Pith.

Instrumentation and Methods for Astrophysics · arXiv:2611.17229 v1 · submitted 12 Nov 2026 · 41 references

Peak Performance in Matched Filtering: Fused Correlation and Statistically Controlled Screening on CPUs and GPUs

Alexander H. Nitz, Keisi Kacanja, Collin D. Capano

The paper says

“Batched matched filtering and cross-correlation over large template banks are a primary computational cost in gravitational-wave astronomy, seismology, and radio transient searches.”

more of the abstract

Standard pipelines pair a general-purpose inverse Fast Fourier Transform (FFT) with separate multiplication and peak-finding kernels, writing intermediate spectra and full correlation series to memory for every template. We fuse the conjugate product into the first inverse-transform butterflies and the peak reduction into the last so only candidate peaks leave processor registers, and tile batches in 2D so K2 correlations reuse 2K cached spectra. This shifts filtering from a DRAM-bandwidth-bound to an on-chip cache-bandwidth-bound regime. In some situations, the signal-to-noise ratio (SNR) is concentrated toward lower frequencies while higher frequencies dominate the transform length. We develop a hierarchical screening method that exploits this via a sequence of short low-frequency transforms before evaluating the full band, deriving stage thresholds from the spectral power profile and matched-filter noise autocovariance to target a specified false-dismissal probability. We demonstrate these methods in the open-source matchedfilter library across eight CPU architectures and two integrated GPUs (AMD Radeon 8060S via Vulkan and Apple M2 via Metal). On an AMD Ryzen AI Max+ 395 at N = 4096, fused peak-only filtering is 1.4 × faster than a standalone AOCL-FFTW inverse transform on one CPU core and 6.2 × faster than rocFFT on its Radeon 8060S GPU (3.1 × over MLX FFT on the M2 GPU). For a fiducial gravitational-wave search (ρth = 5.5, 10−2 false-dismissal target), hierarchical screening yields a 26 × speedup over AOCL-FFTW on CPU, 21 × over rocFFT on the 8060S GPU, and 19 × over MLX FFT on the M2 GPU.

from the abstract

Pith says

Machine review · deepseek-v4-flash · 13 Nov 2026 · what was read was not recorded · not checked by a person

It can be wrong. Challenge this review or reply to it.

The passes disagreed. Reader: holds up. Skeptic: agrees with the reader. Referee: major revision.

4 major objections · 7 minor

How often objections from reviews like this one are right: not yet measured

Are you an author? What you can do.

Not checked: figures, code, what the 41 references say.

The paper says, at §4.3, Fig. 2(c), sentence b95

The review did not quote this passage.

Pith says: Objection 1 of 4 · major

The heterogeneous-bank stress test (right side of Fig. 2(c)) shows that a single reference envelope can produce false-dismissal rates far exceeding the target (∼27% bank-wide for an edge reference at α=10^{-2}). The authors propose partitioning the bank into reference groups ordered by spectral slope, but provide no algorithm or tolerance for determining when a real bank is 'similar enough' to use a single reference.

more of objection 1

This is a load-bearing limitation because the speedups claimed for hierarchical screening (26×, 21×, 19×) and the closing guarantee in b144 ('without sacrificing signal recovery beyond a user-specified false-dismissal rate') both depend on the FDR bound holding in deployment. The paper should either specify a concrete procedure for grouping templates (e.g., a distance metric on q[k] with a tolerance bound) or state explicitly that the method is only validated for banks where a single reference profile is representative, and that users must validate FDR for their specific bank.

The paper says, at §4.1, Eqs. (7)–(9), and §4.2

The review did not quote this passage.

A classifier read this sentence of the paper as acknowledging the concern (probability 0.68):

“The rejection rates hold for noise-dominated data which is approximately Gaussian; rejection rates would be more permissive where these assumptions do not hold. 4.3.”

Pith says: Objection 2 of 4 · major

The covariance model assumes stationary Gaussian noise and independent whitened frequency bins. The authors acknowledge in b95 that 'rejection rates would be more permissive where these assumptions do not hold', but the direction of the failure is not guaranteed: non-Gaussian noise or non-stationary features could produce either more or fewer false dismissals than the model predicts.

more of objection 2

The paper should discuss whether the screening thresholds are conservative or anti-conservative under realistic noise deviations (e.g., spectral lines, glitches, scattered light), and whether the method can be adapted to estimate the noise distribution empirically from the data rather than relying solely on the Gaussian model.

