Collatz cycles with few descents (Q4486458)

Consider the function \(T_1\) defined on odd integers \(x\) by \(T_1(x)=2^{-k(x)}(3x+1)\), where the integer \(k(x)\geq 1\) is fixed by the condition that \(T_1(x)\) should be an odd integer. If \(T\) denotes the well-known \(3n+1\) function, one has the equation \(T_1(x)=T^{k(x)}(x)\). An odd integer \(x\) is called \textit{descending} if \(k(x)\geq 2\) (or, equivalently, if \(|T_1(x)|\leq|x|\)). A cycle of \(T_1\) is called a \textit{Collatz cycle}, denoted by \(\Gamma=(x_1,\ldots,x_n)\) where the \(x_i\) are pairwise distinct odd integers. Denote the length of \(\Gamma\) by \(|\Gamma|=n\), and let \(\delta(\Gamma)\) be the number of descending elements of \(\Gamma\). The main result of the paper under review is Theorem 1.1: The number of Collatz cycles satisfying \(\delta(\Gamma)<2\log|\Gamma|\) is finite. An extension of this result to \(3x+d\) mappings, where \(d\) is a positive integer \(\equiv\pm 1\bmod 6\), is briefly sketched. The paper starts with a nice argument proving that, for fixed \(n\), the number of Collatz cycles with length \(n\) is finite. Then an algebraic reformulation is carried out resulting in a characterization of Collatz cycles with \(|\Gamma|=n\) and \(\delta(\Gamma)=\nu\). The remaining parts of the proof are sophisticated estimates; the central point is a result of A.~Baker and N.~Feldman asserting that there is an effectively computable constant \(C_0>0\) such that \(|k\log 2-n\log 3|>(\max(k,n))^{-C_0}\) for all \(k\geq 2\) and \(n\geq 1\).