Symmetry

Symmetry is where understanding begins, and where understanding deepens when ideal assumptions break [...]

Forbush decreases (FDs) are transient reductions in galactic cosmic ray intensity near Earth, commonly associated with solar-driven interplanetary disturbances. Their measured properties can depend on the observing station and the procedures used to identify and characterise individual events. We analyse long-term hourly neutron monitor observations from Oulu, Lomnický štít, Newark, and Dourbes using a common detection and empirical profile-fitting framework. Candidate FDs are identified from relative count-rate reductions over 12–48 h intervals and manually verified, while duplicate, complex, and stepwise events unsuitable for the adopted single-component model are excluded. Each retained event is fitted with a modified skew-normal profile to determine its amplitude and characteristic times. The resulting catalogues contain 157, 115, 163, and 240 events for Oulu, Lomnický štít, Newark, and Dourbes, respectively, including 49 events identified at all four stations. Median amplitudes are broadly similar among the datasets, whereas duration distributions show stronger station-to-station differences. These differences are reduced, but not eliminated, for the matched-event subset. The amplitude–duration relationship is weak and station-dependent, while onset-to-minimum time correlates positively with total duration but only weakly with the post-minimum recovery duration. The results show that station response and operational parameter definitions must be considered when comparing FD properties derived from different neutron monitor observations.

Large language models (LLMs) are increasingly deployed in second-language (L2) writing assessment. However, substantial disparities persist in how different architectures operationalize shared scoring criteria. This study investigates model-specific assessment asymmetry across ChatGPT-4o, Gemini-2.5 Pro, and DeepSeek-V3, and evaluates whether a calibrated multi-LLM consensus framework can mitigate these scoring discrepancies. Forty argumentative essays composed by Chinese EFL learners were independently evaluated by the three models using a standardized prompt and an analytic rubric encompassing task relevance, clarity of ideas, coherence, and accuracy, benchmarked against human mean scores. Individual models and ensemble configurations were systematically validated across 100 iterations of five-fold nested cross-validation. Cross-model comparisons revealed that inter-model divergence varied significantly by dimension, exhibiting pronounced asymmetry in holistic ratings and task relevance compared to clarity of ideas, coherence, and accuracy. The full consensus framework—integrating model-specific calibration, error-based weighting, and meta-calibration—yielded an RMSE of 1.160 and an MAE of 0.889, relative to 1.144 and 0.874 for unweighted averaging. The paired RMSE difference of 0.016 (95% case-cluster bootstrap CI [−0.114, 0.149]) indicated no statistically significant advantage in overall predictive accuracy over equal weighting. Nevertheless, the calibrated framework achieved superior bias mitigation, evidenced by a reduced absolute mean bias and a narrower median-split residual gap. Supplementary analyses further delineated direct inter-model discrepancies and proficiency-dependent residual bias. These findings demonstrate that model-specific evaluative asymmetries transcend simple mean-level shifts, and that while calibration-driven consensus enhances bias regulation across score bands, this error control does not automatically translate into improved global predictive precision.