Isabella Yin | Searcharxiv

arXiv2026

A High-Precision Numerical Framework for Time-Varying Solar Neutrino Flux with Full Earth Matter Oscillation Corrections for Global Underground Laboratories

Solar neutrinos have been studied for over half a century to test both the Standard Solar Model and the electroweak sector of the Standard Model of particle physics. Contemporary experiments are now entering an era of high-precision measurements, demanding corresponding theoretical predictions with sub-percent accuracy to enable meaningful comparison. In this paper, we identify and analyze the essential physical and computational components required to compute solar neutrino fluxes with high fidelity, and present a unified, computationally efficient framework. This framework incorporates: (i) the time-varying Earth-Sun distance; (ii) Earth matter effects modeled using both one-dimensional (1D) and three-dimensional (3D) Earth electron-density profiles; and (iii) a fast, Strang-splitting-based implementation of the Mikheyev-Smirnov-Wolfenstein (MSW) neutrino propagation formalism, enabling rapid, large-scale scans over neutrino trajectories and energy grids. We deliver site-specific predictions for the China Jinping Underground Laboratory (CJPL) and other underground laboratories actively engaged in solar neutrino programs.

hep-ph↗

arXiv2026

Code over Words: Overcoming Semantic Inertia via Code-Grounded Reasoning

LLMs struggle with Semantic Inertia: the inability to inhibit pre-trained priors (e.g., "Lava is Dangerous") when dynamic, in-context rules contradict them. We probe this phenomenon using Baba Is You, where physical laws are mutable text rules, enabling precise evaluation of models' ability to override learned priors when rules change. We quantatively observe that larger models can exhibit inverse scaling: they perform worse than smaller models when natural language reasoning requires suppressing pre-trained associations (e.g., accepting "Lava is Safe"). Our analysis attributes this to natural language encoding, which entangles descriptive semantics and logical rules, leading to persistent hallucinations of familiar physics despite explicit contradictory rules. Here we show that representing dynamics as executable code, rather than descriptive text, reverses this trend and enables effective prior inhibition. We introduce Code-Grounded Vistas (LCV), which fine-tunes models on counterfactual pairs and identifies states with contradictory rules, thereby forcing attention to logical constraints rather than visual semantics. This training-time approach outperforms expensive inference-time search methods in both efficiency and accuracy. Our results demonstrate that representation fundamentally determines whether scaling improves or impairs contextual reasoning. This challenges the assumption that larger models are universally better, with implications for domains that require dynamic overriding of learned priors.

cs.CL↗