Findafactor

Originally, smooth perfect squares were _forced_ on the left-hand side of the congruence-of-squares equation. This guaranteed that the left-hand side always formed a perfect square, but this was the _only_ modification that could potentially reduce the likelihood of finding other smooth numbers exactly as they fall on the sieving interval. At least for an A/B test, it's useful to have this release that does not force smooth perfect squares, until that attempted optimization is better tested.

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v5.0.4...v5.0.5

sha1sum results:
1295283d0cc99aaee1396a52b4bc3ae5470abb89 FindAFactor-5.0.5-cp310-cp310-manylinux_2_35_x86_64.whl
c9b4f286f6706e353faa66a6f6b5d3a40e5702bc FindAFactor-5.0.5-cp312-cp312-manylinux_2_39_x86_64.whl
094ccfca873e81b2216004f7124386aed28026c1 FindAFactor-5.0.5-cp312-cp312-win_amd64.whl
07c63668de42b62318f922a7bb09f5f3c92830dc FindAFactor-5.0.5-cp313-cp313-macosx_13_0_x86_64.whl
7a3fda1ee36c87336852be426aa94005c5a737c8 FindAFactor-5.0.5-cp313-cp313-macosx_14_0_arm64.whl
8b8d990d299d0ec4996f848bc5aa4c92bec5394b FindAFactor-5.0.5-cp313-cp313-macosx_15_0_arm64.whl
37caac5e3edb2ab63335fba8dc36ab737670868e FindAFactor-5.0.5-cp38-cp38-manylinux_2_31_x86_64.whl
da7e0bb33aed730e93ff18fa9a8d9f859b338674 findafactor-5.0.5.tar.gz

This releases introduces separate smooth wheel radii for the case of with and without excluded small primes from initial `wheel_factorization_level` handling in preamble phase. (I want to have the code in perfectly readable and optimized condition before investing the time to search for optimal settings, at least for the specific 64-bit semiprime example.)

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v5.0.3...v5.0.4

sha1sum results:
a2642939c291a8a6b08647d7d232c077a6e5e77c FindAFactor-5.0.4-cp310-cp310-manylinux_2_35_x86_64.whl
8f04acda72ca2d0dd3571e81003f17c04aba36d1 FindAFactor-5.0.4-cp312-cp312-manylinux_2_39_x86_64.whl
7cc28266d5379daaa09460deece6fb7a68b2ad72 FindAFactor-5.0.4-cp312-cp312-win_amd64.whl
1c82089265a624587230a42228d91bd8fb7236fc FindAFactor-5.0.4-cp313-cp313-macosx_13_0_x86_64.whl
cef540726ea2d7163a698cfdffb038384e8b5b0e FindAFactor-5.0.4-cp313-cp313-macosx_14_0_arm64.whl
1711cda18b2639412e13ffb74423caaf0dfdb35b FindAFactor-5.0.4-cp313-cp313-macosx_15_0_arm64.whl
286616bb2a8d11d401ec576aacb2229fa83c57e7 FindAFactor-5.0.4-cp38-cp38-manylinux_2_31_x86_64.whl
0d333764fb2e48c513df3efe04a5dc6f2fb37d6a findafactor-5.0.4.tar.gz

This patch mostly just removes various lines of code that are no longer necessary. (I still anticipate a v5.1.0 release, very soon, with defaults tuned for a 64-bit semiprime example.)

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v5.0.2...v5.0.3

sha1sum results:
85cf13fc8f7981867c6b348d22621f8fa93ccde3 FindAFactor-5.0.3-cp310-cp310-manylinux_2_35_x86_64.whl
3f261a5002fe44c66047dbe5c97199b211275d5a FindAFactor-5.0.3-cp312-cp312-manylinux_2_39_x86_64.whl
1cb088b2d535bad3f3ce62e2c0352bd1ad49560d FindAFactor-5.0.3-cp312-cp312-win_amd64.whl
4b4d1c2f3ed18b230dda7771c713cb93885e074b FindAFactor-5.0.3-cp313-cp313-macosx_13_0_x86_64.whl
ea1df139ab5cbe7339c779b163e1e2864b8d5fa0 FindAFactor-5.0.3-cp313-cp313-macosx_14_0_arm64.whl
2632268da7643c4efabace1050587e1431f46a43 FindAFactor-5.0.3-cp313-cp313-macosx_15_0_arm64.whl
8ddf84b0787331503153adb399802ffccc2ad6f0 FindAFactor-5.0.3-cp38-cp38-manylinux_2_31_x86_64.whl
b4c8b8f504067cc51345fef29336812688cffead findafactor-5.0.3.tar.gz

Pardon if there's a deeper theoretical prohibition I'm missing, but it doesn't seem to hurt anything to allow congruence-of-squares checks on all squares of single rows coming out of Gaussian elimination since these squares (at least trivially) result in fully-even parity in the residue factorization vectors.

