Renato Augusto Tavares
English

A calibrated computational study of the accelerated 5x + 1 predecessor-count exponent

Enumera exatamente a árvore reversa do mapa 5x+1 e calibra o estimador do expoente de contagem com seis construções sintéticas de alvo conhecido. A leitura calibrada é compatível com a previsão de Kontorovich–Lagarias; trata-se de evidência computacional sob hipóteses declaradas, não de demonstração.

Acesso

Resumo

Two stochastic models predict different counting exponents for the 5x+1 reverse tree: about 0.650919 from Kontorovich and Lagarias, an exponent whose existence is their Conjecture 7.2, and about 0.678 from Volkov, for an expectation and conjecturally for the count of non-escaping integers. The gap is 0.027, and Kontorovich and Lagarias write that Volkov’s own data seems insufficient to discriminate between the values. We enumerate the tree exactly. A fixed-window estimator reads 0.639 and excludes neither value, but on an unconstrained construction with a known target it reads 0.038 low, more than the gap, so a raw reading cannot decide the question. We calibrate it with six matched constructions that keep the tree’s sibling structure, exactly or in relaxed form, four assigned 0.650919 and two assigned 0.678. At the deepest matched checkpoint decade the arithmetic tree reads 0.64926. That lies inside the span of the four constructions assigned the Kontorovich-Lagarias value, and 0.02432 below the lower of the two assigned Volkov’s, with 95% bootstrap interval [0.0208, 0.0273]. The reading assumes that each construction’s realized exponent equals its target and that the calibrated bias transfers to the deterministic tree. Neither is proved here, and three further inputs are open problems.

Resumo no idioma original da publicação. Uma apresentação em português, com os resultados classificados por situação, está em O expoente do mapa 5x+1.

Palavras-chave

5x+1 problem, predecessor trees, counting exponents, branching random walks, Collatz-type maps.

Como citar

Tavares, R. A. (2026). A calibrated computational study of the accelerated 5x + 1 predecessor-count exponent. Preprint. doi:10.2139/ssrn.7445805.

@misc{Tavares2026FiveXPlusOne,
  author = {Tavares, Renato Augusto},
  title = {{A calibrated computational study of the accelerated 5x + 1 predecessor-count exponent}},
  year = {2026},
  month = sep,
  howpublished = {SSRN},
  doi = {10.2139/ssrn.7445805},
  note = {Preprint}
}