Basins of attraction
When a system has several possible long-term outcomes, each starting state belongs to the basin of the outcome it approaches. Coloring those basins turns prediction into geometry.
Interior confidence, boundary uncertainty
Deep inside a basin, small errors leave the outcome unchanged. Near a fractal boundary, arbitrarily small perturbations can cross into another basin. The boundary’s dimension can quantify how uncertainty scales with measurement precision.
Deep inside a basin, small changes of initial state usually reach the same attractor. At the boundary, every neighborhood may contain points with different destinations. This local unpredictability creates the fine interlocking pattern without making the algorithm uncertain everywhere. A map should distinguish basin interior from boundary proximity, for example by combining destination with convergence time or distance.
More than two outcomes
Some systems have Wada boundaries: every boundary point borders three or more basins. The same visible frontier then belongs to all outcomes, making local prediction especially delicate.
With three or more attractors, a Wada boundary may occur, where every boundary point borders all basins at once. Visually, every color repeatedly penetrates the same edge regions. A raster image can only suggest the Wada property; establishing it requires topological or numerically controlled tests. A colorful tangle by itself is not sufficient.
Numerical caution
A basin image depends on stopping criteria, integration accuracy and maximum time. Points that converge slowly may be misclassified as a separate outcome. Resolution studies and independent checks are essential before treating colored filaments as system properties.
Stopping conditions shape the image. An orbit may converge slowly, jump near a singularity or leave the computed domain. Those cases need separate states rather than an arbitrary basin color. Greater precision and iteration depth should keep major boundaries stable while adding finer nesting. If whole color fields move, the classifier rather than the dynamics is probably visible.
Sources and further reading
This article summarizes the following specialist sources in original wording. Accessed and editorially reviewed 12 August 2026.
- Newton's MethodWolfram MathWorld
- Fractal Geometry: Mathematical Foundations and ApplicationsWiley
Continue the idea
Newton fractals
How Newton’s root-finding method divides the complex plane into colored basins with fractal boundaries.
Strange attractors
Why some dissipative chaotic systems settle onto bounded sets with sensitive motion and fractal geometry.
Fractal dimension
What fractal dimension measures, how box counting works and why one object can have more than one meaningful dimension.