Bounding the number of zeros of certain abelian integrals (Q635439): Difference between revisions

Take real-valued analytic functions \(A(x)\), \(B(x)\) and \(f_i(x)\), \(0\leq i\leq n-1\), where \(x\) belongs to some open interval on the real line. Consider the interval \((0,h_0)\), where the level sets \(H=h\) of the Hamiltonian function \(H(x,y)=A(x)+B(x)y^{2m}\), \(m\) natural number, \(H(0,0)=0\), contain a continuous family of ovals \(\gamma_h\) around the critical point (a nondegenerate center) at the origin. The authors estimate the number of zeros of linear combinations of the form \[ \alpha_0I_0(h)+\alpha_1I_1(h)+\ldots+\alpha_{n-1}I_{n-1}(h),\;\; h\in(0,h_0),\tag{1} \] where \(I_i(h) =\int_{\gamma_h}f_i(x)y^{2s-1}dx,\; i=0,1,\ldots,n-1,\; n,s\in\mathbb{N}.\) For this purpose, they define explicitly (through \(A\), \(B\) and \(f_i\)) certain functions \(\ell_i(x)\) and corresponding Wronskians. The main result of the paper is as follows. Assume that \(W[\ell_0,\ldots,\ell_i]\) are nonvanishing for \(i<n-1\) and \(W[\ell_0,\ldots,\ell_{n-1}]\) has \(k\) zeros. Take \(s>m(n+k-2)\). Then the linear combinations of functions (1) form a Chebyshev system with accuracy \(k\). That is, any nontrivial function (1) has at most \(n+k-1\) zeros in \((0, h_0)\). As an application, the authors give a shorter proof of the main result in a paper by \textit{F. Dumortier, C. Li} and \textit{Z. Zhang} [J. Differ. Equations 139, No.1, 146--193 (1997; Zbl 0883.34035)].