.
The semi-dwarf (sd1) allele was central to the Green Revolution, but its widespread use in US tropical japonica rice (Oryza sativa L.) may have introduced unintended effects through linkage drag. This study examined the genomic region surrounding sd1 using genome-wide association and haplotype analyses in a multiparent population (MP6/8) evaluated across environments and in a biparental population (MP-D) for validation. Genomic and phenotypic evidence indicated that an indica-derived haplotype upstream of sd1 persists in US tropical japonica and global indica semi-dwarf germplasm and is associated with reduced grain yield. Genome-wide association analysis confirmed sd1 as the primary determinant of reduced plant height and identified a quantitative trait locus at 36.69 Mb upstream of sd1 that was consistently associated with reduced grain yield (0.6-0.9 t ha-1) across populations and environments. Haplotype analysis identified an indica-derived Presidio haplotype spanning this upstream region was associated with significant yield reductions in both populations. Pedigree analysis traced this unfavorable haplotype to the Taichung Native 1 donor lineage, representing one historically deployed sd1-linked introgression class. Although its frequency has declined in modern cultivars, its persistence highlights the long-term impact of linkage drag following major trait introgression. These findings emphasize the value of marker-assisted selection and targeted recombination to separate favorable dwarfing alleles from linked yield penalties and sustain yield gains in US rice improvement programs.
Human migrationHuman migration is a significant global phenomenon that affects countries of origin, destination, and their surrounding regions. Despite its impact, migration has historically been a challenge to predict reliably. In this paper, we examine the feasibility of applying machine learningMachine learning to predict migration trends using historical data and various socioeconomic indicatorsSocioeconomic indicators. Two engineered indices were developed to assess the interdependencies of these factors. Among the machine learningMachine learning models tested for binary directional prediction, extreme gradient boosted trees (XGBoost)XGBoost demonstrated the highest accuracy. The XGBoost model achieved an average directional prediction accuracy of 93.32
Prior research has consistently shown that students’ SAT scores are influenced by factors beyond academic ability, including socioeconomic background and ethnicity. Using aggregated school-level data from Massachusetts and New York City (NYC), this study assesses the quantitative relationships between average SAT scores and school-level demographics, academic preparation, and funding to inform education policy and equity efforts. Three analytical methods, multiple linear regression, relaxed Least Absolute Shrinkage and Selection Operator (LASSO), and decision trees, were applied sequentially to capture the linear and nonlinear associations. Across all three methods, socioeconomic disadvantage exhibited the strongest and most robust correlation of lower SAT scores, with racial composition and academic preparation identified as secondary factors. Schools with high percentages of Black, Hispanic, and low-income students tend to have lower average scores than schools with high percentages of White, Asian, and well-off students. Moreover, schools with higher college attendance rates and greater funding tend to exhibit higher average SAT scores. These findings represent school-level correlations rather than causal effects and indicate that SAT score disparities are closely intertwined with broader structural inequities already embedded within the K–12 education system. Accordingly, while targeted SAT preparation initiatives may offer modest benefits, meaningful reductions in observed disparities likely require broader policy interventions to expand equitable access to educational opportunities.
The use of metrology and tribology methods for archaeological stone tool microwear analysis has provided opportunities to revisit unresolved issues, such as wear formation processes, the exclusivity of polishes derived from different worked materials, and damage to stone tool surfaces produced by post-depositional environments. In this paper, we provide a brief history of research on post-depositional damage, present a summary of stone tool microwear quantification, and review the development of current methods employed to mathematically characterize stone tool surfaces altered through natural and cultural processes of post-deposition. Through reviewing past work we provide thoughts on the next steps in method development for the mathematical characterization of post-depositional alteration on chipped stone tool surfaces. Ultimately, archaeologists studying use-related microwear using quantification of surface structure must contend with post-depositional wear, just as their visual microscopic microwear analysis colleagues have. One primary obstacle to the widespread adoption of quantitative methods for lithic microwear analysis is the ability to distinguish use-related microwear from microwear resulting from post-deposition, and this review article provides a critical overview of the state of the field.
Diffusion-based Text-to-Image (T2I) models have achieved impressive success in generating high-quality images from textual prompts. While large language models (LLMs) effectively leverage Direct Preference Optimization (DPO) for fine-tuning on human preference data without the need for reward models, diffusion models have not been extensively explored in this area. Current preference learning methods applied to T2I diffusion models immediately adapt existing techniques from LLMs. However, this direct adaptation introduces an estimated loss specific to T2I diffusion models. This estimation can potentially lead to suboptimal performance through our empirical results. In this work, we propose Direct Score Preference Optimization (DSPO), a novel algorithm that aligns the pretraining and fine-tuning objectives of diffusion models by leveraging score matching, the same objective used during pretraining. It introduces a new perspective on preference learning for diffusion models. Specifically, DSPO distills the score function of human-preferred image distributions into pretrained diffusion models, fine-tuning the model to generate outputs that align with human preferences. We theoretically show that DSPO shares the same optimization direction as reinforcement learning algorithms in diffusion models under certain conditions. Our experimental results demonstrate that DSPO outperforms preference learning baselines for T2I diffusion models in human preference evaluation tasks and enhances both visual appeal and prompt alignment of generated images.
