Heisenberg Walk

symbolic-dynamics · 37 views
symbolic-dynamicssymbolic

What It Is

Random walk on the discrete Heisenberg group H(ℤ) with generators drawn uniformly from {(±1,0,0), (0,±1,0)} (the standard symmetric Carnot generating set). Emits the (x,y) coordinates interleaved as a byte stream (out[2i]=x_i mod 256, out[2i+1]=y_i mod 256), which is exactly the encoding the framework's Heisenberg geometry consumes — its embed() builds the path from consecutive byte pairs and accumulates the discrete Levy area z_{n+1} = z_n + x_n·Δy_{n+1} (accumulated x-position times the y-increment) itself. Each step fires exactly one generator, so x and y are NOT independent: exactly one coordinate moves by ±1 while the other holds (Δx,Δy ∈ {-1,0,+1} with probabilities 1/4, 1/2, 1/4), so each marginally is a ±1 walk idle ~half the steps. This coupling is the defining anisotropy: the Levy area grows like Var(z_n) ~ n²/8 in the number n of (x,y) steps (the vertical z scales as n while the horizontal x,y scale as √n) — the Carnot–Carathéodory homogeneity of nilpotent geometry.

Interpretation

Standard analysis sees: left-skewed; anti-correlated (alternating); homoskedastic; nonstationary / drifting. The atlas finds no named structure, but the source is distinctively extreme on Ordinal Partition:memory_order (+7.8z) — beyond what the standard bank predicts for it.

What standard analysis sees
tail heaviness0.46
asymmetry0.05
occupancy0.52
short-range corr0.13
long-range memory0.74
spectral colour0.50
periodicity0.74
complexity0.32
time-irreversibility0.59
volatility clustering0.05
multifractality0.64
dimensionality0.17
nonstationarity0.93
What the atlas adds
Atlas-extreme metrics the standard bank can’t predict for this source
Ordinal Partition:memory_order+7.8zbank-miss 1.8σ
Nonstationarity:ac1_trend-2.7zbank-miss 1.2σ
Wasserstein:recurrence_distance+2.7zbank-miss 1.0σ
Spectral Analysis:spectral_bandwidth+2.6zbank-miss 1.0σ
Wasserstein:distributional_stationarity-2.5zbank-miss 3.0σ
Projective ℙ²:distance_std+2.5zbank-miss 2.8σ

Composition

dtypeuint8
range[0, 255]
unique values94 / 16384
mean ± std165 ± 102

Render Gallery

Atlas Position

Nearest neighborDistance
Copeland-Erdős5.05cross-origin
Windows PE x86-645.57cross-origin
Champernowne5.59cross-origin

Open in Atlas →

Which Geometries Light Up

AutoRegressive › AutoRegressive:ar_coef_2rank 2/3060.6900
Boltzmann › Boltzmann:spectral_gap_Jrank 5/3060.9927
Boltzmann › Boltzmann:nn_dominancerank 303/3060.0090
H² × ℝ (Thurston) › H² × ℝ (Thurston):boundary_dynamicsrank 5/3060.1003
Hölder Regularity › Hölder Regularity:alpha_autocorrelationrank 3/3060.3738
Hölder Regularity › Hölder Regularity:holder_meanrank 303/306-1.8842
Isochronicity › Isochronicity:amplitude_explorationrank 2/3062.0540
Laplacian › Laplacian:biharmonic_ratiorank 2/30615.9087
Lorentzian › Lorentzian:causal_persistencerank 1/3060.9856
Möbius-S³ › Möbius-S³:hopf_fiber_coherencerank 305/3060.0106
Nonstationarity › Nonstationarity:ac1_trendrank 304/306-0.1898
Ordinal Partition › Ordinal Partition:memory_orderrank 1/3061.4040
Spectral Analysis › Spectral Analysis:spectral_bandwidthrank 1/3060.2394
Spirograph › Spirograph:rational_grid_proximityrank 305/3060.4454
in stochastic-process
alphabetical
← / → within domain · ⇧← / ⇧→ alphabetical · ⇧← / ⇧→ inside an open render = same view across sources