Vitale Salvatore, Gerosa Davide, Haster Carl-Johan, Chatziioannou Katerina, Zimmerman Aaron
LIGO, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, Massachusetts 02139, USA.
TAPIR 350-17, California Institute of Technology, 1200 East California Boulevard, Pasadena, California 91125, USA.
Phys Rev Lett. 2017 Dec 22;119(25):251103. doi: 10.1103/PhysRevLett.119.251103. Epub 2017 Dec 20.
In a regime where data are only mildly informative, prior choices can play a significant role in Bayesian statistical inference, potentially affecting the inferred physics. We show this is indeed the case for some of the parameters inferred from current gravitational-wave measurements of binary black hole coalescences. We reanalyze the first detections performed by the twin LIGO interferometers using alternative (and astrophysically motivated) prior assumptions. We find different prior distributions can introduce deviations in the resulting posteriors that impact the physical interpretation of these systems. For instance, (i) limits on the 90% credible interval on the effective black hole spin χ_{eff} are subject to variations of ∼10% if a prior with black hole spins mostly aligned to the binary's angular momentum is considered instead of the standard choice of isotropic spin directions, and (ii) under priors motivated by the initial stellar mass function, we infer tighter constraints on the black hole masses, and in particular, we find no support for any of the inferred masses within the putative mass gap M≲5 M_{⊙}.
在数据信息含量仅为中等程度的情况下，先验选择在贝叶斯统计推断中可能发挥重要作用，这有可能影响所推断出的物理特性。我们表明，从当前双黑洞并合的引力波测量中推断出的一些参数确实如此。我们使用替代性（且由天体物理学激发）的先验假设，重新分析了孪生LIGO干涉仪进行的首次探测。我们发现，不同的先验分布会在最终的后验分布中引入偏差，从而影响对这些系统的物理解释。例如，（i）如果考虑黑洞自旋大多与双星角动量对齐的先验，而不是各向同性自旋方向的标准选择，有效黑洞自旋χ_{eff}的90%可信区间的限制会有大约10%的变化；（ii）在由初始恒星质量函数激发的先验条件下，我们推断出对黑洞质量更严格的限制，特别是，我们发现在假定的质量间隙M≲5 M_{⊙}内，所推断出的任何质量都没有得到支持。
