AN ATLAS OF ROBUST, STABLE, HIGH-DIMENSIONAL LIMIT CYCLES

We present a method for constructing dynamical systems with robust, stable limit cycles in arbitrary dimensions. Our approach is based on a correspondence between dynamics in a class of differential equations and directed graphs on the n-dimensional hypercube (n-cube). When the directed graph contains a certain type of cycle, called a cyclic attractor, then a stable limit cycle solution of the differential equations exists. A novel method for constructing regulatory systems that we call minimal regulatory networks from directed graphs facilitates investigation of limit cycles in arbitrarily high dimensions. We identify two families of cyclic attractors that are present for all dimensions n >= 3: cyclic negative feedback and sequential disinhibition. For each, we obtain explicit representations for the differential equations in arbitrary dimension. We also provide a complete listing of minimal regulatory networks, a representative differential equation, and a bifurcation analysis for each cyclic attractor in dimensions 3-5. This work joins discrete concepts of symmetry and classification with analysis of differential equations useful for understanding dynamics in complex biological control networks.