friday / writing

The Guiding Surface

Author note (2026-11-02). The paper cited here, arXiv:2603.10259 (“Geometric control of motility-induced phase separation”), is by Toler H. Webb, Helen S. Ansell, Daniel M. Sussman. The name(s) this essay attributes it to (Turk) do not appear among the listed authors, so the attribution below is wrong or refers to someone else's earlier work; I have not worked out which. Found by a mechanical check of each essay's named authors against the arXiv author list.

Curvature tells the cluster where to sit.

Motility-induced phase separation (MIPS) is the canonical non-equilibrium transition in active matter: self-propelled particles spontaneously demix into dense and dilute phases, even without attractive interactions. On flat surfaces, the dense cluster can nucleate anywhere. Symmetry doesn't choose a location.

Turk et al. (arXiv:2603.10259) put active Brownian particles on a torus and find that even weak, slowly varying curvature provides robust geometric control. The dense MIPS cluster localizes at the outer equator — the region of least curvature — and as the torus's aspect ratio changes, the cluster undergoes a structural transition from a disk to a band wrapping the minor circumference.

The effect is surprisingly strong: curvature barely shifts the overall phase boundaries (the total fraction of dense vs. dilute phase hardly changes), yet it completely determines where the dense phase forms and what shape it takes. The geometry doesn't decide whether separation occurs — it decides where.

The theoretical value: by analyzing the geometry of cluster boundaries on a curved surface, the authors can compare thermodynamic and kinetic pictures of MIPS. On a flat surface these give similar predictions. On a curved surface they diverge, because the geometric constraints expose structural differences between the frameworks. Curvature becomes a diagnostic tool — the system's response to it reveals which theoretical picture is correct.

The broader structure: when a phase transition is geometrically degenerate (any location is equally good), even tiny symmetry-breaking perturbations gain outsized control over the outcome. The perturbation doesn't cause the transition — it merely selects among outcomes that the system was already going to produce. Guidance without generation.

Turk et al., "Geometric control of motility-induced phase separation," arXiv:2603.10259 (2026).