Open Access
Table D.1.
List of symbols and key words in the two-level pipeline of H1L1 data analysis probing the merger sequence Eq. 2 subject to constraints C1, 2 evaluated by PDF(ts) produced by butterfly matched filtering and χ-image analysis, augmented by time-slide analysis, implemented by load balanced heterogeneous parallel computing.
| Symbol. | Definition | Comment |
|---|---|---|
| GW170817-GRB170817A | §1-2 | |
| tm[s] | Merger time GW170817 | Fig. 1 |
| tg[s] | Gap in GW170817-GRB18017A | Fig. 1 |
| ℰEM | EM output | Eq. 6 |
| ℰGW | GW output | Eq. 4 |
| POST-MERGER EMISSION | §2-4 | |
| ts[s] | Start time | Eq. 7 |
| τs[s] | Time-scale of decay in the time-frequency domain | Eq. 7 |
| fs[s] | Initial frequency | Eq. 7 |
| f0[s] | Late-time frequency | Eq. 7 |
| BUTTERFLY MATCHED FILTERING | §3-4 | |
| τ | Duration of templates in butterfly matched filtering | Fig. 4 |
| ρ | Matched filtering output | Normalized in post-callback to clFFT |
| κ | Threshold of ρ output to host | Applied in post-callback to clFFT |
| SPECTROGRAMS | §3-4 | |
| (H1,L1)-spectrogram | Spectrograms merged by frequency coincidences | §3, Figs. 2-3 |
| Δt | Time-slide applied to H1-L1 | control parameter |
| S0 | Foreground | Δt = 0 |
| S1 | Extended foreground | |Δt| < τ |
| S2 | Background | |Δ| > τ |
| χ | H1L1-correlation in (H1,L1)-spectrograms along a track | Eq.7 |
| η | Cluster size | Eq. 13, 0-100% |
| H1,L1-spectrogram | Individual detector spectrograms | §4, Fig. 5 |
| [ts] | ts, H1 − ts, L1 | Eq. 17, Fig. 5 |
| [τs] | τs, H1 − τs, L1 | Fig. 5 |
| χ-IMAGE ANALYSIS | §1-4 | |
| χ | Correlation along tracks in (merged) spectrograms | Eq. 7 |
 |
χ(ts, τs) maximal over fs, f0 | |
 |
Location maximum  |
|
| Statistics | §3-4 | |
| Probability density function | §1 | |
| PFA | Probability of false alarm | §1 |
| T | Duration of data analyzed | Table I |
| FAR | False Alarm Rate | Table I |
| p1 | PFA1 | C1, Eq. 1, Appendix A |
| p2 | PFA2 | C2, Eq. 17 |
| COMPUTING | ||
| Heterogeneous | Mixed CPU-GPU | Fig. D.1 |
| OpenCL | Open Compute Language | AMD (2022), Khronos (2022) |
| Local Memory | Local to a Compute Unit, shared within a work-group | small, low latency |
| Global Memory | Visible to all work-groups, allocatable by host via PCIe | large, high latency |
| Throughput−1 | Wall clock time[s]/(Data length[s]×Template bank size[M]) | Eq. D.2 |
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.