The paper says, at §5, Tables 1 and 2, Fig. 3

The review did not quote this passage.

Pith says: Objection 3 of 4 · major

The fused peak-only speedups are reported relative to a standalone inverse FFT (AOCL-FFTW or rocFFT) that does not include the conjugate product or peak search. The authors note this in the caption of Fig. 3(a) and in the text, but the headline speedups (1.4×, 6.2×, 3.1×) compare against the inverse FFT only, not against a complete unfused pipeline that includes multiplication and peak finding.

more of objection 3

For a fair comparison, the baseline should include the full pipeline cost. The paper does provide some comparisons against the complete unfused pipeline on GPU (6–13× for fused peak-only vs. unfused pipeline on Radeon 8060S, §5.2), but the CPU comparisons are only against the standalone FFT. The authors should either provide CPU benchmarks of the complete unfused pipeline or clarify that the speedups are relative to the transform-only baseline and that the advantage over a full pipeline would be larger.

The paper says, at §4.2, Eq. (10) and chain selection

The review did not quote this passage.

Pith says: Objection 4 of 4 · major

The chain selection minimizes expected cost using modeled cumulative noise survival fractions π_i. However, the π_i depend on the detection threshold ρ_th and the false-dismissal budget α, and the model assumes Gaussian noise. The paper does not report how sensitive the optimal chain is to deviations from the Gaussian assumption or to errors in the estimated survival fractions.

more of objection 4

If the model underestimates π_i, the chain may be too aggressive and cause excessive false dismissals; if it overestimates π_i, the chain may be too conservative and reduce the speedup. The authors should discuss the robustness of the chain selection to model misspecification.

7 minor objections

The paper says, at Abstract and §1

Pith says: minor objection 1 of 7

The abstract states 'fused peak-only filtering is 1.4× faster than a standalone AOCL-FFTW inverse transform on one CPU core and 6.2× faster than rocFFT on its Radeon 8060S GPU (3.1× over MLX FFT on the M2 GPU).' The phrase 'standalone AOCL-FFTW inverse transform' is clear, but the GPU comparisons should also specify that the baseline is the inverse transform only, not the full pipeline, for consistency.

The paper says, at §4.1, Eq. (7)

Pith says: minor objection 2 of 7

The definition of A(x) as m F^{-1}(q_hat)[x] is given, but the normalization of the inverse transform is not explicitly stated. It would be helpful to note that the 1/m normalization of the length-m inverse transform cancels the factor m, so that A(0)=1.

show 5 more

The paper says, at §4.2, Eq. (9)

Pith says: minor objection 3 of 7

The notation P_FD(ρ_c) uses ρ_c as a vector of thresholds, but the equation shows a union over tiers. Clarify that ρ_c is the set of tier thresholds (ρ_{c,1}, ρ_{c,2}, ...).

The paper says, at §5.1, Fig. 3(a)

Pith says: minor objection 4 of 7

The shaded bands span the 10th–90th percentile of timing blocks, but the number of timing blocks is not stated. Adding this information would help assess the variability.

The paper says, at §5.2, Fig. 4(c)–(d)

Pith says: minor objection 5 of 7

The 'effective problem throughput' is defined as 48N bytes divided by execution time for the fused and hierarchical cases, but the hierarchical case skips operations and memory reads on rejected pairs, so the throughput metric is not directly comparable to the memory bandwidth ceilings. The authors note this in the caption, but it would be clearer to state in the text that the hierarchical curves are not physical memory throughputs.

The paper says, at §6, last paragraph

Pith says: minor objection 6 of 7

The suggestion that reduced-precision formats (float16, bfloat16, fp8) could be used for early screening tiers is interesting but speculative. The paper does not test this, so it should be clearly marked as future work rather than presented as a near-term possibility.

The paper says, at References

Pith says: minor objection 7 of 7

Reference [40] (Google Highway) is listed as 2024, but the URL points to a GitHub repository. The reference should include the version or commit hash used for reproducibility.

