Findafactor

An option has been added that allows users to select the number up to which primes are sieved for trial division. If more primes are needed to choose the smooth factor base, simply set the trial division level higher. More importantly, a conceptual mistake has been fixed in the selection of the smooth prime factor base: primes should only be admitted into the base if the residue of the number to factor, modulo a given prime, is a perfect square.

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v3.5.1...v3.6.0

sha1sum results:
d779de5e108cc16f5fa38a6ddbea449207d9f0d4 FindAFactor-3.6.0-cp310-cp310-manylinux_2_35_x86_64.whl
8b5eeb741794d47836fc86fc109b74e5bc362447 FindAFactor-3.6.0-cp312-cp312-manylinux_2_39_x86_64.whl
e4286da3f12d264f3d8fdcb78e2056fd78c2bd52 FindAFactor-3.6.0-cp312-cp312-win_amd64.whl
ec54b8e1d9e3ad9ea850a0476fb0a1d62e31cabf FindAFactor-3.6.0-cp313-cp313-macosx_13_0_x86_64.whl
2a78fdd6c350277c4d063b02d72855527316fd4c FindAFactor-3.6.0-cp313-cp313-macosx_14_0_arm64.whl
f80a7ea157bc1d35665c806cb50fb484c231c93f FindAFactor-3.6.0-cp313-cp313-macosx_15_0_arm64.whl
83053749e7735f982973b1279da4b6ac6e11613d FindAFactor-3.6.0-cp38-cp38-manylinux_2_31_x86_64.whl
23a183b8b93c59eea61546fac2d72c7e1a2c29c5 findafactor-3.6.0.tar.gz

smoothNumberKeySet` served a function, in previous versions. By now, 99% of what it does is simply pose a liability to long-running large number factorization with congruence of squares, to run out of memory. As such, it should be removed, as in this release.

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v3.5.0...v3.5.1

sha1sum results:
ed86682b499c86b1c49cd5ec583c5e8fa998fc1e FindAFactor-3.5.1-cp310-cp310-manylinux_2_35_x86_64.whl
d4ed6043d1fdc05fea15a812a129ab263105a69f FindAFactor-3.5.1-cp312-cp312-manylinux_2_39_x86_64.whl
e45d82f6519a35176a6abd11a33bee64b0bc87be FindAFactor-3.5.1-cp312-cp312-win_amd64.whl
94ef421109ee7a34513f2aacc3b2694f1cdd455f FindAFactor-3.5.1-cp313-cp313-macosx_13_0_x86_64.whl
d3f1aae7a1d5bae7ac02d2c8e74b7ced95e3f262 FindAFactor-3.5.1-cp313-cp313-macosx_14_0_arm64.whl
9a32dc051c167c1d0ad04d6b6c47b101ff496398 FindAFactor-3.5.1-cp313-cp313-macosx_15_0_arm64.whl
360ae7a4300dec52d9527ea45dd6a5a367c4649e FindAFactor-3.5.1-cp38-cp38-manylinux_2_31_x86_64.whl
312eae7e8a463ae8f5652967ed2a123106e1e0cb findafactor-3.5.1.tar.gz

The default value for `batch_size_multiple` has been set to 64. For Gaussian elimination, this seems to give the best performance at small scales on a Core i9. (Please do not expect to rely on default settings. Scale of the number to factor and the systems on which it's factored make a qualitative difference in optimal settings. A `batch_size_multiple` of 64 "won't completely suck," in general, but it's necessary for users to tune this and the other options on a case-by-case basis.)

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v3.4.0...v3.5.0

sha1sum results:
0a29bb0fa3617d4b4781d0b3fc14daa44a6e5ff0 FindAFactor-3.5.0-cp310-cp310-manylinux_2_35_x86_64.whl
3d6a355f29e531f36b9dfbbe89dd408919a304d7 FindAFactor-3.5.0-cp312-cp312-manylinux_2_39_x86_64.whl
d16e045f439e465eb094357544c18c0e12567cdc FindAFactor-3.5.0-cp312-cp312-win_amd64.whl
a6653a488af88d6024b855f716668b0c536156a1 FindAFactor-3.5.0-cp313-cp313-macosx_13_0_x86_64.whl
82df67fdbc4584b8aa31b72f5e60d5d00a3c23bd FindAFactor-3.5.0-cp313-cp313-macosx_14_0_arm64.whl
5403e5b2c75c7fc53cba99b5edc9752a0b4f367e FindAFactor-3.5.0-cp313-cp313-macosx_15_0_arm64.whl
4b7fe7a356afbd404e0f8bed5097627cc189b00e FindAFactor-3.5.0-cp38-cp38-manylinux_2_31_x86_64.whl
c31b3f79c87aa5ff9cd4e75cd410612f3b2e26e1 findafactor-3.5.0.tar.gz

use_gaussian_elimination=True` switches Gaussian (O(N^3) for rows worst case) elimination techniques on, with the `False` option of only a check of all rows for duplicate factorization parity, only once per pair of rows (for O(N^2) for rows worst case, but with _many_ more rows, larger "N").

