Inter-detector correlations in LIGO-Virgo strain: an exploratory study and possible astrophysical implications
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: General Relativity and Quantum Cosmology. The day in general relativity and quantum cosmology.
Kai: I'm Kai.
Mira: And I'm Mira, here with Lev.
Kai: We're looking at the paper: "Inter-detector correlations in LIGO-Virgo strain".
Title and authors: Kai: So what the paper actually summarizes is that they’re looking at these events and finding something pretty dominant in their results. In twenty-six out of twenty-seven pair measurements, the peak correlation in the event window falls right into the middle of where we expect noise to be.
Mira: That means they conclude that in most cases, their peak isn't statistically strong enough to claim it's a genuine signal above the noise floor when you look at this paper "Inter-detector correlations in LIGO-Virgo strain: an exploratory study and possible astrophysical implications".
Lev: If I were building this system, and I saw these results, I’d be saying that the expected signal amplitude is buried too deep in our noise model for us to pull it out reliably right now.
Kai: Exactly. The main point they make is that even when you look at pairs involving the independent Virgo detector, like H1 and V1 or L1 and V1, those results are similarly quiet under their method.
Mira: They also highlight that for the H1 and L1 pair with event GW two hundred thousand four hundred fifty-eight they found a very strong peak with an empirical p-value of zero point zero zero five.
Lev: That low p-value is interesting because it points to one specific case where the time delay might be more significant than in the other events.
Kai: Yeah, that specific exception is what they focus on when they discuss how different detector pairs behave under this test of the stochastic gravitational-wave background hypothesis.
The paper's summary: Mira: Now, the authors suggest some ways to improve this kind of analysis, and they’re focused on making the method more robust by using a model-independent approach to their p-value calculation.
Kai: They use an empirical p-value defined as pemp equals k plus one over N plus one, where k is the number of noise windows that have a correlation stronger than the event peak. That’s a way to handle finite samples better than just using a standard statistical test.
Lev: I like that idea of using the Laplace correction, which prevents you from getting zero for your p-value when you only have twenty events or so; it gives you more confidence in what you're saying about the noise being truly dominant.
Mira: They also point out a limitation in standard processing methods, specifically how power spectral density whitening affects things. Applying standard whitening doesn't clean up the data; it actually removes that one isolated excess, like for GW two hundred thousand four hundred fifty-eight.
Kai: That’s a key point because it shows that correlations might extend into the surrounding strain data, not just stay confined to the event window when you apply standard power spectral density whitening.
Lev: So what this implies is that we should look at those surrounding windows too, instead of just focusing on the narrow band around the event time.
The paper's improvements: Kai: To wrap things up on this paper, they conclude that the inter-detector time delay isn't a strong enough standalone test by itself for validating a signal. They say looking at the data under whitening doesn't help separate event from noise; it just removes the excess.
Mira: So their final interpretation points toward something else: they suggest that the most plausible physical explanation is the presence of a continuous correlated component in the strain data, which they naturally call a stochastic gravitational-wave background.
Lev: If we were building a detector network and saw this pattern across all pairs, it tells us we aren't just seeing random noise; we’re looking at some kind of underlying physical process that spans the whole network.
Kai: And because of that, the authors strongly suggest the next step is to test this idea using a four-detector setup involving KAGRA. If they see the same correlated structure there, it would be a very strong indicator for them.
Mira: They also mention that high-amplitude transient events, like mergers or rare background fluctuations, could be sitting on top of this continuous background component. So the transient event doesn't disprove the background; it just sits on top of it.
Lev: That makes sense from a hardware perspective—you have your fundamental noise floor plus whatever physical process is causing those long-term correlations we’re trying to measure.
Kai: So this paper, "Inter-detector correlations in LIGO-Virgo strain: an exploratory study and possible astrophysical implications," shows that while individual time delays are tricky, the overall pattern across all detectors points toward a background noise hypothesis.
Conclusion: Kai: So, to wrap up on "Inter-detector correlations in LIGO-Virgo strain: an exploratory study and possible astrophysical implications," they’ve found that for most detector pairs, the event peak isn't statistically separated from the noise level when you look at all those surrounding windows.
Mira: That means the standard way we check for a signal using just time delays isn't a great standalone test. They’re seeing that in most cases, their peak falls right into where the empirical noise distribution expects it to be.
