A reading series for physics-loving high-schoolers and undergraduates
Why the world still works when you throw things away ── you don't have to track 10²³ molecules; temperature and pressure will do. You can do chemistry without knowing a thing about quarks. This property of the universe ── that discarding the fine detail still gets the coarse answer right ── has a name, and the name is renormalization. This series re-threads the measurement problem of "Quantum That Clicks" and the holography of "Black Holes That Click" onto a single line.
Discard the position and velocity of all 10²⁴ molecules in a glass of water and keep just two numbers, temperature and pressure. The prediction still lands. The reason is 1/√N ── averages sharpen as the count grows. Entropy is just another name for "how much you threw away."fluctuation ∝ 1/√N
The neutron's 880-second lifetime is not a dice roll. Make the mass slightly complex, m−iΓ/2, and decay is simply the phase rotating into the complex plane. What separates oscillation from decay is whether the destination is discrete or continuous ── and that distinction reaches all the way to the neutron lifetime puzzle.m → m − iΓ/2
Why can an electron be in superposition and a cat cannot? The phase is not destroyed ── it is diluted into the environment. And the dilution rate explodes with the size of the system. "Treat it as probability once it's big" is not a mood; it is a theorem with a computable clock.τ_dec ∝ 1/N
The critical point of water, the critical point of an iron magnet, and a toy model of arrows on a lattice. Their microscopics have nothing in common, yet the critical exponents agree exactly. Repeat coarse-graining and theories flow to a fixed point, discarding their origins on the way ── the heart of this series.ν ≈ 0.630 (3D Ising)
Coarse-graining is not universal. At a critical point the correlation length diverges and all scales couple; in turbulence energy flows between scales; in chaos the microscopic is amplified into the macroscopic. Counting the places where "throw it away" fails is what gives the claim its content.t_pred ≈ (1/λ)ln(1/ε)
Only the Higgs mass and the cosmological constant fail to shield themselves from the floor above. Cancellations to 34 and 120 digits. Modern physics' two greatest puzzles turn out to have exactly the same shape ── and reading that shape tells you what "naturalness" really means.Λ_obs / Λ_theory ≈ 10⁻¹²⁰
In AdS/CFT the radial direction of the bulk corresponds to the renormalization scale of the boundary theory. Going deeper is coarse-graining. If one dimension of space is really an axis of resolution, then "the universe is coarse-grainable" and "the universe is holographic" become two descriptions of one claim.radial coordinate z ↔ RG scale
Being coarse-grainable and having something interesting happen are two different things. The arrow of time exists only because the universe began in an absurdly special state ── 1 part in 10^(10¹²³) by Penrose's estimate. A universe that began at equilibrium has no layers, no structure, no observers.1 / 10^(10¹²³)
Why is the universe coarse-grainable? A third answer: because a universe that isn't has nobody in it to ask. Memory, learning and prediction are all impossible without discarding information. An observer is just a name for a machine that predicts the future from coarse-grained variables ── this one shakes hands with "Learning That Clicks."prediction = discard, then get it right
Treat "add them and divide by √N" as a transformation and the Gaussian becomes its fixed point, with the central limit theorem as the claim that it attracts. The k-th cumulant shrinks as N^(1−k/2) ── mean relevant, variance marginal, skewness and above all irrelevant. The episode where 1 and 4 join up from behind.κ_k → κ_k N^(1−k/2)
"Coarse-graining loses information" is a theorem (the data processing inequality); "how far may you squash" is answered by rate–distortion theory. And the heart of it ── having layers means having a knee in the rate–distortion curve. Ending with how an information inequality proved the renormalization group's c-theorem.a knee in R(D) = a layer
Nothing can be learned without assumptions (no free lunch), so if learning works the world must be layered. Infinite width = free field, 1/width = interaction, depth = renormalization flow ── sorted into rigorous / promising / metaphor, and then: 〈why is learning possible〉 = 〈why is the universe coarse-grainable〉.learning = finding what to discard
The linking episode with the sister series. The D of "The Universe Is a Computer"'s F=1/(Cn)^D is a scaling dimension (its complex dimension = a renormalization eigenvalue gone complex), and "Cosmology That Clicks"'s c·t=constant shares its argument with the Callan–Symanzik equation. All three had already crossed at the neutron, at Landauer, and at the continuum limit.D ≡ scaling dimension / μ d/dμ = 0