Multifractals
A monofractal is summarized by one dominant scaling exponent. A multifractal contains interwoven regions with different local scaling strengths and requires a spectrum rather than one number.
Measures, not only shapes
Multifractal analysis often studies how mass, probability, energy or activity is distributed over boxes as scale changes. Dense and sparse regions follow different exponents even when they occupy the same geometric support.
A multifractal analysis studies not only the shape of a support but how mass or probability is distributed over it. Small boxes can have different local exponents. A uniformly occupied geometric carpet may need one dimension; intermittent turbulence or unequal cascades need a spectrum. Preprocessing must therefore define the measure being analyzed explicitly.
From moments to a spectrum
Partition functions weight box probabilities by powers q. Positive q emphasizes dense regions; negative q emphasizes sparse ones and is sensitive to noise. Scaling exponents can be transformed into a singularity spectrum describing the dimensions of local-strength subsets.
Powers of box masses are summed for different moments q. Positive q emphasizes dense regions and negative q sparse ones. Scaling yields τ(q) and, through a Legendre transform, an f(α) spectrum. Negative moments are extremely sensitive to empty or noisy boxes. A broad spectrum can reveal real heterogeneity or merely amplify finite sampling and detection limits.
A demanding empirical method
Reliable spectra need broad scale range, sufficient samples and robustness checks against trends and finite-size effects. A wide estimated curve is not automatically evidence of a complex cascade; surrogates and uncertainty matter.
Reliable multifractal analysis needs many scales, sufficient data and stability under window choice and detrending. Synthetic surrogates test whether a spectrum exceeds what linear correlation or distribution alone could produce. A colorful f(α) curve without uncertainty is easy to generate but weak evidence. Method comparison belongs to the result.
Sources and further reading
This article summarizes the following specialist sources in original wording. Accessed and editorially reviewed 12 August 2026.
- Fractal Geometry: Mathematical Foundations and ApplicationsWiley
- Fractal dynamics in physiologyPhysiological Reviews
Continue the idea
Fractal dimension
What fractal dimension measures, how box counting works and why one object can have more than one meaningful dimension.
Fractal patterns in physiological signals
How scale-related fluctuations are studied in heartbeat and other physiological time series—with necessary methodological caution.
Clouds, turbulence and scale
Why cloud boundaries and turbulent fields show statistical scaling rather than exact geometric self-similarity.