Lev: From an engineering standpoint, if you saw this in real hardware, it would tell you that any correlation we measure across those detector pairs could just be instrumental noise bouncing around in the shared infrastructure.
Kai: Exactly. The main point they make is that even when you look at pairs involving the independent Virgo detector, like H1 and V1 or L1 and V1, those results are similarly quiet under their method.
Mira: And they highlight one exception for the H1 and L1 pair with event GW two hundred thousand four hundred fifty-eight. They found a very strong peak there with an empirical p-value of zero point zero zero five.
Lev: That low p-value is interesting because it points to one specific case where the time delay might be more significant than in the other events.
Kai: Yeah, that specific exception is what they focus on when they discuss how different detector pairs behave under this test of the stochastic gravitational-wave background hypothesis.
Mira: They conclude that the most plausible physical explanation for this persistent correlation structure is a continuous correlated component in the strain data, which they call a stochastic gravitational-wave background.
Lev: If you were building a detector network and saw this pattern across all pairs, it tells you that you aren't just seeing random noise; you’re looking at some kind of underlying physical process that spans the whole network.
Kai: And because of that, the authors strongly suggest the next step is to test this idea using a four-detector setup involving KAGRA. If they see the same correlated structure there, it would be a very strong indicator for them.
Mira: They also mention that high-amplitude transient events, like mergers or rare background fluctuations, could be sitting on top of this continuous background component. So the transient event doesn't disprove the background; it just sits on top of it.
Lev: That makes sense from a hardware perspective—you have your fundamental noise floor plus whatever physical process is causing those long-term correlations you’re trying to measure.
Kai: So the paper, "Inter-detector correlations in LIGO-Virgo strain: an exploratory study and possible astrophysical implications," shows that while individual time delays are tricky, the overall pattern across all detectors points toward a background noise hypothesis.
Mira: It's a good reminder that sometimes we have to look beyond just the transient event and consider what’s happening continuously in the data stream.
Instituto de Física, Universidade de Brasília · Canadian Quantum Research Center · Instituto Federal do Tocantins · Universidade Federal do Norte do Tocantins
gr-qc
Submitted: 2026-07-12
Updated: 2026-11-17
Importance score: 80/100
The gist: On correlated noise in the LIGO-Virgo network: a test of the stochastic gravitational-wave background hypothesis is an independent cross-correlation analysis of publicly available LIGO–Virgo strain…
Key concepts
- Cross-correlation analysis
- This method compares the strain data recorded by two different detectors (like LIGO and Virgo) at different times to see if they move together. It is used here to check if a signal is real or just noise that happens independently in each detector.
- Empirical p-value (pemp)
- This statistical measure determines how likely it is that the observed peak correlation between two detectors happened purely by chance within the surrounding noise. A very low p-value suggests the event peak is statistically significant and not just a random fluctuation in the noise.
- Stochastic Gravitational-Wave Background
- This hypothesis suggests that gravitational waves are constantly present from distant sources, creating a persistent, continuous background hum in spacetime. The study investigates whether this continuous background could be causing the correlated noise observed across detectors.
- Detector Independence Test
- The researchers used multiple detector pairs (H1×L1, H1×V1, L1×V1) and included the independent Virgo detector to test if the correlation is a physical phenomenon or just a result of shared equipment. If it appears in all independent setups, it strengthens the case for a physical background.
Terminology
Summary
On correlated noise in the LIGO-Virgo network: a test of the stochastic gravitational-wave background hypothesis is an independent cross-correlation analysis of publicly available LIGO–Virgo strain data for nine gravitational-wave candidate events from the O2 and O3 observing runs, extending the methodology of Creswell et al. p. 0 p. 7 [1] to all three detector pairs (H1×L1, H1×V1, L1×V1) and to a sample five times larger than the original study. p. 7 In 26 of 27 pair-measurements, the peak cross-correlation C∗ev in the event window lies within the bulk of the empirical noise distribution, with pemp ≥ 0. p. 312 in every such case. p. 3 The central strength of our argument lies in the role of Virgo as an experimental control. p. 8
Methodology and Analysis Approach
The analysis employed a model-independent approach by computing the normalized cross-correlation function C(τ) in two contexts for each detector pair (d1, d2) ∈ 3 pairs. For each pair, the authors computed C(τ) [Eqs. p. 2 (2)–(3)] in two contexts: in the event window, a w = 0. p. 22 s segment centred on the GPS time where they record the signed peak C∗ev and corresponding lag τev. p. 2 They then drew N = 200 noise windows of identical width uniformly at random from the surrounding 256 s, excluding a ±2 guard zone around the event and enforcing a minimum separation of 1 s between consecutive windows. p. 3 The empirical p-value is defined as pemp = k + 1 / N + 1, where k = card j: C∗j ≥ C∗ev. p. 3 A value pemp ≲ 0. p. 301 would indicate that the event window produces a peak that the noise essentially never reaches. p. 3
Evaluation of Results Across Detector Pairs
The dominant result is negative: in 26 of 27 pair-measurements, including all pairs involving the independent Virgo detector, the event window produces a peak C∗ev that falls well within the bulk of the empirical noise distribution. No detector pair involving Virgo (H1×V1 or L1×V1) yields pemp < 0. p. 320 for any event in the sample. p. 3 The single exception is the H1×L1 pair for GW190412, with pemp = 0. p. 8005, precisely in the detector pair most susceptible to shared-infrastructure confounds. p. 8
Implications for Noise Structure
The observation that off-source windows in H1×V1 and L1×V1 produce cross-correlation amplitudes statistically indistinguishable from those of the event windows, consistently across nine events, calls for a nontrivial explanation. The authors interpret this consistency as evidence that the most plausible physical explanation is the presence of a continuous correlated component in the strain data, with a stochastic gravitational-wave background as the most natural candidate. p. 6 This interpretation suggests that high-amplitude transient events may be superimposed on a continuous correlated background. p. 8
Future Testing and Constraints
The argument from detector independence points directly to a decisive observational test involving the KAGRA detector. If the correlated structure observed in our analysis reflects a physical stochastic background, it must appear in KAGRA data as well, with time delays consistent with the KAGRA baseline geometry. p. 7 A null result in KAGRA pairs would be equally informative, since it would constrain the range of viable physical and instrumental explanations. p. 8 The four-detector H1–L1–V1–K1 network provides the decisive experimental test of the hypothesis advanced in this paper. p. 8
Conclusion on Validation Criteria
The authors conclude that the inter-detector time delay is not a statistically robust standalone validation criterion under the present analysis. The behavior of the data under whitening shows that applying standard power-spectral-density whitening does not restore a clean separation between event and noise; it removes the only isolated excess in the sample, returning all three detector pairs to the correlation range defined by the surrounding noise. p. 8 This is consistent with a correlation structure that is not confined to the nominal event window but extends into the surrounding strain data. p. 8 The paper suggests that a continuous correlated component in the strain data, with a stochastic gravitational-wave background as the most natural candidate, is the most plausible physical explanation. p. 8 The analysis should be extended to include the KAGRA detector for a decisive test of this hypothesis. p. 7
--- Page 1 ---
On correlated noise in the LIGO-Virgo network: a test of the stochastic gravitational-wave background hypothesis S. C. Ulhoa,1, 2, ∗ B. p. 0 C. C. Carneiro,3, † F. p. 0 L. Carneiro,4, ‡ and L. p. 0 V. A. Cunha1, § 1 Instituto de Física, Universidade de Brasília
2 Canadian Quantum Research Center, 204-3002 32 Ave Vernon, BC V1T 2L7 Canada
3 Instituto Federal do Tocantins, 77760-000, Colinas do Tocantins, TO, Brazil
4 Universidade Federal do Norte do Tocantins, Centro de Ciências Integradas, 77824-838 Araguaína, TO, Brazil
