Laplacian

Curvature cascade, non-periodic boundary content, monotone runs
scaleencoding-invariantdim 1D8 metrics

What It Measures

The curvature cascade of a 1D signal — how energy propagates through successive discrete derivatives.

Applies the discrete Laplacian (second difference) and its iterates to the signal. The biharmonic ratio measures how much energy survives two additional derivative orders; the Poisson recovery error quantifies non-periodic boundary content by solving Δu = Δf with periodic FFT and measuring the residual. Gradient metrics capture monotone run structure and its coupling to curvature.

Metrics

biharmonic_ratio

Energy ratio ||f''''||²/||f''||². Logistic Period-2 (16.0) and Period-4 (15.9) score highest: their sharp alternating jumps propagate maximum energy through successive derivatives. Bearing Outer (1.05) scores lowest among non-constant sources — its smooth vibration envelope dies fast under differentiation.

poisson_recovery_error

Residual when solving Δu = Δf with periodic boundary conditions via FFT. GUE Spacings (9.7M) and Fibonacci Quasicrystal (8.8M) score highest: their structure lives at the boundaries, not in the bulk periodic component. Quantum Walk (0.02) scores near zero — its probability distribution is well-captured by periodic Fourier modes.

gradient_sign_persistence

Fraction of consecutive first differences with the same sign (monotone runs). Square Wave (0.99) nearly always continues in the same direction. Logistic Edge-of-Chaos (0.0) and Fibonacci Word (0.0) reverse direction at every step.

laplacian_spectral_ratio

Low/high frequency energy ratio of the Laplacian. Standard Map Mixed (143.3) dominates: its curvature energy is concentrated at low frequencies (broad smooth bends in the mixed regular/chaotic dynamics). Logistic Period-2 (0.0) has zero low-frequency curvature — all its energy is at the Nyquist frequency.

curvature_autocorrelation

Lag-1 autocorrelation of |f''|. Devil's Staircase (0.80) scores highest: its long constant plateaus create persistent curvature regimes. Logistic Period-3 (-0.50) is maximally anti-persistent — curvature events alternate with flat stretches.

cross_scale_curvature_coherence

Correlation of |Laplacian| computed at scale 1 vs scale 2 (dilated kernel). Bearing Inner (0.79) and Ocean Swell (0.77) score highest: their curvature structure is hierarchically consistent. Phyllotaxis (-0.59) and Circle Map QP (-0.59) have anti-correlated curvature across scales.

gradient_curvature_anticorrelation

Correlation between monotone-run mask and |f''|, negated. Measures whether smooth runs coincide with low curvature. Phyllotaxis (1.0) and Circle Map QP (1.0) have perfect coupling. Lotka-Volterra (-0.31) has the opposite: its smooth runs carry high curvature (curved oscillation arcs).

laplacian_evolutionary_index

Product of curvature_autocorrelation and gradient_curvature_anticorrelation. Shuffled Blocks (0.54) scores highest: its random block boundaries create curvature events that cluster AND couple to gradient structure. Logistic Period-3 (-0.50) is the most negative — strong anti-persistent curvature with inverted coupling.

Atlas Rankings

biharmonic_ratio
SourceDomainValue
Logistic r=3.2 (Period-2)chaos15.9980
Logistic r=3.5 (Period-4)chaos15.8529
Noisy Period-2chaos15.7787
···
Constant 0xFFnoise0.0000
Bearing Outerbearing1.0489
Ocean Swellgeophysics1.2218
cross_scale_curvature_coherence
SourceDomainValue
Bearing Innerbearing0.7878
Ocean Swellgeophysics0.7715
ECG Fusionmedical0.7458
···
Phyllotaxisbio-0.5878
Circle Map Quasiperiodicchaos-0.5877
Critical Circle Mapchaos-0.5832
curvature_autocorrelation
SourceDomainValue
Devil's Staircaseexotic0.8013
Logistic r=3.68 (Banded Chaos)chaos0.7660
Rössler Hyperchaoschaos0.7477
···
Logistic r=3.83 (Period-3 Window)chaos-0.5000
Critical Circle Mapchaos-0.3448
Circle Map Quasiperiodicchaos-0.3090
gradient_curvature_anticorrelation
SourceDomainValue
Phyllotaxisbio1.0000
Circle Map Quasiperiodicchaos1.0000
Logistic r=3.83 (Period-3 Window)chaos0.9965
···
Lotka-Volterrabio-0.3057
Hodgkin-Huxleybio-0.2281
Clipped Sinewaveform-0.2260
gradient_sign_persistence
SourceDomainValue
Constant 0x00noise1.0000
Square Wavewaveform0.9919
Triangle Wavewaveform0.9797
···
Logistic Edge-of-Chaoschaos0.0000
Logistic r=3.2 (Period-2)chaos0.0000
Fibonacci Wordexotic0.0000
laplacian_evolutionary_index
SourceDomainValue
Shuffled Blocksexotic0.5405
Coupled Map Latticechaos0.3372
Henon Mapchaos0.3065
···
Logistic r=3.83 (Period-3 Window)chaos-0.4983
Critical Circle Mapchaos-0.3377
Circle Map Quasiperiodicchaos-0.3090
laplacian_spectral_ratio
SourceDomainValue
Standard Map K=0.5 (Mixed)chaos143.2595
Ocean Swellgeophysics20.6555
Bearing Outerbearing16.2628
···
Constant 0xFFnoise0.0000
Logistic r=3.2 (Period-2)chaos0.0000
Logistic r=3.5 (Period-4)chaos0.0000
poisson_recovery_error
SourceDomainValue
GUE Spacingsquantum9739279.8412
Fibonacci Quasicrystalnumber_theory8845418.5383
Phyllotaxisbio8593151.5043
···
Constant 0xFFnoise0.0000
Quantum Walkquantum0.0173
Random Stepsexotic0.1925

When It Lights Up

Laplacian sits in the Scale lens and captures derivative-order energy cascade — complementary to Hölder (pointwise smoothness) and p-Variation (path roughness). In the atlas, biharmonic_ratio separates periodic signals with sharp transitions (high) from smooth oscillators (low). The poisson_recovery_error is unique: it's the only metric that specifically measures non-periodic boundary content, lighting up on quasicrystal and number-theoretic sequences whose structure doesn't fit into periodic Fourier decomposition.

Open in Atlas
← Ammann-Beenker (Octagonal)Hodge–Laplacian →