Findafactor

Latest version: v6.7.0

Safety actively analyzes 714736 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

Page 10 of 26

4.7.0

If the `FACTOR_FINDER` "ladder" (of smooth square primes) process relaxes like a Markov mixing process, one could drive against relaxation by adding several multiples of a single square prime component at once (through random selection and multiplication).

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v4.6.0...v4.7.0

sha1sum results:
9b5f8075b20570bc41fbdf3e7939bfa93ae9c6cb FindAFactor-4.7.0-cp310-cp310-manylinux_2_35_x86_64.whl
32d153e66802de5e30d07f1fffa3a8b025a2bf55 FindAFactor-4.7.0-cp312-cp312-manylinux_2_39_x86_64.whl
d007c1b0659e80fdcff4565ac5449c2384147e76 FindAFactor-4.7.0-cp312-cp312-win_amd64.whl
bece63f1edf829ef96a557f24ffe79b09d6316a5 FindAFactor-4.7.0-cp313-cp313-macosx_13_0_x86_64.whl
5ad6b17805adaa24ebbe60a07a4d8cae9b541d63 FindAFactor-4.7.0-cp313-cp313-macosx_14_0_arm64.whl
e9ccd3f47581425f2aff4530d5cd71b67d7b6f6f FindAFactor-4.7.0-cp313-cp313-macosx_15_0_arm64.whl
f9a95c696859f135acd8a56cef3b0677480759cd FindAFactor-4.7.0-cp38-cp38-manylinux_2_31_x86_64.whl
ae7f99624a566b3de5f4b45ce4a525f75ab0b741 findafactor-4.7.0.tar.gz

4.6.0

Elara (GPT) has suggested that the "ladder" algorithm in `FACTOR_FACTOR` method is similar to a Markov mixing process, which has a relaxation time proportional or similar to **`O(k log k)`** for `k` count of distinct elements to mix. As such, `batch_size_multiplier` is now a multiplier on `k log k`, where `k` is the count of smooth primes. Mixing operation count, for comparison, is tallied as half the sum of multiplication and division operations on a full ascend/descend cycle of the "ladder."

However, even after tuning the default batch size multiplier, there is still significant variance in what the best batch size could be for any random semiprime, even if several semiprimes are almost exactly the same size. Users could tune the batch size up or down, but there's no feedback on the effect of this tuning until success. Hence, this motivates the new `batch_size_variance` argument: it is an integer `m` for which the batch size `b` is automatically varied as `b * (2 ** i)` for `i` in `range(m)`, repeating from the beginning at the end of the sequence. This allows users to target a contiguous range of batch sizes at once, where the motivation for power-of-two increments come from what the developers have typically used to successfully identify bespoke batch sizes for different numbers to factor.

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v4.5.0...v4.6.0

sha1sum results:
e2cc3d52bd6cd8d9294947c236185ef56f8215d4 FindAFactor-4.6.0-cp310-cp310-manylinux_2_35_x86_64.whl
67b4d5c908fffdfe4276a2fe117908ac96dfcc4c FindAFactor-4.6.0-cp312-cp312-manylinux_2_39_x86_64.whl
a3591866d4d63e5474dc013bd729451566b4e773 FindAFactor-4.6.0-cp312-cp312-win_amd64.whl
85606af43f97d530d37c2b111c1ca8740088cfbd FindAFactor-4.6.0-cp313-cp313-macosx_13_0_x86_64.whl
7716e8f5472f25beeebdda2c22d24fd8f287c389 FindAFactor-4.6.0-cp313-cp313-macosx_14_0_arm64.whl
f31f41350a013e68f4a42fd6d4c1f1cda7014a04 FindAFactor-4.6.0-cp313-cp313-macosx_15_0_arm64.whl
b7648867bd9af87c222d840f3b551b1f47849e96 FindAFactor-4.6.0-cp38-cp38-manylinux_2_31_x86_64.whl
5f1e8d95f9a71ca4da3e2ad52040804e1792f13e findafactor-4.6.0.tar.gz

4.5.0

This release changes the `batch_size_multiplier` default. Note that even if two numbers are almost the same size, they might still respectively best benefit from significantly different `batch_size_multiplier` values. It also expands multi-node parallelism to `FACTOR_FINDER` mode. (For `FACTOR_FINDER` mode, users can even run multiple solvers with 1/0 settings for total/ID node values, without much or any redundancy, but assigning different IDs to different nodes will segment their smooth number seed range.)

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v4.4.0...v4.5.0

sha1sum results:
c23760215c3480809a641aa264e5e96a34d643ee FindAFactor-4.5.0-cp310-cp310-manylinux_2_35_x86_64.whl
d5e9f1aa4a776fcebdda0d412c9ca26c4375f721 FindAFactor-4.5.0-cp312-cp312-manylinux_2_39_x86_64.whl
5773177c0f8b724a4a811bbad36d877bde0b12c3 FindAFactor-4.5.0-cp312-cp312-win_amd64.whl
3a37e501a0725f7f64f58a91a21eba51fcf8efc0 FindAFactor-4.5.0-cp313-cp313-macosx_13_0_x86_64.whl
d95069723a02800730213ee2d77d8bb5a3da283b FindAFactor-4.5.0-cp313-cp313-macosx_14_0_arm64.whl
6dd1e2360b45beb81219754f0c41292eaed7f32b FindAFactor-4.5.0-cp313-cp313-macosx_15_0_arm64.whl
7c955739f2f619b387798263c97ed5c367fcce2b FindAFactor-4.5.0-cp38-cp38-manylinux_2_31_x86_64.whl
916ce57b17ccd0e5a4c714d7454d27d3471271c4 findafactor-4.5.0.tar.gz

4.4.0

MIXED` method had limited foreseeable application: if one wants prime proving capabilities, `PRIME_PROVER` seems to be faster in general, with no sign of any inversion at scale; meanwhile, `FACTOR_FINDER` seems even slower at small scales, but there's good theoretical reason to believe it scales better than `MIXED` mode (or at least should after refinements).

