July 7, 2026
Kepler? I barely knew her
Pure-Python symbolic regression that rediscovered Kepler's law from 8 data point
This tiny Python tool found a planet rule fast — and the comments came for blood
TLDR: GP_ELITE is a simple Python tool that claims it can uncover human-readable math rules from small datasets, even recreating a classic law of planetary motion from just eight examples. Commenters were split between impressed and skeptical, arguing over whether this is real scientific insight or just a neat curve-fitting party trick.
A new tool called GP_ELITE is being pitched as the easy button for finding simple math rules in messy real-world data. No giant setup, no special hardware, just a normal install and go. Its big flex? Feed it the distances and years of the eight planets, and it spits back Kepler’s famous rule for how planets move around the Sun. For fans, that’s a mic-drop moment: a small, readable answer instead of a mysterious machine guess.
But the real action was in the comments, where the applause instantly turned into a science-side-eye festival. One commenter started calmly poking holes: what if you add more planetary info, use only the six planets Kepler actually knew, or throw in noisy data instead of the clean classroom version? Another went straight for the jugular on the battery demo, calling the formula suspicious because it mixes things that maybe shouldn’t be mixed at all. In plain English: yes, it may fit the numbers, but does it mean anything in the real world?
And then came the big dreamer energy: if this thing can rediscover a planet rule, could it tackle galaxy mysteries better than today’s theories? That got balanced by a wonderfully dismissive hot take saying, basically, “Calm down, you can eyeball a power law on a graphing site.” So the vibe is clear: half the crowd sees a delightfully accessible discovery machine, and the other half is yelling, “Cool trick, now prove it’s not just fancy curve-fitting.”
Key Points
- •GP_ELITE is a pure-Python symbolic regression package for discovering interpretable equations from small experimental datasets.
- •The article says the tool rediscovered Kepler’s Third Law from eight planetary data points in seconds.
- •GP_ELITE can be used via a console workflow for CSV files or programmatically through a Python API that returns expressions, validation scores, and model size.
- •In a NASA battery example, the article reports a discovered state-of-health law with validation R² of 0.996 from 168 charge cycles.
- •The article reports benchmark results on 16 Feynman physics equations, with 81% solved at R² above 0.999 and mean R² of 0.993 in fast mode.