Searcharxiv⌕ Search

arXiv subjects

Henrique Oliveira

Publications and source records attributed to Henrique Oliveira.

4 recordsLinked to original sources

F3-AGI v4.0: Bioenergetically-Gated Active Inference, Metaplastic Preconditioning, and Residual Program Synthesis for Sample-Efficient Abstract Reasoning

Few-shot abstract reasoning on ARC-AGI requires inducing algorithmic transformations from 2 to 5 demonstration pairs without task-specific pretraining. Because few-shot demonstrations under-constrain the hypothesis space, program synthesis suffers from the Rashomon effect (predictive multiplicity): heuristic search finds spurious programs with low training error that fail on test inputs, while unconstrained library growth degrades latency. We introduce F3-AGI v4.0, a CPU-native neuro-symbolic and active inference architecture comprising three mechanisms: (1) F3-Metabolic-Optim, combining mistake-gated plasticity, Fisher-EMA preconditioning, and virtual ATP gating; (2) Adaptive Complexity-Guided Program Synthesis (AACR) with a Cycle-3 ResidualPatchSynthesizer and smallest-pair early rejection; and (3) a 512D L2-normalized random projection driving Expected Free Energy (EFE) exploration with dead-signature pruning. Across three empirical studies: (i) on a synthetic 5-task class-incremental stream without replay, F3-Metabolic-Optim achieves 91.67% final accuracy with 9.17% forgetting, skipping 93.47% of updates, cutting L1 plasticity cost by 65.7%, and running 2.19x faster than EWC; (ii) on the 400-task ARC-AGI-1 public evaluation set (offline CPU, zero LLM calls), F3-AGI v4.0 solves 57/400 tasks (14.25% pass@2; 56/400 pass@1) with a mean latency of 1.62 s/task, decomposed into 19/50 (38.00%, 1.19 s/task, a 42.5x paired speedup over Horizon 2) on the 50-task calibration slice and 38/350 (10.86%) on the 350 strictly uninspected hold-out tasks (with 7 autonomous Cycle-3 patch solutions); and (iii) in a local 5-game (15-level) ARC-AGI-3-inspired interactive suite, our agent completes 15/15 levels with 89.44% level-weighted RHAE (90.72% unweighted).

cs.AI↗

Evolutionary Entropy Shapes Reproductive Lifespan in Age-Structured Populations

Evolutionary entropy measures the temporal organization of reproductive contributions along the life cycle of an age-structured population. We develop a mathematical and empirical framework showing that, in iteroparous animal populations represented by Leslie-type demographic matrices, reproductive windows are frequently organized near the age classes selected by entropy maximization. Evolutionary entropy complements the classical net reproductive number and asymptotic growth rate: whereas these measure lifetime replacement and growth, entropy measures the temporal dispersion of the growth-adjusted reproductive distribution. Our central result is a reduction principle: under Euler--Lotka normalization, evolutionary entropy and generation time are invariant under multiplicative rescaling of survivorship and fertility on the reproductive interval. The relevant entropy is determined not by absolute survivorship, fertility, or juvenile mortality, but by the normalized post-maturity reproductive distribution. We derive explicit entropy functionals for finite and open-group Leslie models, including geometric reproductive tails. For the geometric regime, governed by we prove a sharp critical threshold separating populations with a unique finite entropy-maximizing endpoint from those whose entropy increases toward an asymptotic value in terms solely of the age at first reproduction. The theory is tested on 130 animal species. Entropy-derived predictions, computed from the demographic matrices alone, are compared with independent life-history variables. Predicted and observed reproductive medians coincide exactly for a majority of species, over 90% are predicted within three reproductive classes, and associations remain strong after phylogenetic correction. These results identify a quantitative regularity across taxa, with geometric reproductive distributions playing a central role.

q-bio.PE↗

Population Dynamics with Infinite Leslie Matrices

Infinite Leslie matrices, introduced by Demetrius forty years ago are mathematical models of age-structured populations defined by a countable infinite number of age classes. This article is concerned with determining solutions of the discrete dynamical system in finite time. We address this problem by appealing to the concept of kneading matrices and kneading determinants. Our analysis is applicable not only to populations models, but to models of self-reproducing machines and self-reproducing computer programs. The dynamics of theses systems can also be described in terms of infinite Leslie matrices.uation.

q-bio.PE↗

Products of 2X2 matrices related to non autonomous Fibonacci difference equations

A technique to compute arbitrary products of a class of Fibonacci $2\times2$ square matrices is proved in this work. General explicit solutions for non autonomous Fibonacci difference equations are obtained from these products. In the periodic non autonomous Fibonacci difference equations the monodromy matrix, the Floquet multipliers and the Binet's formulas are obtained. In the periodic case explicit solutions are obtained and the solutions are analyzed.

math.DS↗