(Dated: July 14, 2026) We revisit the inter-detector cross-correlation criterion used in gravitational-wave validation by extending the analysis of Creswell et al. [1] to nine LIGO–Virgo candidate events from the O2 and O3 observing runs. p. 0 For each event, we analyze all three detector pairs (H1×L1, H1×V1, L1×V1) and compare the normalized cross-correlation function C(τ) in the event window with an empirical ensemble of N = 200 surrounding off-source windows, using a band-pass-only pipeline. p. 0 In 26 of 27 pair-measurements, the event peak is statistically indistinguishable from the corresponding noise ensemble, including all pairs involving the independent Virgo detector. p. 0 The only exception is GW190412 in the H1×L1 pair. p. 0 We conclude that the cross-correlation statistic does not provide a robust standalone separation between event and noise. p. 0 We interpret the persistence of comparable correlation structure in the LIGO–Virgo pairs as evidence that the strain data may contain a continuous correlated physical component, for which a stochastic gravitational-wave background is a plausible candidate. p. 0
--- Page 2 ---
I. INTRODUCTION The direct detection of gravitational waves by the LIGO Scientific Collaboration in September 2015, reported in February 2016 [2], stands as one of the most consequential experimental results in modern physics. p. 0 The signal, designated GW150914 and interpreted as the merger of two stellar-mass black holes, provided the first direct confirmation of a prediction of Einstein’s general relativity that had remained unverified for a century. p. 0 The achievement was recognized with the Nobel Prize in Physics in 2017. p. 0 The road to that detection, however, was neither short nor uncontroversial. p. 0 The theoretical status of gravitational waves was itself disputed for decades after Einstein first derived them in 1916 [3]. p. 0 A central question was whether the waves carried physical energy or were merely coordinate artefacts of the linearized theory. p. 0 At the Chapel Hill conference in 1957, Richard Feynman introduced what became known as the “sticky bead argument”: a bead free to slide along a rod would be heated by a passing gravitational wave—an argument that gave the debate a particularly vivid and intuitive resolution [4, 5]. p. 0 The same period saw the question of whether gravitational radiation transports energy addressed in formal terms, notably by Trautman, who developed a treatment of the energy of gravitational radiation and showed it to be non-negative under appropriate asymptotic conditions [6]. p. 0 The modern mathematical formulation of the observable effects of gravitational waves in the linearized theory—including the geodesic deviation equation as the basis for detector response—was laid out by Pirani in 1956 [7]. p. 0 The first serious proposal for a gravitational-wave detector came from Joseph Weber in the 1960s, who constructed resonant-bar detectors and reported candidate detections [8, 9]. p. 0 Weber’s claims were ultimately not reproduced, but his work established the experimental program that would eventually lead to the kilometer-scale laser interferometers of today. p. 0 The LIGO detectors — two Michelson interferometers with 4 km arms located in Hanford, Washington, and Livingston, Louisiana — and the Virgo detector near Cascina, Italy were designed precisely to measure the fractional length change h = ∆L/L ∼ 10−21 induced by a passing gravitational wave. p. 0 GW150914 appeared to vindicate this design, and was widely regarded as closing the century-long controversy over the physical reality of gravitational radiation. p. 0 The validation of a gravitational-wave detection rests on two independent pillars. p. 0 The first is the matched-filter signal-to-noise ratio: the data are correlated with a bank of theoretical waveform templates derived from models of compact binary coalescence, incorporating generalrelativistic, post-Newtonian, and numerical-relativity ingredients, and a detection is claimed when the network SNR exceeds a threshold consistent with a prescribed falsealarm rate. p. 0 While the initial identification of a transient may be made by model-independent searches, as in the case of GW150914, whose first alert was produced by an unmodelled burst pipeline [2], and in particular the separation of signal from noise, rests on the matched filter: the LIGO-Virgo collaboration itself establishes that the residuals are consistent with Gaussian noise only after subtraction of the maximum-likelihood waveform. p. 0 The second, and conceptually simpler, pillar is the inter-detector time delay: if a real wave front propagates at the speed of light, it must arrive at spatially separated detectors with a lag τ consistent with the baseline geometry, at most ±10 ms for the H1–L1 pair, and up to ±27 ms for baselines involving Virgo. p. 1 For GW150914, the reported delay of τ ≈ 6. p. 19 ms between Livingston and Hanford was presented as a key piece of evidence for a genuine astrophysical signal [2]. p. 1 It was precisely this second criterion that Creswell, von Hausegger, Jackson, Liu, and Naselsky called into question in 2017. p. 1 Analyzing the publicly