This is an earlier experiment. Read the amendments below before treating its findings as current.
The question
Pairing strong settings, storm strength 10 plus fog gave 16 collisions where each alone gave none.
What was tested
Nine pairs of settings, fourteen families, four layouts. Then two fixes to how fog shortens ranges.
What was found then
Fog had shrunk the close-range push apart to 1.4 m/s², against a 4 m/s² pull toward the goal. Using the clear-air range for that push (fog limits how far you see, not how hard you push once close) took eight layouts from 32 collisions to 1, with work down 0.06 %. Storm plus fog plus a slow link then gave 29 collisions in 56 cases: the rules applied fog to neighbours heard by radio. Ignoring fog for the push away from radio-heard neighbours took 36 to 11, with work up 1.4 %. Doing the same for speed-matching also gave 11 but cost 6.3 % of work, so it was left out.
What changed
The close-range push ignores fog; so does the push away from neighbours heard by radio.
Numbers and source references
- 16
- collisions from storm 10 + fogREADME.md:730
- Nine
- pairs of settingsREADME.md:730
- fourteen
- familiesREADME.md:730
- four seeds
- seedsREADME.md:730
- 1.4 m/s²
- fogged close-range push at one standardREADME.md:730
- 4 m/s²
- goal pullREADME.md:730
- 32 to 1
- eight seeds, storm 10 + fog 0.8README.md:730; git e8ffc79
- −0.06 %
- workREADME.md:730
- 29 in 56 rows
- storm + fog + 2 Hz link tripleREADME.md:730
- 36 → 11
- the three link rows over four seedsREADME.md:730; git 38ec347
- +1.4 %
- workREADME.md:730
- −6.3 %
- work with alignment fog-blind tooREADME.md:730
- eight seeds
- seedsREADME.md:730