A narrow, exponentially thin region of phase space—the "singular funnel"—can make common reductions of fast–slow dynamical systems give misleading predictions about which attractor a system will reach.
Reduced models that remove fast variables can miss an exponentially thin basin-of-attraction feature called a singular funnel.
Multistability and resilience assessments based on the reduced description can be wrong because the reduced model omits the funnel's narrow basin tongue.
# What the paper shows Systems with both fast and slow variables are often simplified by eliminating the fast dynamics and substituting either their steady-state or time-averaged behavior back into the slow equation. This reduction lowers dimension and makes analysis tractable. Yanchuk and colleagues demonstrate that this standard strategy can fail: the full system can contain an exponentially thin region of phase space—a "singular funnel"—that the reduced model does not capture.
# Why the funnel matters In dissipative dynamical systems the long-term outcomes are attractors, and each attractor has a basin of attraction: the set of initial conditions that evolve toward that attractor. For multistable systems, understanding basins determines which attractor a given initial condition will reach.
The singular funnel is a narrow tongue in one basin that clings to the axis where the fast variable equals zero and extends arbitrarily far in the slow-variable direction. Because it is exponentially thin, a reduced model that removes the fast variable typically does not represent it. That omission changes basin geometry and therefore changes which attractor the system is predicted to approach.
# Simple example used
# Practical implications
# What to watch for in applications Researchers and modelers who simplify multiscale systems by eliminating fast variables should not assume the reduced model preserves basin topology. When multistability matters—when initial conditions, perturbations, or resilience are central questions—check whether the omitted fast dynamics could contain fine-scale features like a funnel that affect outcomes.
# Bottom line Eliminating fast variables can produce a useful low-dimensional model, but that model can miss exponentially thin structures in phase space that change which attractor the full system reaches. When outcomes depend on basin geometry, the full fast–slow dynamics may be necessary to avoid misleading conclusions.
Feature-rich products can overwhelm new users. Staged access, visible choices, and one useful first result can make adoption easier and more lasting.
A story about geometry, fragility, and the slow realization that “distance” in finance was never the ordinary kind. Continue reading on Medium »
First in a series on probabilistic forecasting for physical signals. Next: what happens when you roll the forecast forward more than one step. The post Your Model’s MSE Is Lying to You appeared first on Towards Data Science .
Almost every conversation about observability budgets I have been in ultimately arrives at the same conclusion: “we need to reduce our telemetry volume.” That sentence is usually followed by a number. Thirty percent. Half. Whatever the finance spreadsheet needs it to be. Then someone says the thing that makes everyone
Anyone who has tried to correct a conspiracy theory with facts has likely met the “backfire effect”: direct correction can harden the very belief it was meant to change.
While judges may be concerned about how art is created (see the first article in the series: Creators, Spectators and Judges: The art establishment), spectators are not. The post Creators, Spectators and Judges: Part 2 appeared first on On Landscape .
Loading more related stories...
Open the app view to save this story, compare related coverage, and continue from the same source.