What would settle it

Measure the Section 4.3 conditional FDR on real data: inject ~1e5 signals at SNR 5.5-8 at random lags into 16 public LIGO O4 noise segments (N=4096, Delta f=0.5 Hz) and run the (512, 1024, 2048) chain at rho_th=5.5, alpha=1e-2 under two reference configurations: the single BNS-matched envelope and a slope-partitioned multi-reference grouping. If the measured FDR exceeds ~2e-2 in either configuration, Section 6's guarantee fails in the paper's fiducial search regime and the mitigation needs a specified grouping procedure plus real-data calibration; if the FDR stays within the Wilson bounds of the Figure 2(c) targets - and deviations show up, per b95, as reduced speedup rather than increased FDR - the concern does not land.

from the skeptic's reading

How the review went

3 readings by deepseek-v4-flash · 13 Nov 2026

  1. Reader holds up confidence moderate
  2. Skeptic agrees 3 min 6 s · 18,438 tokens read, 27,285 written
  3. Referee major revision: 4 major objections, 7 minor 47 s · 18,913 tokens read, 1,862 written

Not run on this paper: the editor, the simulated authors' defence.

Pith did not check this review's objections against the paper.

The review in full

The strongest objection

The screening speedups (26x/21x/19x) and the closing guarantee (b144: 'without sacrificing signal recovery beyond a user-specified false-dismissal rate') both rest on the FDR bound holding when the method is deployed. Within the tested regime the model is internally consistent: re-deriving Eqs. (7)-(8) with the paper's unit-variance coarse normalization gives E[z_c[L]] = rho*sqrt(mu) and A(x) = m*F^{-1}{q_hat}, and the Section 4.2 cross-tier covariance sqrt(mu_i/mu_j)*A_bi(x) matches the shared-band overlap, so the Figure 2(c) matched-reference validation across three decades of alpha with >2.5e5 injections per profile is genuine evidence. The fragile pillar is reference representativeness, and Section 4.3 demonstrates the failure: a single edge reference over an f^gamma bank, gamma in [-2.33, -1.33], yields ~27% bank-wide FDR against a 1e-2 target (27x), and even the bank-mean reference gives ~3.5%. The proposed mitigation ('partition the bank into four reference groups ordered by spectral slope') is validated only on this synthetic bank; no algorithm or tolerance for choosing the groups is specified, so a user cannot determine when a real bank is 'similar enough'. Second, the failure-direction claim (b95) that rejection rates 'would be more permissive where these assumptions do not hold' is asserted without argument and is not directionally guaranteed: non-Gaussian noise concentrated in the high-frequency part of the band (glitches, scattered light, spectral lines above 256 Hz) raises the full-band peak without raising the coarse-stage response, which would inflate the FDR above target while simultaneously reducing the screening speedup - the worst case for the advertised claim. All validation is simulated whitened Gaussian noise; real-data FDR is entirely untested.

The central claim decomposes into (a) fused peak-only throughput gains and (b) hierarchical screening speedups at a statistically controlled false-dismissal rate. I checked the arithmetic of (a) and the internal consistency of (b). Table 2 is self-consistent: 32.84/22.78 = 1.44x, 32.84/1.24 = 26.5x, 22.78/1.24 = 18.4x, matching the abstract; the Figure 3(b) throughput values (8192 pairs over 1.24 ms = 6.6 Mpairs/s) agree; the traffic model Eqs. (3)-(5) evaluates correctly (R ~ 95x at K=32, N=4096). Re-deriving Eqs. (7)-(8) under the paper's unit-variance coarse normalization, the coarse response peak E[z_c[L]] = rho*sqrt(mu) and autocovariance A(x) = m*F^{-1}{q_hat} follow, and the cross-tier covariance sqrt(mu_i/mu_j)*A_bi(x) matches the shared-band overlap, so the screening model is coherent in its stated regime. The Figure 2(c) matched-reference validation (three decades of alpha, >2.5e5 injections per profile, disclosed Wilson intervals) is genuine supporting evidence. The baselines are honest: FFTW_PATIENT on CPU, rocFFT and MLX on GPU, with the standalone inverse transform timed without the conjugate product or peak search, making the fused comparison conservative. The Section 4.3 stress test that fails (27% FDR at a 1e-2 target) is disclosed rather than hidden, which further supports good faith. The load-bearing risk is the transfer of the FDR guarantee: reference representativeness is the controlling variable, the grouping mitigation is validated only on a synthetic f^gamma bank with no specified procedure, and the b95 direction claim is unproven - high-frequency-concentrated non-Gaussianity would inflate FDR while shrinking the speedup. This does not refute the claims as stated for the tested regime, and the main limitation is acknowledged by the authors, so the reader's ACCEPT stands unchanged; the concrete test above defines the condition under which the concern would land.

