Scalable data-analysis framework for long-duration gravitational waves from compact binaries using short Fourier transforms
基于短傅里叶变换的可扩展长时标致密双星引力波数据分析框架
We introduce a framework based on short Fourier transforms (SFTs) to analyze long-duration gravitational wave signals from compact binaries. Targeted systems include binary neutron stars observed by third-generation ground-based detectors and massive black hole binaries observed by the LISA space mission. In short, ours is an extremely fast, scalable, and parallelizable implementation of the gravitational wave inner product, a core operation of gravitational wave matched filtering. By operating on disjoint data segments, SFTs allow for efficient handling of noise nonstationarities, data gaps, and detector-induced signal modulations. We present a pilot application to early warning problems in both ground- and space-based next-generation detectors. Overall, SFTs reduce the computing cost of evaluating an inner product by three to five orders of magnitude, depending on the specific application, with respect to a nonoptimized approach. We release public tools to operate using the SFT framework, including a vectorized and hardware-accelerated reimplementation of a time-domain waveform. The inner product is the key building block of all gravitational wave data treatments; by speeding up this low-level element so massively, SFTs provide an extremely promising solution for current and future gravitational wave data-analysis problems.
展开 ▾借鉴连续波搜索中的 SFT 分解策略,将内积转化为对各独立时间段的频域加权求和,无需近似探测器响应函数;结合 GPU 向量化后整体计算成本降低 3–5 个数量级,已在 BNS 和 LISA 大质量双黑洞的早期预警模拟中验证可行性。
连续引力波搜索领域早已利用 SFT 处理长期数据的非平稳性和间断(Williams+ 2000、Prix+ 2011、Prix 2019),近期更展示了相干搜索的 GPU 加速效果(Tenorio+ 2024)。本文将 SFT 框架推广到双星长时标旋近波形,用 Fresnel 核实现内积的高效近似,一举克服了传统频域方法在噪声变化、数据缺口和探测器调制等方面面临的难题,并将波形采样率压缩 3–4 个数量级。展望未来,该方案可与多频带(Vinciguerra+ 2017)和相对分箱(Zackay+ 2018)等加速策略互补,以及适配如 IMRPhenomT(Estelles+ 2021)等更复杂的波形模型,有望成为第三代地面与空间引力波探测器(ET、CE、LISA)实时分析和全球拟合的关键基础模块。
预印本 2025-02-17 · 刊出 2025-05-14 · 收录 2026-10-12