双中子星系统的改进种群与核态方程联合推断
Improved Population and EOS Joint Inference for Binary Neutron Star Systems
- Rochester Institute of Technology（罗切斯特理工学院）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
本研究扩展RIFT算法（Hyperpipe）以联合推断双中子星系统的核态方程与种群参数，通过多种观测数据验证，发现更宽松的先验边界可显著改善超参数后验。
AI中文摘要：
中子星内部的极端环境为在高密度下探测核态方程提供了机会，同时也能研究这些恒星残骸的性质。历史上，核态方程和中子星质量分布是分别推断的，后者在研究中往往被简单地假定为固定不变，但核态方程对中子星质量的依赖性表明它们应该被联合推断。在本工作中，我们扩展了流行的RIFT算法的广义版本（称为Hyperpipe），使其能够与灵活的、用户提供的先验接口，从而促进核态方程和双中子星种群超参数的联合推断。我们通过将这一框架应用于若干现有观测的组合来展示其实用性，特别是包括大质量银河系脉冲星、双中子星系统、毫秒X射线脉冲星、引力波事件GW170817以及核对称能，在广泛采用的参数化EOS族和简单高斯种群模型的背景下。我们恢复了与先前对EOS和双中子星种群分析一致的参数，发现后者拟合为双变量正态分布，均值为$(\mu_1,\mu_2) = (1.39,1.27) M_{\odot}$，宽度为$\sigma = 0.08 M_{\odot}$。此外，我们在流程中实现了新颖的坐标变换，通过该变换我们发现该EOS族所使用的临时先验边界可能过于严格。我们展示了在放宽但仍物理的先验边界下的结果，注意到超参数后验有显著改善，而核态方程推断有适度差异。
英文摘要：
The extreme environment within neutron stars presents the opportunity to probe the nuclear equation of state at high densities while studying the properties of these stellar remnants. Historically, the equation of state and neutron star mass distribution have been inferred separately, with the latter often simply assumed to be fixed in studies of the former, but the dependence of the equation of state on neutron star mass indicates they should be inferred together. In this work, we extend the generalized version of the popular RIFT algorithm known as Hyperpipe to interface with flexible, user-provided priors that will facilitate joint inference of equation of state and binary neutron star population hyperparameters. We demonstrate this framework's utility via application to a combination of several existing observations, particularly including massive galactic pulsars, double neutron star systems, millisecond X-ray pulsars, the gravitational wave event GW170817 and the nuclear symmetry energy, within the context of a widely-adopted parametric EOS family and a simple Gaussian population model. We recover parameters consistent with previous analyses of the EOS and binary neutron star population, finding the latter to fit a bivariate normal distribution with mean $(μ_1,μ_2) = (1.39,1.27) M_{\odot}$ and width $σ= 0.08 M_{\odot}$. Furthermore, we implement novel coordinate transformations in our pipeline, with which we have discovered that the ad hoc prior boundaries used for this EOS family may be too restrictive. We present results with relaxed yet still physical prior boundaries, noting vastly improved hyperparameter posteriors and modestly different equation of state inferences.
