Active Learning and Genetic Search for Discrete Abelian Flavor Model Building

The sequential breaking of discrete flavor symmetries provides a theoretical framework that can explain quark mass and mixing angle hierarchies. Each model is identified by its field content, charge assignments and symmetry breaking pattern, leading to an enormous space of possible models; this motivates the investigation of computational approaches for identifying promising candidates. In this work, we focus on quark flavor models with the flavor symmetry group $G_F=(\mathbb{Z}_2)^p$, where $p$ is a positive integer; operators involving flavon fields which spontaneously break the flavor symmetry determine the size of the Yukawa matrix entries. We consider two search strategies: an active learning search employing a random forest regressor and a genetic search. The generation of candidate models is automated and models are evaluated using a loss function that takes into account the logarithmic discrepancy between the model prediction and target observable values as well as penalties for model complexity. We find that both our active learning and genetic algorithms significantly outperform random model generation, which we use as a control; our active learning approach identifies lower-loss models faster (i.e., with fewer exact model evaluations) than our genetic search, though each ultimately finds successful models. Our results support the effectiveness of active learning and evolutionary optimization as approaches to automate the search for promising models based on discrete flavor symmetries.

Authors

Publication Details

Journal
arXiv (Cornell University)
Published
2026-11-18
DOI
https://doi.org/10.48550/arxiv.2611.24537
Primary Topic
Particle physics theoretical and experimental studies
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Active Learning and Genetic Search for Discrete Abelian Flavor Model Building

Christopher D. Carone, Selin Aşmanoğlu
arXiv (Cornell University)
Particle physics theoretical and experimental studies
preprint

Active Learning and Genetic Search for Discrete Abelian Flavor Model Building

Christopher D. Carone, Selin Aşmanoğlu
preprint en

Abstract

The sequential breaking of discrete flavor symmetries provides a theoretical framework that can explain quark mass and mixing angle hierarchies. Each model is identified by its field content, charge assignments and symmetry breaking pattern, leading to an enormous space of possible models; this motivates the investigation of computational approaches for identifying promising candidates. In this work, we focus on quark flavor models with the flavor symmetry group $G_F=(\mathbb{Z}_2)^p$, where $p$ is a positive integer; operators involving flavon fields which spontaneously break the flavor symmetry determine the size of the Yukawa matrix entries. We consider two search strategies: an active learning search employing a random forest regressor and a genetic search. The generation of candidate models is automated and models are evaluated using a loss function that takes into account the logarithmic discrepancy between the model prediction and target observable values as well as penalties for model complexity. We find that both our active learning and genetic algorithms significantly outperform random model generation, which we use as a control; our active learning approach identifies lower-loss models faster (i.e., with fewer exact model evaluations) than our genetic search, though each ultimately finds successful models. Our results support the effectiveness of active learning and evolutionary optimization as approaches to automate the search for promising models based on discrete flavor symmetries.

arXiv (Cornell University)
Particle physics theoretical and experimental studies
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon