• abd3db7 Reduce memory use while calculating norms and increase caching
• 6f0d783 Release v1.0.0
• ee570b4 Remove debug which was causing GB of logs!
• 6f3a790 Remove failed norms cache experiment
• d120dec Rename to without 64 suffix as it was originally
• 998daf3 Return bits in opposite order and rework to use masks all the way through
• 06b9a51 Rework NormCombinations to return candidates with most first
• b1e1113 Rework NormCombinations to use bit masks.
• aec39e2 Seperate out -verbose and -pslq-verbose
• 04f33d9 Show the stats at the end of the computation
• 6d836d0 Sort the norms for consistency when running through them with -search-x
• 3cf79a1 Speed iqrt128 up by factor of 4 by using isqrt64 as initial guess
• 095ec5a Speed up Mul128x64 slightly
• c8d720e Speed up group filtering by 11x
• 9aaeccb Speed up group filtering by another 2x
• 57ce226 Speed up isqrt by 16% and add classic algorithms for comparison
• e225431 Tidy, refactor, add tests and add -max-lehmer-measure
• 6a2201e Update README
• d1e845a Update the default -alpha and -min-items to be more sensible
• 6815ad1 Use a bitset to keep track of the zeros for efficiency
• 912cef6 Use a cache for the norm calculation
• f9a1247 fixup! Add factoring part to normscache