The exact bounds on the number of zeros of complete hyperelliptic integrals of the first kind (Q695066)

The purpose of the paper is to study the number of zeros of the integral \[ J(h)=\oint_{\Gamma(h)}\frac{\alpha_0 +\alpha_1x}{y}dx \] in the interval \(I=(h_1, h_2)\), where \(\{\Gamma(h)\}\) forms a continuous family of ovals contained in the level sets \(\{H(x,y)=h\}\). The Hamiltonian is \(H=\frac12y^2+U(x)\) with a polynomial \(U\) of degree five. This problem was initially investigated by \textit{L. Gavrilov} and \textit{I. D. Iliev} [Trans. Am. Math. Soc. 356, No. 3, 1185--1207 (2004; Zbl 1043.34031)]. They introduced a normal form of \(H\) in which the derivative of the potential factorizes into \(U'(x)=x(x-\lambda)(x-\mu)(x-1)\) with parameters \(\lambda,\mu\) that are either complex conjugated (complex case) or \(0\leq\mu\leq\lambda\leq 1\) (real case) and established that \(J(h)\) can have two zeros, but for some families of ovals (called there exceptional) \(J(h)\) has at most one zero. Here, the authors take three degenerate Hamiltonians corresponding to parameters \((\lambda,\mu)=(0,0), (1,1), (1,\frac25)\) from the bifurcation diagram which are not exceptional and prove for them that \(J(h)\) has at most one zero in \(I\). The proof uses some results by \textit{M. Grau} et al. [Trans. Am. Math. Soc. 363, No. 1, 109--129 (2011; Zbl 1217.34052)], and \textit{F. Mañosas} and \textit{J. Villadelprat} [J. Differ. Equations 251, No. 6, 1656--1669 (2011; Zbl 1237.34039)] to establish first that \(J(h)\) has no more than two zeros. Then, by using the behaviour of \(J(h)\) at the endpoints of \(I\), the number of zeros is reduced to one. As known, this kind of proof proceeds through huge formulas even when the Hamiltonian contains no parameters.