USDR Crosscheck · desktop
The check is whether that breaking point follows p_c(L) = p_c(∞) + c × L^(−1/ν), with ν within 15% of 4/3.
Fill a grid until one habitat wraps all the way around. That filled percent is the breaking point for that width. A huge landscape breaks near 59.3%. A smaller one should break somewhere else, and the gap should shrink in a known way as the width grows.
The conclusion appears here after the four widths are measured.
This run writes that sentence from the dots on the chart.
Press Run. One real landscape fills over about 15 seconds, then starts over. Each color is a label for one connected habitat, so you can tell them apart. The color is not height, density, or quality. Gold, at the end, is only the habitat that wrapped around the landscape. Dark cells are empty. This picture is not the average, and it is not a photograph of a real park.
Waiting.
Filled fraction. The gold tick is 59.27%.
Each dot is the average breaking point from 400 runs at that width. The number on the dot is the filled percent where the habitat wrapped. The gold dashed line is 59.3%, where a huge landscape breaks. When the run finishes, the teal line is the curve fitted to these dots.
Press Run. One landscape, 48 patches wide, fills over about 15 seconds and then repeats. Colors name connected habitats. They are not a measurement. The chart is 400 real runs at each size. The result is that fit, not a preset.
git clone https://github.com/KR8ZYSHO3/Universal-Science-Discovery.git cd Universal-Science-Discovery/repro/p-b-habitat-percolation-ecology-fss pip install -r requirements.txt python simulate_percolation_fss.py
Estimated runtime: ~15–40 s in Python, or a few minutes in the browser, at 400 Newman–Ziff samples/L. Both report the fit. Neither presets the result. Exit code is always 0; inspect stdout for CONFIRMED vs INCONCLUSIVE.