No label defined (Q4030596)

Consider the \(C^ 3\) differential system (1) \(\dot x=f(x,y)+\alpha f_ 0(x,y,\alpha,\delta)\), \(\dot y=g(x,y)+\alpha g_ 0(x,y,\alpha,\delta)\), \(\alpha\in R\), \(\delta\in R^ n\). Suppose when \(\alpha=0\) system (1) has a singular closed orbit \(L_ 0\) passing through a unique saddle point \(N\). Let \(\sigma_ 0=(f_ x+g_ y)_ N\), \(\sigma=\oint_{L_ 0}(f_ x+g_ y)dt\), \(M(\delta)=\oint_{L_ 0} e^{-\int_ 0^ t(f_ x+g_ y)dr}(fg_ 0-gf_ 0)dt\). Theorem 1. If \(\sigma_ 0=0\), \(\sigma\neq 0\), \(M(\delta_ 0)\neq 0\) for \(\delta_ 0\in R^ n\), then for \(|\alpha|\) and \(|\delta- \delta_ 0|\) sufficiently small, (1) has a unique limit cycle in a small neighbourhood of \(L_ 0\), iff \(\alpha\sigma M(\delta_ 0)>0\). The stability of this limit cycle is determined by \(\text{sgn }\sigma\). Now suppose when \(\alpha=0\) system (1) has a family of closed orbits: \(L_ \lambda\): \(z=z(t,\lambda)=(x(t,\lambda),y(t,\lambda))\), \(0\leq t\leq T_ \lambda\), \(0<\lambda<K\), \(L_ \lambda\to L_ 0\) as \(\lambda\to 0\). Let \(\Phi(\lambda,\delta)=\oint_{L_ \lambda} e^{-\int_ 0^ t (f_ x+g_ y)dr}(fg_ 0-gf_ 0)_{\alpha=0}dt\), and \(\Delta(\alpha,\delta)=\lambda_ 1(\alpha,\delta)-\lambda_ 2(\alpha,\delta)\), where \(\lambda_ 1(\alpha,\delta)\) and \(-\lambda_ 2(\alpha,\delta)\) are the characteristic roots of (1) at \(N\). Theorem 2. (i) \(M(\delta)=\Phi(0+0,\delta)\); (ii) If \(M(\delta_ 0)\neq 0\), then for \(|\alpha|\neq 0\) and \(|\delta-\delta_ 0|\) sufficiently small, (1) has no closed orbit in a small neighbourhood of \(L_ 0\); (iii) If \(M(\delta_ 0)=0\), \(\Delta_ \alpha '(0,\delta_ 0)\neq 0\), then \(L_ 0\) can bifurcate at most one limit cycle when \(|\alpha|\neq 0\) and \(|\delta-\delta_ 0|\) sufficiently small. There are also two similar theorems for \(\delta\in R\), in which \(M'(\delta_ 0)\neq 0\) is used.