available strain data for GW150914, GW151226, and GW170104, they showed that the cross-correlation function C(τ) between the H1 and L1 detectors exhibits a peak at τ ≈ 7 ms not only in the event window, but also in noise windows before and after the event — and even in the residuals after subtraction of the best-fit waveform template. p. 1 If the inter-detector time delay appears in what the LVK pipeline designates as noise, it cannot serve as unambiguous evidence for a propagating gravitationalwave front. p. 2 The LIGO-Virgo collaboration responded to these findings in a technical note and subsequently in a peerreviewed publication [11]. p. 1 Their principal arguments were that the phase correlations reported by Creswell et al. p. 1 arise from spectral leakage due to the absence of a Tukey window before the FFT, and that the correlated residuals result from the use of a suboptimal template rather than the maximum-likelihood waveform. p. 1 With proper windowing and the optimal template, they argued, the residuals are consistent with Gaussian noise. p. 1 We find this response, while technically valid on its own terms, to reinforce rather than dissolve the underlying concern. p. 1 The matched filter — including the maximum-likelihood template — is constructed by correlating the data with waveforms derived from generalrelativistic models of compact binary coalescence. p. 1 To require the optimal template in order to render the residuals noise-like is to require that a model built from the theory be used to confirm a prediction of that same theory. p. 1 More pointedly: the physical effect of a gravitational wave passing through a detector network — the differential strain it induces, propagating at the speed of light — cannot, in principle, depend on what filter the analyst applies to the data. p. 1 A time delay that is physically real should be statistically distinguishable from the noise background without invoking the theoretical waveform. p. 1 If it is not, one faces a genuine dichotomy: either the event window contains nothing that the noise does not also contain, or the correlations present throughout the data reflect something more fundamental than instrumental noise — possibly a stochastic gravitational-wave background of cosmological or astrophysical origin. p. 1 The analysis of Creswell et al. p. 0 was restricted to the H1–L1 detector pair. p. 1 The two LIGO detectors share design heritage, data acquisition systems, and analysis pipelines; one could therefore argue that spurious correlations might arise from shared infrastructure rather than from any physical process. p. 1 The Virgo detector, operated by a separate European collaboration with independent hardware, software, and geographic location, provides a qualitatively different test. p. 1 Inter-detector correlations between a LIGO detector and Virgo cannot be attributed to shared processing infrastructure, and are therefore a more stringent probe of whether the noise itself carries a correlated time structure. p. 1
--- Page 3 ---
II. METHODOLOGY The standard LVK data analysis pipeline transforms the raw detector output — a strain time series s(t) = h(t)+n(t), where h(t) is the gravitational-wave strain and n(t) is the instrumental noise through a sequence of conditioning steps before any detection statistic is computed. p. 1 A Tukey (cosine-tapered) window is applied to the data segment prior to the discrete Fourier transform, suppressing spectral leakage between frequency bins. p. 1 A fourth-order Butterworth band-pass filter retains frequencies in the range [35, 350] Hz, where detector sensitivity is highest. p. 1 The data are then whitened by dividing each Fourier coefficient by the square root of the one-sided power spectral density Sn(f), estimated from an off-source segment, so that all frequencies contribute equally to subsequent statistics. p. 1 Narrow spectral lines from mechanical resonances, power-line harmonics, and intentionally injected calibration tones are removed by notch filters. p. 1 The primary detection statistic is the matched-filter signal-to-noise ratio, obtained by correlating the conditioned data with a bank of theoretical waveform templates h(t; θ) derived from models of compact binary coalescence, incorporating generalrelativistic, post-Newtonian, and numerical-relativity ingredients. p. 1 The product (ab) is defined by (ab) = 2 Z ∞ 0 a˜(f) b∗(f) + a˜∗(f) b(f) Sn(f) df. p. 2 (1). A candidate is promoted to a detection when the network SNR exceeds a threshold consistent with the target falsealarm rate, and when the inter-detector time delay between SNR peaks is consistent with the light-travel time between the relevant detector pair — at most ±10 ms for H1–L1 and up to ±27 ms for baselines involving Virgo. p. 2 This time delay is the key quantity under scrutiny in the present work. p. 2 Creswell et al. [1] subjected this framework to an independent analysis that identified a potentially serious problem. p. 2 Working with the publicly available strain data for GW150914, GW151226, and GW170104, they applied