Correlation Dimension¶

A fractal-dimension statistic given by the small-radius scaling exponent of the probability or normalized count that two sampled points lie within distance epsilon.

Version: v4 · 2026-09-19 · History
Domain-specific #: 1567
Domain group: Formal Sciences
Origin domain: Mathematics
Subdomains: Chaos Theory, Strange Attractor Characterization → Mathematics
Aliases: Correlation exponent, Grassberger–Procaccia dimension

Core Idea¶

The correlation dimension is a fractal dimension defined by how the probability that two independently sampled points lie within distance r scales as r becomes small. For a probability measure μ on a metric space, define a correlation integral.

C(r) = ∫ μ(B(x,r)) dμ(x) = P(d(X,Y) < r)

for independent X,Y ~ μ, subject to boundary and convention details. When a stable limit exists,

D_2 = lim_{r→0} log C(r) / log r.

For finite data, the Grassberger–Procaccia estimator counts close pairs over a range of radii and estimates the slope of log C_N(r) against log r in a defensible scaling region.

How would you explain it like I'm…

The Close-Pairs Measure

Sprinkle lots of dots along a shape. Pick two dots at random and check if they are close together. If you keep making 'close' tinier and tinier, the way that chance shrinks tells you what kind of shape the dots make, like a thin line, a flat sheet, or something wiggly in between.

Close-Pairs Spread Number

Some shapes in nature are so wrinkly that they're not quite a line and not quite a flat sheet — their 'dimension' can be somewhere in between. The correlation dimension is one way to measure that. Pick two points from the shape at random and ask: what's the chance they are closer than some small distance r? As you make r smaller and smaller, that chance shrinks, and how fast it shrinks gives the dimension. With real data, scientists count close pairs for many distances, draw the counts on a special graph, and measure the steepness of the straight part, being careful that noise or too little data isn't fooling them.

Pair-Scaling Fractal Dimension

The correlation dimension is a kind of fractal dimension. Take a set of points spread according to some probability pattern, and pick two points independently. The correlation integral C(r) is the probability that they lie within distance r of each other. For small r, C(r) behaves roughly like r raised to some power, and that power is the correlation dimension: D₂ = the limit of log C(r) / log r as r goes to 0, when that limit exists. With finite data — for example, points on a chaotic attractor reconstructed from a time series — the Grassberger–Procaccia method counts close pairs at many radii and fits the slope of log C against log r in a range where it looks straight. That slope is only meaningful if you check that it reflects real geometry, not limited resolution, noise, or points that are close just because they were measured close together in time.

The correlation dimension D_2 of a probability measure mu on a metric space is defined from the correlation integral C(r) = integral of mu(B(x, r)) d mu(x), which equals P(d(X, Y) < r) for independent X, Y drawn from mu (up to boundary and convention details). When the limit exists, D_2 = lim as r goes to 0 of log C(r) / log r, describing how the probability of close pairs scales with radius. For finite data, typically points on an attractor reconstructed from a time series, the Grassberger–Procaccia estimator computes the empirical fraction C_N(r) of pairs within r across a range of radii and estimates the slope of log C_N(r) versus log r in a defensible scaling region. The concept is more than that plotted slope: it depends on the measure or sampled attractor, the metric and embedding, close-pair counting, the asymptotic scaling assumption, and diagnostics that separate genuine geometric scaling from finite resolution, noise, and temporal dependence among samples.

Scope of Application¶

Correlation dimension is used in nonlinear dynamics, fractal geometry, experimental time-series analysis, recurrence analysis, and characterization of invariant measures. Its computational accessibility made it a prominent tool for assessing low-dimensional attractor hypotheses from observations.

For points sampled independently from a smooth k-dimensional region with regular density, close-pair probability typically scales like r^k away from boundaries. Multifractal measures can yield a correlation dimension reflecting order-two mass concentration and differing from support dimensions.

Clarity¶

A common empirical estimator for N distinct sample indices is

C_N(r) = [2/(N(N−1))] Σ_{i<j} 1{d(x_i,x_j)<r}.

Some conventions include ordered pairs or self-pairs with a different normalization. Self-pairs are dangerous because their zero distance dominates as r shrinks. For trajectories, pairs with |i−j| below a chosen exclusion lag should often be omitted.

Manages Complexity¶

A strange attractor may occupy a complicated subset of a high-dimensional phase space. Correlation dimension compresses its mass distribution into one exponent that can distinguish a line-like, surface-like, or fractional concentration pattern. Pair counting avoids explicit construction of an optimal cover.

That efficiency trades away spatial detail. Different measures can share D_2; local dimensions may vary; topology and dynamical equations are not recovered.

Abstract Reasoning¶

Start from probability, not the graph. Interpret C(r) as close-pair mass before fitting a line.

Locate scale boundaries. Reject radii dominated by instrument resolution, noise, self-pairs, or finite diameter.

Check slope stability. A candidate exponent should persist over a meaningful interval and across increasing data size.

Knowledge Transfer¶

The portable skeleton is scale exponent from pairwise neighborhoods: count how rapidly the probability of local coincidence grows with radius and infer an effective dimension. This can organize analyses of point clouds, spatial events, and representation manifolds.

Literal transfer requires a metric, a sampling measure, and a defensible asymptotic or scaling regime. A log–log slope in network statistics or market data is only analogous until those elements and dependence controls are supplied.

Relationships to Other Abstractions¶

Current abstraction Correlation Dimension Domain-specific

Parents (3) — outgoing typed relations

Correlation Dimension is a kind of Dimension Prime

Dimension is instantiated as the sought scaling exponent.
Correlation Dimension presupposes Allometry and Scaling Law Prime

Dimension is instantiated as the sought scaling exponent.
Correlation Dimension presupposes Metric Prime

Dimension is instantiated as the sought scaling exponent.

Hierarchy paths (3) — routes to 3 parentless roots

Correlation Dimension → is a kind of Dimension

Show alternative paths (2)

Neighborhood in Abstraction Space¶

Correlation Dimension sits in a sparse region of the domain-specific corpus (63^rd percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Foundations of Probability & Inference (29 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-25