The referee's report

If the claims hold, this work provides a practical, open-source acceleration of a core computational kernel in gravitational-wave astronomy, seismology, and radio transient searches. The operator-fusion and 2D tiling approach is a clean application of known IO-aware optimization principles to the matched-filtering context, and the statistical screening model with covariance-calibrated thresholds is a principled alternative to ad-hoc hierarchical search thresholds. The validation includes extensive Monte Carlo testing (2.5×10^5 injections per profile) and cross-platform benchmarks. The paper is clearly written and the code is released under an open-source license, supporting reproducibility.

Pith's reading of the paper (a machine's summary, not the authors')

Matched filtering compares a stretch of noisy measurement data against many template signals to find the best match. The standard method works in Fourier space: multiply the data spectrum by a template spectrum, run an inverse FFT to get a correlation curve, then scan the curve for the largest peak. For every data-template pair, the old pipeline writes the intermediate product and the full correlation curve to memory and reads them back, and those memory trips dominate the cost.

This paper removes most of those trips. The multiplication is folded into the first FFT stage and the peak search into the last stage, so for each pair only the input spectra are read and only the winning peak record is written. Batching is tiled in two dimensions, so the same loaded spectra serve many pairs at once. On eight CPUs and two integrated GPUs these fused kernels beat a bare vendor inverse FFT by 1.2 to 6.2 times, and peak accuracy stayed at about one part in ten million in single precision.

The second trick exploits the fact that for gravitational-wave inspiral signals most of the signal power sits at low frequencies while the highest frequencies dominate the transform length. A screening chain evaluates short low-frequency correlations first and runs the expensive full-band filter only on surviving pairs. Stage thresholds are computed from the template power profile and the matched-filter noise autocovariance to hit a chosen false-dismissal rate, for example one percent. In the fiducial test only 0.22 percent of pairs reach the full filter, giving 19 to 26 times total search speedups. The authors state that the gains shrink for templates whose power is spread across the whole band, and that the thresholds assume approximately Gaussian, stationary noise.

Core claim. Central claim (abstract and Sections 3 to 5): 'We fuse the conjugate product into the first inverse-transform butterflies and the peak reduction into the last so that only candidate peaks leave the processor registers, and tile batches in 2D so K^2 correlations reuse 2K cached spectra.' Concretely, fused peak-only filtering is 1.4x faster than a standalone AOCL-FFTW inverse transform on one CPU core at N=4096, 6.2x faster than rocFFT on the Radeon 8060S, and 3.1x faster than MLX on the Apple M2; hierarchical screening at rho_th=5.5 with a 1e-2 false-dismissal target gives 26x, 21x, and 19x speedups respectively. The accuracy claim is that single-precision fused filtering matches a float64 reference to a maximum relative peak error of 2.30e-7 with complete agreement on the peak sample index in every trial.

The premise it rests on. The screening thresholds are calibrated from a model in which whitened stationary Gaussian noise produces coarse-band correlations that are jointly Gaussian with covariance A(x) = m F^{-1}(q_hat)[x] fixed by the spectral power profile alone (Eqs. 7 to 9, Sections 4.1 to 4.2). If this covariance model is wrong, or if the reference envelope for a template bank is unrepresentative, the realized false-dismissal rate exceeds the target: the paper's own Section 4.3 heterogeneous-bank test measures about 27 percent bank-wide false dismissals against a 1e-2 target when a single edge template serves as reference. The model is validated only on simulated whitened Gaussian noise, and the authors note (Section 4.3, sentence b95) that rejection rates 'would be more permissive where these assumptions do not hold', so real-data behavior is untested.

Check this review

Pith stored no record of this review's passes and took no fingerprint of it. Reviews made by code from 13 Nov 2026 on carry both.

How often reviews like this one are right

Reviews like this one: written by deepseek-v4-flash, where what was read was not recorded.

How often objections from reviews like this one are right: not yet measured

Every value above, in the review's public record (JSON)

Reviews by people

This review was written by a machine. No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.

