Other meanings of Self-similarity
MATHEMATICS
Self-similarity is the property in which an object resembles a part of itself at different scales. The resemblance may be exact, as in ideal mathematical constructions, or approximate, as in many natural forms. It is a central organizing idea in fractal geometry and helps describe structures whose detail persists when magnified.
Self-similarity describes a correspondence between an entire object and one or more of its parts after scaling, rotation, translation, or other permitted transformations. Exact self-similarity occurs when the smaller pieces are identical copies of the whole in a strict mathematical sense; approximate self-similarity preserves broad patterns while allowing local differences. Statistical self-similarity, common in turbulence and natural landscapes, means that measured properties have similar distributions across scales rather than identical outlines.1
The concept is closely associated with fractal geometry, although not every self-similar object is treated as a fractal. A line segment, for example, is self-similar under ordinary Euclidean scaling but has an integer dimension. Fractals typically combine repeated structure with exceptional detail, irregular boundaries, or dimensions that are not whole numbers.2
Many exact examples arise by repeatedly applying a finite set of scale transformations. An iterated function system generates a limiting set by replacing each figure with transformed copies of itself; the Sierpiński triangle and Koch snowflake are familiar examples. Their visible complexity comes from an unbounded succession of operations, even though the generating rules are short and finite.3
The dimension of such a set can be calculated from its scaling ratios and may be fractional. The Hausdorff dimension of the Sierpiński triangle is log 3 divided by log 2, while the Koch curve has dimension log 4 divided by log 3. Hutchinson’s mathematical framework established general conditions under which these constructions possess a unique compact attractor.3
Natural systems often exhibit approximate or statistical self-similarity rather than perfect repetition. Branching in trees, river networks, blood vessels, lungs, and some coastlines can show related patterns over a limited range of scales, reflecting shared constraints on transport, growth, or optimization. These examples are not infinitely self-similar: physical materials, cellular structure, and finite size impose lower and upper limits.1
Self-similar models are used in image compression, signal analysis, network science, geophysics, and the study of turbulence. In physics, the renormalization group explains how descriptions can acquire scale-invariant behavior near critical points. In complex dynamics, Julia sets display recurring structure under repeated iteration, making self-similarity a visual expression of the underlying dynamical rules.
Self-similarity does not require a visually obvious pattern, and resemblance can be encoded in statistics, measures, or mathematical laws. A probability distribution may be self-similar under rescaling even when individual samples look unrelated. Likewise, a self-affine object may scale by different factors in different directions, so ordinary similarity is replaced by anisotropic stretching; many rough surfaces are better described this way than by isotropic fractals.
The apparent scale-free character of a fractal is also usually bounded. A real coastline stops displaying the same behavior below the grain size of sediment, while a digital image has a finite pixel scale. The modern study of fractals was strongly shaped by Benoît Mandelbrot, who connected mathematical self-similarity with irregular forms in nature, economics, and physical measurement.2 Distinguishing exact, statistical, and finite-range self-similarity prevents exaggerated claims about natural patterns.
Self-similarity is a scale-related property, not a claim that every part of an object is identical to the whole at every possible scale.
Help improve the encyclopedia. Reports go straight to the site manager.