repulsive. Since the divergence is C 2 < 0, system (1)
admits a compact global attractor A¼Ap of measure zero.
Both for p ¼ 0 and p ¼ 0:83, system (1) has three equilibria
and Ap looks like an upside down and slightly distorted
copy of the standard Lorenz butterfly. For p – 0, rotation
symmetry [5] is broken and one wing becomes less developed
than the other. Please see Figs. 1 and 2.
Topological approaches to chaos detection in ordinary
differential equations are based on analyzing return maps
associated to Poincaré sections. A further dimension reduction,
if possible, plays a vital role here.
The aim of the present paper is to provide a computerassisted
proof for chaos in (1), both for p ¼ 0 and p ¼ 0:83.
We follow the Mischaikow–Mrozek–Zgliczynski approach
[6–9] for chaos detection in Poincaré return maps. This
approach does not require global Poincaré sections. It
requires disjoint, carefully selected quadrangles L and R
on a carefully selected local Poincaré plane such that the
return map on L [ R—in most cases, not the return map
itself, but some iterates thereof—is subject to a number
of topological constraints.
Preparing for a computer-assisted proof for chaos in
system (1), we pass to the set