What's Changed
* Gaussian elimination by WrathfulSpatula in https://github.com/vm6502q/FindAFactor/pull/1

New Contributors
* WrathfulSpatula made their first contribution in https://github.com/vm6502q/FindAFactor/pull/1

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v3.3.2...v3.4.0
5b813241a030dc689ff918898e329e0b288a33db FindAFactor-3.4.0-cp310-cp310-manylinux_2_35_x86_64.whl
64aef1d03449d0772b931d5174117921727bb413 FindAFactor-3.4.0-cp312-cp312-manylinux_2_39_x86_64.whl
639fa9f21cbd20b8f8e3b1c70a5bdd8f823fb6b3 FindAFactor-3.4.0-cp312-cp312-win_amd64.whl
983d6a02b2b02b7c1310cb6a170428308ca6cb3c FindAFactor-3.4.0-cp313-cp313-macosx_13_0_x86_64.whl
6f7257ab586f80311b0a51068754ca97b52ff698 FindAFactor-3.4.0-cp313-cp313-macosx_14_0_arm64.whl
2a9b27a89b24f4e083f63806d62157ecbd862a67 FindAFactor-3.4.0-cp313-cp313-macosx_15_0_arm64.whl
cb90438d28c78dc0d91c403c09826afeba2188a3 FindAFactor-3.4.0-cp38-cp38-manylinux_2_31_x86_64.whl
9c530f9e3d877c2799222ac7cf875abef08443ca findafactor-3.4.0.tar.gz
sha1sum results:

A tiny bit of additional work can be reused, in the loop to combine rows into perfect squares.

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v.3.3.1...v3.3.2

sha1sum results:
f3f146f3ca420cf65a584bfac7d97fc220e95b4b FindAFactor-3.3.2-cp310-cp310-manylinux_2_35_x86_64.whl
ef0a382dbd8dc0b773d4280807fd12016c0e3d86 FindAFactor-3.3.2-cp312-cp312-manylinux_2_39_x86_64.whl
42d6c725ef1ffac2677091a01e83af5b0fe1c196 FindAFactor-3.3.2-cp312-cp312-win_amd64.whl
42957b3328fc7c5bacaeec7e3fd70fde2fde1520 FindAFactor-3.3.2-cp313-cp313-macosx_13_0_x86_64.whl
5f796b6e2e696209085d9f634598c032cec9e5f7 FindAFactor-3.3.2-cp313-cp313-macosx_14_0_arm64.whl
fcda888434e1c52ae78557589a55bd8fa606cf06 FindAFactor-3.3.2-cp313-cp313-macosx_15_0_arm64.whl
3583ea4b215cdbc4f530d3030071c61062e795de FindAFactor-3.3.2-cp38-cp38-manylinux_2_31_x86_64.whl
724033b28f351463fcccfd7e7443393199561c9f findafactor-3.3.2.tar.gz

rowOffset` was a good idea, but it wasn't implemented correctly before this release iteration. The `batch_size_multiple` default has also been re-tuned, but, in general, users should tune `batch_size_multiple` for the specific system and number to factor, every time.

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v3.2.4...v3.3.0

sha1sum results:
4518406eb8967cf1be29dd4a568dd2a97d9d2fc9 FindAFactor-3.3.0-cp310-cp310-manylinux_2_35_x86_64.whl
5b13a376385f2ec428a91f2b5d1df60a8cda2823 FindAFactor-3.3.0-cp312-cp312-manylinux_2_39_x86_64.whl
bf9b98e3714be44708012c1d043762e2e8e14a99 FindAFactor-3.3.0-cp312-cp312-win_amd64.whl
8100e4d0ade5acf7bb3b006bad0b52df11dafd97 FindAFactor-3.3.0-cp313-cp313-macosx_13_0_x86_64.whl
a5138410ec0a43a2e944132835429a462f2d8c06 FindAFactor-3.3.0-cp313-cp313-macosx_14_0_arm64.whl
d976e598c5fd56a0fbad3e936c03ca79ad338c77 FindAFactor-3.3.0-cp313-cp313-macosx_15_0_arm64.whl
bd92f2b473babfc1eab34f2b36ba49823ba7964e FindAFactor-3.3.0-cp38-cp38-manylinux_2_31_x86_64.whl
d6d556b3c90a6cab4bec5aa3a8a135592d41ea93 findafactor-3.3.0.tar.gz