Request a human review

How Pith checks papers · The full audit of this paper

The paper

Abstract

Batched matched filtering and cross-correlation over large template banks are a primary computational cost in gravitational-wave astronomy, seismology, and radio transient searches. Standard pipelines pair a general-purpose inverse Fast Fourier Transform (FFT) with separate multiplication and peak-finding kernels, writing intermediate spectra and full correlation series to memory for every template. We fuse the conjugate product into the first inverse-transform butterflies and the peak reduction into the last so only candidate peaks leave processor registers, and tile batches in 2D so K2 correlations reuse 2K cached spectra. This shifts filtering from a DRAM-bandwidth-bound to an on-chip cache-bandwidth-bound regime. In some situations, the signal-to-noise ratio (SNR) is concentrated toward lower frequencies while higher frequencies dominate the transform length. We develop a hierarchical screening method that exploits this via a sequence of short low-frequency transforms before evaluating the full band, deriving stage thresholds from the spectral power profile and matched-filter noise autocovariance to target a specified false-dismissal probability. We demonstrate these methods in the open-source matchedfilter library across eight CPU architectures and two integrated GPUs (AMD Radeon 8060S via Vulkan and Apple M2 via Metal). On an AMD Ryzen AI Max+ 395 at N = 4096, fused peak-only filtering is 1.4 × faster than a standalone AOCL-FFTW inverse transform on one CPU core and 6.2 × faster than rocFFT on its Radeon 8060S GPU (3.1 × over MLX FFT on the M2 GPU). For a fiducial gravitational-wave search (ρth = 5.5, 10−2 false-dismissal target), hierarchical screening yields a 26 × speedup over AOCL-FFTW on CPU, 21 × over rocFFT on the 8060S GPU, and 19 × over MLX FFT on the M2 GPU.

The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. No sentence in this record was written by a model.

abstract compiled · 2026-11-14

The authors' main claim, as the classifier picked it out 0.96

“For a fiducial gravitational-wave search ($\rho_\mathrm{th}=5.5$, $10^{-2}$ false-dismissal target), hierarchical screening yields a $26\times$ speedup over AOCL-FFTW on CPU, $21\times$ over rocFFT on the 8060S GPU, and $19\times$ over MLX FFT on the M2 GPU.”

How the main claim is stated: asserted 1.00

What the abstract itself states, as probabilities: claims first 0.04 · states limitation 0.35 · human participants 0.02 · controlled comparison 0.82 · randomized assignment 0.01 · states quantitative result 0.99 · compares with prior results 0.98 · mentions code or data availability 0.80

The abstract, sentence by sentence

This table labels the abstract only (arXiv listing abstract v1). The number is the probability the classifier gave the label.

SentenceRoleThe authors' own workStates the main conclusionStrength
Batched matched filtering and cross-correlation over large template banks are a primary computational cost in gravitational-wave astronomy, seismology, and radio transient searches. background 1.00 0.05 0.02 asserted 1.00
Standard pipelines pair a general-purpose inverse Fast Fourier Transform (FFT) with separate multiplication and peak-finding kernels, writing intermediate spectra and full correlation series to memory for every template. background 0.98 0.04 0.02 asserted 1.00
We fuse the conjugate product into the first inverse-transform butterflies and the peak reduction into the last so only candidate peaks leave processor registers, and tile batches in 2D so $K^2$ correlations reuse $2K$ cached spectra. method 0.91 0.94 0.17 asserted 1.00
This shifts filtering from a DRAM-bandwidth-bound to an on-chip cache-bandwidth-bound regime. interpretation 0.94 0.78 0.15 asserted 1.00
In some situations, the signal-to-noise ratio (SNR) is concentrated toward lower frequencies while higher frequencies dominate the transform length. background 0.99 0.12 0.02 hedged possibility 0.62
We develop a hierarchical screening method that exploits this via a sequence of short low-frequency transforms before evaluating the full band, deriving stage thresholds from the spectral power profile and matched-filter noise autocovariance to target a specified false-dismissal probability. method 0.82 0.96 0.08 asserted 1.00
We demonstrate these methods in the open-source matchedfilter library across eight CPU architectures and two integrated GPUs (AMD Radeon 8060S via Vulkan and Apple M2 via Metal). method 0.57 0.95 0.05 asserted 1.00
On an AMD Ryzen AI Max+ 395 at $N=4096$, fused peak-only filtering is $1.4\times$ faster than a standalone AOCL-FFTW inverse transform on one CPU core and $6.2\times$ faster than rocFFT on its Radeon 8060S GPU ($3.1\times$ over MLX FFT on the M2 GPU). result 1.00 0.96 0.19 asserted 1.00
For a fiducial gravitational-wave search ($\rho_\mathrm{th}=5.5$, $10^{-2}$ false-dismissal target), hierarchical screening yields a $26\times$ speedup over AOCL-FFTW on CPU, $21\times$ over rocFFT on the 8060S GPU, and $19\times$ over MLX FFT on the M2 GPU. result 1.00 0.96 0.44 asserted 1.00