`MIXED` method has been cut. Any `method` argument values other than `PRIME_PROVER`/`0` will default to `FACTOR_FINDER`/`1`.

What's Changed
* Cut mixed mode by WrathfulSpatula in https://github.com/vm6502q/FindAFactor/pull/3


**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v4.3.1...v4.4.0


sha1sum results:
87348112240c70b922faf8e2d8c9933756da600e FindAFactor-4.4.0-cp310-cp310-manylinux_2_35_x86_64.whl
c8c6de306074315b1c71d23b82be53a8a1f19e53 FindAFactor-4.4.0-cp312-cp312-manylinux_2_39_x86_64.whl
e66dcc87d39411aaf1805d7d6737fa83440bd2a2 FindAFactor-4.4.0-cp312-cp312-win_amd64.whl
885519272187bf669f0a0f257d7f517b228fd97a FindAFactor-4.4.0-cp313-cp313-macosx_13_0_x86_64.whl
3b68a0ff0839a2312adf1d1750a4a86150203cf7 FindAFactor-4.4.0-cp313-cp313-macosx_14_0_arm64.whl
87340db084fcc5b198d97eafe81c938feb2ddf4f FindAFactor-4.4.0-cp313-cp313-macosx_15_0_arm64.whl
06ff802172d442916c3890603d3d262c53cb4cd4 FindAFactor-4.4.0-cp38-cp38-manylinux_2_31_x86_64.whl
8bd3e323e92cc007d161a6df0bf3964fa9ab3bb8 findafactor-4.4.0.tar.gz

4.3.1

It's actually quite important that seeding and reseeding are managed with insight such that the statistical distribution stays relatively close to uniform on each and every valid perfect square in the range `(N, N^2)`. This is improved in this release by using the initial seeding algorithm to produce a larger seed, starting above `N`.

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v4.3.0...v4.3.1

sha1sum results:
2ebd9501163e6bd0a5fa2f523e6e653e821e0429 FindAFactor-4.3.1-cp310-cp310-manylinux_2_35_x86_64.whl
b114283225426e0b442d25e5c000e7b41077f783 FindAFactor-4.3.1-cp312-cp312-manylinux_2_39_x86_64.whl
206d3416e04b3a1eb71a0cbec13dcf10948b581f FindAFactor-4.3.1-cp312-cp312-win_amd64.whl
a4c902aeddddd6ee2773b130d32b6d2a49d6ebcb FindAFactor-4.3.1-cp313-cp313-macosx_13_0_x86_64.whl
2e96e39d3fc945986048ad5aaf7722bd83012a6d FindAFactor-4.3.1-cp313-cp313-macosx_14_0_arm64.whl
717636c9969071b44e279575ada606e2ca9778f0 FindAFactor-4.3.1-cp313-cp313-macosx_15_0_arm64.whl
4e35fccf94e99fe608cdab042c54b480905b62f4 FindAFactor-4.3.1-cp38-cp38-manylinux_2_31_x86_64.whl
9bfa77614c090fec057c28bc4230357ce51a64b2 findafactor-4.3.1.tar.gz

4.3.0

A reseeding interval for the initial smooth number seed per thread likely benefits approach to uniform randomness of smooth perfect squares. This is controlled by `batch_size_multiplier` and proportional to the square of the count of smooth primes, for cycle count of up-and-down traversal. (Coincidentally, the existing default multiplier setting of `8` seems to remain reasonable).

**Full Changelog**: https://github.com/vm6502q/FindAFactor/compare/v4.2.7...v4.3.0

sha1sum results:
830ae722ddceee07c9d7c5058dd487e1b25e375d FindAFactor-4.3.0-cp310-cp310-manylinux_2_35_x86_64.whl
8d2df4edf6d51437bc8b0e9facc6025c17156279 FindAFactor-4.3.0-cp312-cp312-manylinux_2_39_x86_64.whl
d7d98baf4f8533a92409887176ebd01a815a97c2 FindAFactor-4.3.0-cp312-cp312-win_amd64.whl
26b7c673cb4f9f7cda89ac275b5ac1cdcc51e26d FindAFactor-4.3.0-cp313-cp313-macosx_13_0_x86_64.whl
01e2946b36a4e81b90b2d0b59b415695fe0783d8 FindAFactor-4.3.0-cp313-cp313-macosx_14_0_arm64.whl
fdb4f475dee73aea30f0ad83b50055dd5a50e888 FindAFactor-4.3.0-cp313-cp313-macosx_15_0_arm64.whl
fa257d2b8e78eb8a1c7a751c4c7e52452725ad9d FindAFactor-4.3.0-cp38-cp38-manylinux_2_31_x86_64.whl
d39c6ab55f197786700c82137fb91f9d1f9dbf15 findafactor-4.3.0.tar.gz

Page 10 of 26

© 2025 Safety CLI Cybersecurity Inc. All Rights Reserved.