Reachability of maximal number of critical periods without independence (Q2202269)

In this paper, the authors consider the following system \[ \dot x=-y+M(x, y, \lambda), \quad \dot y=x+N(x, y, \lambda), \] where \(\lambda=(\lambda_1, \ldots, \lambda_n)\) and \[ M(x, y, \lambda), N(x, y, \lambda)=O(|x,y|^2). \] Suppose the above system has a center at the origin, then we can define the period function \[ P(r)=2\pi+\sum_{i=2}p_i(\lambda)r^i. \] A classic problem is to determine the number of critical periods, which are the zeros of \(P'(r)=0\) for all \(\lambda\). It is well known that if for \(p_2(\lambda')=p_4(\lambda')=\dots=p_{2k}(\lambda')=0\), but \(p_{2k+2}(\lambda')\not=0\) for some \(\lambda'\), then \(P'(r)\) has at most \(k\) zeros near \(r=0\) when \(|\lambda-\lambda'|\) is small. A natural problem is to ask whether we can find \(\lambda\) so that \(P'(r)\) has exact \(k\) zeros near \(r=0\). This is a difficult problem. Usually one need the condition that \(p_2, p_4, \dots, p_{2k+2}\) are independent of the parament \(\lambda\), which will imply that one can only find at most \(n-1\) critical periods for \(\lambda=(\lambda_1, \ldots, \lambda_n)\). The authors give a new method, which does not need the independence condition, so that they can find more critical periods than before.