透過您的圖書館登入
IP:3.135.198.49
  • 學位論文

以平行加速之巢式抽樣模擬神岡重力波探測器在未來重力波網路之貢獻

Observing scenarios of KAGRA in future gravitational-wave networks using GPU-parallelized nested sampling

指導教授 : 陳凱風
共同指導教授 : 灰野禎一(Sadakazu Haino)

摘要


隨著重力波探測器精密度的提升,未來觀測到重力波事件的頻率將逐年增 加。由於參數估計相當耗時,不足以應付如此大量的事件數,參數估計的加速方 法正在被研發中。本研究的第一部份致力於描述 GPE+,一個利用圖形運算器平行 加速之重力波參數估計程式。GPE+ 運用了兩種加速方式,重力波模型與概似函數 之計算、以及巢式抽樣之平行化。其中,模型與概似函數之加速已被利用在 GPE 程式,並在一個圖形處理器上達到了比 LALInference 在一個中央處理器上快一 百倍的速度。本研究開發了新的演算法以平行化 GPE 內的巢式抽樣法,由此設計 出一個新的重力波參數估計程式:GPE+。GPE+ 表現出比 GPE 快二至四倍的速度, 並且輸出與 GPE 一致的參數估計結果,代表著 GPE+ 能達到比 LALInference 快 二百至四百倍的速度。GPE+ 的高速平行運算將有利於模擬未來重力波探測器之貢 獻,以及電磁波對硬體之觀測。 本研究第二部分運用了 GPE+ 以模擬大量重力波數據來預測神岡重力波探測 器(KAGRA)於不同靈敏度下(KAGRA+:180 百萬秒差距;hyper KAGRA:500 百萬秒差距)在全球重力波網路的貢獻。本研究結果發現 KAGRA 最大的貢獻在 於增加事件定位的準確度;其中,運用 KAGRA+ 能將精準度提升二倍,而 hyper KAGRA 則能將精準度提升四倍。至於距離以及傾角,KAGRA 則稍有貢獻,但 質量以及自旋卻僅有微小貢獻。事件定位的提升將有利於電磁波對應體的觀測, 而距離量測的準確度提升則能增強重力波作為標準警笛之能力。此結果顯示將 KAGRA 加入全球重力波網路能提升未來哈伯常數之觀測。

並列摘要


With more sensitive gravitational-wave detectors under construction, the detection rate of gravitational waves from compact binary coalescence sources will continue to increase in the near future. This era of multi-detection gravitational wave astronomy presents a challenge for gravitational wave parameter estimation, a time-consuming process even in the single-event case. Thus, an acceleration method for parameter estimation is in demand. The first half of this thesis presents GPE+, a GPU-accelerated parameter estimation program for gravitational waves. The GPU parallelization methods implemented in GPE+ are twofold: (1) the waveform and likelihood calculations, (2) and the nested sampling algorithm. The waveform and likelihood accelerations have been employed in the code GPE, which demonstrated a ∼100 times speedup on one GPU compared with LALInference on one CPU. In this thesis, we parallelized the nested sampling algorithm in GPE by parallelizing the prior sampling portion, and designed a new program: GPE+. GPE+ demonstrates a 2-4 times speedup and produces consistent results compared to its predecessor, GPE, which makes GPE+ 200-400 times faster than LALInference. GPE+ offers the opportunity to perform large simulations to estimate observing scenarios for detector upgrades, and generate sky localization confidence areas in a short amount of time for electromagnetic follow-up of gravitational wave events. The second half of this thesis uses GPE+ to run thousands of simulations with future sensitivities of a gravitational-wave detector network. The simulations emphasize the effects of adding the KAGRA detector in the global network at different sensitivities, the KAGRA+ detector (180 Mpc) and the hyper KAGRA detector (500 Mpc). The results show that including the KAGRA detectors will have the most improvement in sky localization, with KAGRA+ providing a factor of two improvements and hyper KAGRA providing a factor of four improvements. Distance and inclination angle measurements show modest improvements, whereas the mass and spin measurements only exhibit minimal improvements. The sky localization improvement implies that adding KAGRA to the global detector network can enhance the electromagnetic counterpart identification, whereas the distance improvement can better the standard siren method of gravitational waves. Both improvements indicate that adding KAGRA can lead to better measurements of the Hubble constant.

參考文獻


1. Aasi, J. et al. Advanced LIGO. Classical and Quantum Gravity 32, 074001 (2015). 

2. Abbott, B. P. et al. Observation of Gravitational Waves from a Binary Black Hole Merger. Phys. Rev. Lett. 116, 061102 (2016). 

3. Abbott, B. P. et al. Tests of General Relativity with GW150914. Phys. Rev. Lett. 116, 221101 (2016). 

4. Abbott, B. P. et al. The Rate of Binary Black Hole Mergers Inferred from Advanced LIGO Observations Surrounding GW150914. The Astrophysical Journal 833, L1 (2016). 

5. Abbott, B. P. et al. Astrophysical Implications of the Binary Black-hole Merger GW150914. The Astrophysical Journal 818, L22 (2016). 


延伸閱讀