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v5.0.1...v5.0.2

sha1sum results:
3f5ed62c505e729b12a83cf2bbbfc81976cc8525 FindAFactor-5.0.2-cp310-cp310-manylinux_2_35_x86_64.whl
42d5bf25fb1a7d52195c7ff48b77f891e7978ab5 FindAFactor-5.0.2-cp312-cp312-manylinux_2_39_x86_64.whl
f94903d9038aaeb2bfb5808c7481279bf792af57 FindAFactor-5.0.2-cp312-cp312-win_amd64.whl
bdb996b0bb10f645e069ce30afbe3e565c99801b FindAFactor-5.0.2-cp313-cp313-macosx_13_0_x86_64.whl
e2467742f194907ac9787338ad1d6f52fc11dc19 FindAFactor-5.0.2-cp313-cp313-macosx_14_0_arm64.whl
5dd9e40d2c4d0541fa8906a421754cfaf408397a FindAFactor-5.0.2-cp313-cp313-macosx_15_0_arm64.whl
f6924f55180359d757db428bf1cc73d2c0b8d0a4 FindAFactor-5.0.2-cp38-cp38-manylinux_2_31_x86_64.whl
4ddabc9d2967a442f93643becaff92e6f7347d74 findafactor-5.0.2.tar.gz

This release (only) fixes the early termination condition on all sieving threads, when enough smooth residues are found. (Expect another likely release today for tuned default settings.)

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v5.0.0...v5.0.1

sha1sum results:
bfec3c48c16863440818f7d370bdbb4f725e7ba3 FindAFactor-5.0.1-cp310-cp310-manylinux_2_35_x86_64.whl
3107099d87b5f9a8c0615555461b84c1640863ea FindAFactor-5.0.1-cp312-cp312-manylinux_2_39_x86_64.whl
1a90b492afbd809d3fa27539f8b8d4bf80f5aed1 FindAFactor-5.0.1-cp312-cp312-win_amd64.whl
6acb321f87c8c308f614109a5822c91a26f285de FindAFactor-5.0.1-cp313-cp313-macosx_13_0_x86_64.whl
357c53e691bb85a1ba1e5a800158c3d97b902cc6 FindAFactor-5.0.1-cp313-cp313-macosx_14_0_arm64.whl
48f3171e071a9fc7152aab3de872e3f22744358d FindAFactor-5.0.1-cp313-cp313-macosx_15_0_arm64.whl
92e32dc4d8bd9a102a0b2b4ca0a76bfdd1455f9a FindAFactor-5.0.1-cp38-cp38-manylinux_2_31_x86_64.whl
d81d6a9af6305e11c10b6399f965a6798d08814f findafactor-5.0.1.tar.gz

Let's actually read the textbook! `FACTOR_FINDER` should now stay true to the theory of Quadratic Sieve (QS). As explained in the README, people familiar with QS will recognize settings for smoothness and sieving bounds, but we put particular care into designing an interface where the "multiplier" settings should all typically be close to `1.0`. (The devil is in the tiny bit of tuning needed close to `1.0` values.)

If necessary, expect a quick update, even later to today, for any bug fixes or tweaks. **This release is experimental.** However, given our usual cadence of releases, there's no reason not to publish an **experimental** release right now.

What's Changed
* Proper Quadratic Sieve by WrathfulSpatula in https://github.com/vm6502q/FindAFactor/pull/5


**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v4.8.0...v5.0.0

sha1sum results:
3f554ad039f76e49c2530ffd7aabbfed1012d31a FindAFactor-5.0.0-cp310-cp310-manylinux_2_35_x86_64.whl
75d4646509360f630c569c6d8b0af389cc38168c FindAFactor-5.0.0-cp312-cp312-manylinux_2_39_x86_64.whl
2f8ed6f4dc80302ddd5d4c8ae22dd8ed943db151 FindAFactor-5.0.0-cp312-cp312-win_amd64.whl
99df4c7ed37b29fcc32061947437872244f82369 FindAFactor-5.0.0-cp313-cp313-macosx_13_0_x86_64.whl
dc7e80e3f8d8f85de001703f296024dc43142fd0 FindAFactor-5.0.0-cp313-cp313-macosx_14_0_arm64.whl
99dfffd5758e88a6045c1c6849028870edf320e4 FindAFactor-5.0.0-cp313-cp313-macosx_15_0_arm64.whl
9ee5ddee7157aa63d60f5970e4edf895e053ec84 FindAFactor-5.0.0-cp38-cp38-manylinux_2_31_x86_64.whl
ba3e71714f91224eb62511151efeec23ba862c41 findafactor-5.0.0.tar.gz