Preperiodic points for quadratic polynomials with small cycles over quadratic fields (Q1650203)

Given a polynomial \(f\) over a number field \(K\), one can construct a finite directed graph \(G(f,K)\) whose vertices are the \(K\)-rational preperiodic points of \(f\), with an edge \(\alpha\to\beta\) if \(f(\alpha)=\beta\). The dynamical uniform boundedness conjecture of \textit{P. Morton} and \textit{J. H. Silverman} [Int. Math. Res. Not. 1994, No. 2, 97--109 (1994; Zbl 0819.11045)] -- the dynamical analogue of the (now proved) strong uniform boundedness conjecture for elliptic curves -- suggests that, as one ranges over all fields and all polynomials of given degrees \(n\geq 1\) and \(d\geq 2\), respectively, one finds only finitely many such graphs, up to isomorphism. Poonen has given a complete classification of all directed graphs for quadratic polynomials over the rational field (\(n=1,d=2\)), under the assumption that the polynomial does not admit rational points of period greater than 3. (Such an assumption is believed to always be satisfied.) The scene is set by a conjecture, formulated by the author, that any quadratic polynomial over a quadratic extension of \(\mathbb{Q}\) has no more that 15 preperiodic points, and that the associated graph is one of a collection of 46 graphs, which are listed explicitly. The main result of this paper (which extends [\textit{J. R. Doyle} et al., New York J. Math. 20, 507--605 (2014; Zbl 1391.37082)]) constitutes substantial progress toward the proof of this conjecture, under a Poonen-like assumption on periods. The author shows that if a quadratic polynomial over a quadratic field does not have rational points of period greater than four, and if its graph \(G\) is not isomorphic to any of the graphs mentioned above, then \(G\) must contain one subgraph chosen from two small lists. This statement is refined by additional constraints on the possible subgraphs.