AI 中文总结
该研究利用数值相对论等模拟数据，构建了适用于非进动、进动及偏心双黑洞并合残余性质的统一解析模型，精度优于现有模型，可通过gwModels包公开供相关领域应用。
AI 中文摘要
我们利用约5000次数值相对论（NR）模拟（质量比q达128）和1200次黑洞微扰理论（BHPT）模拟（质量比q达1000），给出了准圆形双黑洞并合残余性质的全解析模型。模型gwModelRemS（用于最终质量、最终自旋、峰值光度和反冲速度）和gwModelRemP（用于最终质量、最终自旋和峰值光度）分别描述非进动和进动双黑洞。我们的方法结合了后牛顿理论、点粒子极限的解析见解与数据驱动拟合框架，该框架还包含经验证的AI智能体辅助优化。所得拟合模型在其有效域内优于现有解析模型，且精度可与数据驱动模型媲美。我们还构建了gwModelRemP_flow，这是一种针对进动双黑洞反冲速度的归一化流模型，已对平面内自旋取向边缘化，该模型额外利用了通过基于波形的数据增强程序构建的、超出可用NR和BHPT模拟范围的扩大低保真训练集。我们还纳入了简单的偏心扩展，即基于 eccentricity 和轨道异常的 leading-order 依赖。这些模型覆盖从等质量到极端质量比的所有区域，可通过gwModels包公开获取，用于引力波天文学、天体物理种群研究和宇宙学领域。
英文摘要
Using approximately $5000$ numerical-relativity (NR) simulations spanning mass ratios up to $q=128$ and $1200$ black-hole-perturbation-theory (BHPT) simulations extending to $q=1000$, we present fully analytic models for the remnant properties of quasi-circular binary black hole mergers. The models gwModelRemS (for final mass, final spin, peak luminosity, and recoil kick) and gwModelRemP (for final mass, final spin, and peak luminosity) describe nonprecessing and precessing binaries, respectively. Our approach combines analytic insights from post-Newtonian theory and point-particle limits with a data-driven fitting framework which also includes validation-guided AI-agent-assisted optimization. The resulting fits outperform existing analytic models and achieve accuracies comparable to data-driven models within their domains of validity. We also construct gwModelRemP\_flow, a normalizing-flow model for recoil kicks from precessing binaries, marginalized over the in-plane spin orientations. This model additionally utilizes an enlarged low-fidelity training set, beyond the available NR and BHPT simulations, constructed through a waveform-based data-augmentation procedure. We also include simple eccentric extensions through leading-order dependence on eccentricity and orbital anomaly. The models span the equal-mass to extreme-mass-ratio regimes and are publicly available through the gwModels package for applications in gravitational-wave astronomy, astrophysical population studies, and cosmology.
CommentsModels are available through https://github.com/tousifislam/gwModels/