their own data-cleaning procedure — a fourth-order Butterworth band-pass filter over [35, 350] Hz plus notch filtering of calibration lines — and verified that the resulting strain differs from the LVK-cleaned data by no more than 3 × 10−4 in relative amplitude, confirming reproducibility. p. 2 They then computed the normalized crosscorrelation function between the Hanford and Livingston strain time series as a function of time lag τ: C(t, τ, w) = CorrH t+τ+w t+τ Lt+w t, (2) where Hb a and L b a denote the Hanford and Livingston data in the time interval [a, b], w is the window width, and Corr is the Pearson correlation coefficient, defined as Corr(x, y) = Cov(x, y) / sqrt(Cov(x, x) Cov(y, y), (3). p. 2 with covariances computed within the respective windows. p. 2 The lag τ is restricted to the physically meaningful range [−10 ms, +10 ms] for the H1–L1 baseline. p. 2 By varying τ, one shifts the Hanford record relative to Livingston and measures the degree of similarity as a function of delay. p. 2
--- Page 4 ---
III. RESULTS Table I summarises the cross-correlation statistics for all 27 valid detector-pair measurements across the nine events in our sample. p. 3 For each pair we report the signed peak correlation C∗ev, the corresponding lag τev, and the empirical p-value pemp defined in Eq. p. 3 (4). The dominant result is negative: in 26 of 27 pair-measurements, the event window produ… p. 3
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Improved detection of stochastic gravitational-wave background by implementing a model-independent approach where
The only pre-processing applied is a fourth-order Butterworth band-pass filter over [35, 350] Hz, applied forward and backward (scipy.signal.filtfilt) to avoid phase distortion.
This allows the system to test the null hypothesis of instrumental independence without relying on theoretical waveform assumptions or noise models. -
Enhanced discrimination between signal and noise by utilizing an empirical p-value calculation defined as
pemp = k + 1 / N + 1, k = card j: C∗j ≥ C∗ev, (4), where the Laplace correction prevents pemp = 0 for finite samples.
This provides a robust statistical measure to assess whether an event peak isstatistically indistinguishable from noise under this criterion
by comparing it against an ensemble of surrounding noise windows. -
Improved control over confounding factors by systematically testing detector independence through a four-detector network, specifically suggesting that
A four-detector H1–L1–V1–K1 network therefore provides the decisive experimental test of the hypothesis advanced in this paper.
This allows for a stringent test of whetherthe correlations we observe reflect a physical stochastic background
by observing if the structure appears in KAGRA data with expected time delays. -
Refined interpretation of outlier events by explicitly distinguishing between transient and background hypotheses, noting that
The existence of a loud event in H1×L1 does not contradict the background hypothesis; it is compatible with it.
This enables the system to recognize that "a sufficiently intense stochastic gravitational-wave background would manifest as a continuous correlated signal in all detectors, upon which high-amplitude transient events — whether astrophysical compact binary mergers or rare background fluctuations — would be superimposed." -
Better handling of preprocessing artifacts by demonstrating how whitening affects isolated excesses, stating that
Applying the standard power-spectral-density whitening does not restore a clean separation between event and noise; it removes the only isolated excess in the sample, that of GW190412.
This teaches the system that correlations extending into surrounding strain data are suppressed by whitening rather than being purely instrumental artifacts.
Sources
- On the time lags of the LIGO signals
- Observation of Gravitational Waves from a Binary Black Hole Merger
- A guide to LIGO-Virgo detector noise and extraction of transient gravitational-wave signals
- Open data from the third observing run of LIGO, Virgo, KAGRA and GEO
- GWTC-3: Compact Binary Coalescences Observed by LIGO and Virgo During the Second Part of the Third Observing Run
- GWTC-2: Compact Binary Coalescences Observed by LIGO and Virgo During the First Half of the Third Observing Run
- Overview of KAGRA: Detector design and construction history
Related papers
- Accurate waveforms for generic planar-orbit binary black holes: The multipolar effective-one-body model SEOBNRv6EHM
- Analytic Marginalization Over Calibration Uncertainties in Gravitational Wave Inference and Application to Black Hole Ringdown
- A catalog of long, eccentric and precessing binary black hole simulations from ICTS-TIFR
- Pole Skipping, Avoided Crossing, and Resonant Excitation in Kerr Quasinormal Modes near Algebraically Special Frequencies
- Gravitational Waves from Green's Function Decomposition for a Kerr black hole: I. Equatorial ISCO Plunge
- A New Spin on Dissipative Tides: First-Post-Newtonian Effects in Compact Binary Inspirals