What kind of paper the classifier reports this is: method or tool 1.00 · evidence from simulation or computation 0.95

What was examined, and what was not

Sources read

  • arXiv listing abstract v1

Not examined

  • full text
  • figures and tables
  • references and citing sentences
  • relations to other papers

Levels run

  • P abstract

Not run

  • S sections
  • T body sentences
  • C citations
  • R cross-paper relations

How the labels were made

Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-11-14: 1 request, 7,409 input tokens.

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. This record as JSON

Figures (4)
Figure 1 of the paper
Figure 1. Memory traffic in unfused, fused, and tiled matched filtering. (a) A conventional pipeline writes the product P = Dd ⊙ H ∗ t and the full correlation series z[τ] to memory buffers (cache or DRAM) and reads both back, transferring 48N bytes per pair. The fused filter applies the conjugate product in the first butterfly and the |z| 2 maximum in the last, keeping intermediate states in registers, L1 cache, or GPU local memory so only the 16N bytes of input spectra are read. (b) In a K × K cache tile, 2K cached input spectra serve K 2 correlations. (c) Memory traffic per pair at N = 4096
Figure 2 of the paper
Figure 2. Statistically controlled hierarchical screening and empirical false-dismissal rate (FDR) validation at
Figure 3 of the paper
Figure 3. Single-core CPU performance on the AMD Ryzen AI Max
Figure 4 of the paper
Figure 4. GPU performance and effective problem throughput on the integrated AMD Radeon 8060S (40 CUs, Vulkan; left) and Apple M2 GPU (10 cores, Metal; right) for 16 × 512 = 8,192 pairs per batch. (a, b) Time per pair, including device synchronization, for the vendor FFT baseline (rocFFT or MLX inverse transform only), the complete unfused pipeline (conjugate product → inverse FFT → single-pass peak search), and the fused peak-only and hierarchical screening filters (ρth = 5.5, α = 10−2 ). (c, d) Effective problem throughput, defined as 48N conventional problem bytes divided by execution time…

Figures are the authors', reproduced from the paper.

