Limit cycles of a planar vector field (Q1365513): Difference between revisions

Consider a two-dimensional autonomous differential system \[ dx/dt= P(x,y),\quad dy/dt= Q(x, y)\tag{\(*\)} \] with a unique equilibrium point \(E\). Let \(B(x, y)\) be a positive smooth function, let \(d_B(x,y)\) be the divergence of the vector field \((BP,BQ)\). It is assumed that \(d_B(x,y)= \lambda\) defines a closed non-selfintersecting curve \(\alpha(\lambda)\) for some interval \([\underline\lambda,\overline \lambda]\). The region bounded by \(\alpha(\lambda)\) is denoted by \(\Omega_\lambda\), its area by \(A(\lambda)\), let \[ I(\lambda):= \iint_{\Omega_\lambda} d_B(x, y)dx dy. \] \(\alpha(\lambda)\) is called a compensating curve if \(E\subset \Omega_\lambda\), \(0< A(\lambda)<\infty\), \(I(\lambda)= 0\). The authors prove that among all areas of regions bounded by compensating curves, the area of the region bounded by a limit cycle is minimal. This property can be used to estimate the maximal number of limit cycles for a class of vector fields.