Arbitrary order bifurcations for perturbed Hamiltonian planar systems via the reciprocal of an integrating factor (Q5958917): Difference between revisions

This paper is devoted to study planar systems of the form \[ \dot x=-\partial H(x,y)/\partial y+f(x,y,\varepsilon),\quad \dot y=\partial H(x,y)/\partial x+g(x,y,\varepsilon), \] such that when \(\varepsilon=0\) the origin is a center, \(f\) and \(g\) are analytical at the origin and \(f(x,y,0)=g(x,y,0)\equiv 0.\) More specifically, the authors are interested to control which periodic orbits of the center persist for \(\varepsilon\) small enough. It is well known that to study this problem it is useful to compute the return function \(d(h,\varepsilon)=\varepsilon M_1(h)+ \varepsilon^2 M_2(h)+\cdots \) associated to a transversal section to the center, which is usually parameterized by the Hamiltonian function. In fact, just the first nonidentical zero \(M_n(h)\) is needed. This function is called the \(n\)th Melnikov function. Closed expressions for the Melnikov functions are just known for \(M_1\) and for \(M_2\) (when \(M_1=0\)). The main result is to give a constructive recursive procedure to obtain a function, called \(V_0(h),\) such that: (i) coincides essentially with \(M_1(h)\) when \(M_1(h)\not\equiv 0\), (ii) coincides essentially with \(M_2(h)\) when \(M_1(h)\equiv 0\) and \(M_2(h)\not\equiv 0\), (iii) in general, the set of zeros of \(V_0(h)\) contains all real zeros of the first nonidentical zero Melnikov function. The proof of the above result is mainly based on a previous theorem of the same authors [Nonlinearity 9, No.~2, 501-516 (1996; Zbl 0886.58087)] which asserts the following: Let \((P,Q)\) be a \(C^1\) vector field defined in an open subset \(U\) of \({\mathbb{R}}^2.\) Let \(V=V(x,y)\) be a \(C^1\) solution to the linear partial differential equation \(P\partial V/\partial x+ Q\partial V/\partial y=(\partial P/\partial x+\partial Q/\partial y)V.\) If \(\gamma\) is a limit cycle of \((P,Q),\) then \(\gamma\) is contained in \(\{(x,y)\in U : V(x,y)=0 \}.\) In fact, the \(V_0(h)\) that appears above is the constant term of the \(\varepsilon\) expansion of \(V(x,y,\varepsilon)=V_0(x,y)+\varepsilon V_1(x,y)+\varepsilon^2 V_2(x,y)+\cdots,\) where this function is a solution to the above partial differential equation, when \(P=-\partial H(x,y)/\partial y+f(x,y,\varepsilon)\) and \(Q=\partial H(x,y)/\partial x+g(x,y,\varepsilon)\).