References (41)
  1. [1] B. Allen, W. G. Anderson, P. R. Brady, D. A. Brown, J. D. E. Creighton, FINDCHIRP: An algorithm for de- tection of gravitational waves from inspiraling compact binaries, Physical Review D 85 (12) (2012) 122006. 8 on Pith
  2. [2] S. J. Gibbons, M. B. Sørensen, D. B. Harris, F. Ringdal, The detection and location of low magnitude earthquakes in northern Norway using multi-channel waveform cor- relation at regional distances, Physics of the Earth and Planetary Interiors 160 (3–4) (2007) 285–309 on Pith
  3. [3] S. A. Usman, A. H. Nitz, I. W. Harry, C. M. Biwer, D. A. Brown, M. Cabero, C. D. Capano, T. Dal Canton, T. Dent, S. Fairhurst, et al., The PyCBC search for gravitational waves from compact binary coalescence, Classical and Quantum Gravity 33 (21) (2016) 215004 on Pith
  4. [4] C. Messick, K. Blackburn, P. Brady, P. Brockill, K. Cannon, R. Cariou, S. Caudill, S. J. Chamberlin, J. D. E. Creighton, R. Everett, et al., Analysis framework for the prompt dis- covery of compact binary mergers in gravitational-wave data, Physical Review D 95 (4) (2017) 042001 on Pith
  5. [5] A. H. Nitz, T. Dal Canton, D. Davis, S. Reyes, Rapid de- tection of gravitational waves from compact binary merg- ers with PyCBC Live, Physical Review D 98 (2) (2018) 024050 on Pith
  6. [6] A. H. Nitz, T. Dent, G. S. Davies, S. Kumar, C. D. Capano, I. Harry, S. Mozzon, L. Nuttall, A. Lundgren, M. Tápai, 2- OGC: Open gravitational-wave catalog of binary mergers from analysis of public Advanced LIGO and Virgo data, The Astrophysical Journal 891 (2) (2020) 123 on Pith
  7. [7] S. M. Ransom, S. S. Eikenberry, J. Middleditch, Fourier techniques for very long astrophysical time-series analysis, The Astronomical Journal 124 (3) (2002) 1788–1809 on Pith
  8. [8] S. Dimoudi, K. Adámek, P. Thiagaraj, S. M. Ransom, A. Karastergiou, W. Armour, A GPU implementation of the correlation technique for real-time Fourier domain pul- sar acceleration searches, The Astrophysical Journal Sup- plement Series 239 (2) (2018) 28 on Pith
  9. [9] K. Adámek, J. Novotný, S. Dimoudi, W. Armour, Search- ing for pulsars in extreme orbits: GPU acceleration of the Fourier domain “jerk” search, The Astrophysical Journal Supplement Series 247 (2) (2020) 56 on Pith
  10. [10] J. W. Cooley, J. W. Tukey, An algorithm for the machine calculation of complex Fourier series, Mathematics of Computation 19 (90) (1965) 297–301 on Pith
  11. [11] M. Frigo, S. G. Johnson, The design and implementation of FFTW3, Proceedings of the IEEE 93 (2) (2005) 216–231 on Pith
  12. [12] K. Adámek, S. Dimoudi, M. Giles, W. Armour, GPU fast convolution via the overlap-and-save method in shared memory, ACM Transactions on Architecture and Code Optimization 17 (3) (2020) 18:1–18:20 on Pith
  13. [13] K. Cannon, R. Cariou, A. Chapman, M. Crispin-Ortuzar, N. Fotopoulos, M. Frei, C. Hanna, E. Kara, D. Keppel, L. Liao, S. Privitera, A. Searle, L. Singer, A. Weinstein, To- ward early-warning detection of gravitational waves from compact binary coalescence, The Astrophysical Jou... on Pith
  14. [14] T. Adams, D. Buskulic, V . Germain, G. M. Guidi, F. Mar- ion, M. Montani, B. Mours, F. Piergiovanni, G. Wang, Low-latency analysis pipeline for compact binary coales- cences in the advanced gravitational wave detector era, Classical and Quantum Gravity 33 (17) (2016) 175012 on Pith
  15. [15] S. Vinciguerra, J. Veitch, I. Mandel, Accelerating gravi- tational wave parameter estimation with multi-band tem- plate interpolation, Classical and Quantum Gravity 34 (11) (2017) 115006 on Pith
  16. [16] S. Morisaki, Accelerating parameter estimation of grav- itational waves from compact binary coalescence using adaptive frequency resolutions, Physical Review D 104 (4) (2021) 044062 on Pith
  17. [17] R. Dhurkunde, H. Fehrmann, A. H. Nitz, Hierarchical ap- proach to matched filtering using a reduced basis, Physical Review D 105 (10) (2022) 103001 on Pith
  18. [18] A. H. Nitz, K. Kacanja, K. Soni, Beyond FINDCHIRP for gravitational-wave matched-filter searches with ratio-filter dechirping, Physical Review D 113 (8) (2026) 082003 on Pith
  19. [19] F. Franchetti, Y . V oronenko, M. Püschel, Formal loop merg- ing for signal transforms, in: Proceedings of the 2005 ACM SIGPLAN Conference on Programming Language Design and Implementation (PLDI), ACM, 2005, pp. 315–326 on Pith
  20. [20] M. Mathieu, M. Henaff, Y . LeCun, Fast training of convolutional networks through FFTs, in: International Conference on Learning Representations (ICLR), 2014, arXiv:1312.5851 on PithPith's reviewarXiv
show 21 more
