The cellular automata that commute with Rule 54, and rigidity of the second-order automaton 94R.
We find all invertible cellular automata commuting with the integrable Rule 54: two copies of the particle-conserving automorphisms of the golden mean shift.
https://t.co/BaMXQEloJU
The 88 second-order elementary cellular automata: a classification by automorphism group.
60 of the 88 are settled: in 43 the automorphism group contains every finite group; in 17 it is computed exactly and is abelian. 28 remain open.
https://t.co/XjlkJahLpi
Every finite group from a wall: large automorphism groups of second-order cellular automata.
In 42 of 88 classes, reversible rule 54 among them, walls nothing crosses let you permute what lies between: the automorphism group contains every finite group.
https://t.co/FitewoFmnD
@DrJimFan Sora is doing what Bob Ross does. Without foreknowledge of a mountain's exact 3D geometry, he makes the type of brushstrokes that only create plausible rocky surfaces. Sora has similarly learned the plausible "brushstrokes" of turbulent water, fabric, snow, skin, trees, etc.
@TheZvi The trope is so powerful in all text that even if a close friend asks you "how you really feel" you have no choice but to negate the obvious answer or suffer repeat questioning.
Looking for a statistician interested in poring over 11 years of daily food diaries, weights, steps, body comp, and RMR as someone lost 100 lbs. and regained half to find patterns. Zero missing days—everything logged from 4000 kcal binges to fasts. Hoping for novel insights like:
Perhaps a simple control system model can predict daily RMR using rules like: when omega-6 to 3 ratio exceeds 10:1, if protein also exceeds 50g, lower RMR to store all fat consumed, else raise RMR to maintain current set point.
@KilianNicolai @thomasahle We are looking for hidden working memory used for planning that contradicts the claim it's only predicting one word at a time. Multiple revealed steps removes the need to do anything but predict the next word by using the context window as memory.
@sprooos@thomasahle When asked, it can recite the first and last letter of any word, but the same working memory may be evident in an analogous example that asks it to unwind nested, dependent clauses.
For 3 months, the heaviest men were on 3500 calories per day, while the lightest were on 4700 calories per day. What happened to them?
The average change was a 1.76lb LOSS!
The heaviest (remember, max 182lbs) lost 12.5lbs on 3400 cal!
The lightest eating 4700 cal gained 8.4!
This obesity thing is like the biggest possible nerdsnipe. A huge mystery that affects billions of people, with tons of data and no satisfying answers as of yet.
Like how does Iran have an obesity rate *~12 times higher* than Vietnam, a country of very similar wealth?