  1. [21] N. Vasilache, J. Johnson, M. Mathieu, S. Chintala, S. Piantino, Y . LeCun, Fast convolutional nets with fbfft: A GPU performance evaluation, in: International Conference on Learning Representations (ICLR), 2015, arXiv:1412.7580 on PithPith's reviewarXiv
  2. [22] D. Y . Fu, H. Kumbong, E. Nguyen, C. Ré, FlashFFTConv: Efficient convolutions for long sequences with tensor cores, in: International Conference on Learning Representations (ICLR), 2024, arXiv:2311.05908 on PithPith's reviewarXiv
  3. [23] T. Dao, D. Y . Fu, S. Ermon, A. Rudra, C. Ré, FlashAt- tention: Fast and memory-efficient exact attention with IO-awareness, Advances in Neural Information Processing Systems 35 (2022) 16344–16359 on Pith
  4. [24] J. Šochman, J. Matas, WaldBoost: Learning for time con- strained sequential detection, in: Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR), V ol. 2, IEEE, 2005, pp. 150–156 on Pith
  5. [25] T. Rakthanmanon, B. Campana, A. Mueen, G. Batista, B. Westover, Q. Zhu, J. Zakaria, E. Keogh, Searching and mining trillions of time series subsequences under dynamic time warping, in: Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data ... on Pith
  6. [26] K. V . Mishra, Y . C. Eldar, Sub-Nyquist radar: Principles and prototypes, in: Compressed Sensing in Radar Signal Processing, Cambridge University Press, 2019, pp. 1–48, arXiv:1803.01819 on PithPith's reviewarXiv
  7. [27] B. Zackay, E. O. Ofek, An accurate and efficient algorithm for detection of radio bursts with an unknown dispersion measure, for single-dish telescopes and interferometers, The Astrophysical Journal 835 (1) (2017) 11 on Pith
  8. [28] J. White, K. Adámek, J. Roy, S. M. Ransom, W. Armour, Pulscan: Binary pulsar detection using unmatched filters on NVIDIA GPUs, The Astrophysical Journal Supplement Series 276 (1) (2025) 20, arXiv:2406.15186 on PithPith's reviewarXiv
  9. [29] S. D. Mohanty, S. V . Dhurandhar, A hierarchical search strategy for the detection of gravitational waves from coa- lescing binaries, Physical Review D 54 (12) (1996) 7108– 7128 on Pith
  10. [30] S. D. Mohanty, Hierarchical search strategy for the de- tection of gravitational waves from coalescing binaries: Extension to post-Newtonian waveforms, Physical Review D 57 (2) (1998) 630–658 on Pith
  11. [31] T. Tanaka, H. Tagoshi, Use of new coordinates for the tem- plate space in hierarchical search for gravitational waves from inspiraling binaries, Physical Review D 62 (8) (2000) 082001 on Pith
  12. [32] A. S. Sengupta, S. V . Dhurandhar, A. Lazzarini, T. Prince, Extended hierarchical search (EHS) algorithm for detection of gravitational waves from inspiraling compact binaries, Classical and Quantum Gravity 19 (7) (2002) 1507–1512 on Pith
  13. [33] B. U. Gadre, S. Mitra, S. Dhurandhar, Hierarchical search strategy for the efficient detection of gravitational waves from nonprecessing coalescing compact binaries with aligned-spins, Physical Review D 99 (12) (2019) 124035 on Pith
  14. [34] K. Soni, B. U. Gadre, S. Mitra, S. Dhurandhar, Hierarchical search for compact binary coalescences in the Advanced LIGO’s first two observing runs, Physical Review D 105 (6) (2022) 064005 on Pith
  15. [35] K. Cannon, A. Chapman, C. Hanna, D. Keppel, A. C. Searle, A. J. Weinstein, Singular value decomposition ap- plied to compact binary coalescence gravitational-wave signals, Physical Review D 82 (4) (2010) 044025 on Pith
  16. [36] S. Hooper, S. K. Chung, J. Luan, D. Blair, Y . Chen, L. Wen, Summed parallel infinite impulse response filters for low- latency detection of chirping gravitational waves, Physical Review D 86 (2) (2012) 024012 on Pith
  17. [37] D. Tolmachev, VkFFT—A performant, cross-platform and open-source GPU FFT library, IEEE Access 11 (2023) 12039–12058 on Pith
  18. [38] S. Williams, A. Waterman, D. Patterson, Roofline: an insightful visual performance model for multicore architec- tures, Communications of the ACM 52 (4) (2009) 65–76 on Pith
  19. [39] D. H. Bailey, FFTs in external or hierarchical memory, The Journal of Supercomputing 4 (1) (1990) 23–35 on Pith
  20. [40] Google LLC, Google Highway: Efficient and portable SIMD/vector intrinsics, https://github.com/google/ highway(2024) on Pith
  21. [41] Y . He, K. Fatahalian, T. Foley, Slang: Language mech- anisms for extensible real-time shading systems, ACM Transactions on Graphics 37 (4) (2018) 141:1–141:13. 10 on Pith

Challenges and discussion

Disagree with the review, or know something it missed? Leave a comment, or challenge this review formally: a challenge becomes a permanent entry in this paper's record, and its decision is published here.

No comments yet.

Sign in to comment