⚛️ general relativity

A stepping stone toward detecting gravitational wave memory: a cumulative analysis with the full $(\ell=2, m=0)$ spherical harmonic using events from GWTC-4.0 and GWTC-5.0

This paper presents a Bayesian cumulative analysis of GWTC-4.0 and GWTC-5.0 binary black hole events using the IMRPhenomTHM_20 waveform model to test for the $(\ell=2, m=0)$ spherical harmonic mode, finding modest evidence for its presence and projecting that a significantly larger catalog of approximately 166 events will be required to achieve decisive statistical detection of gravitational wave memory.

Original authors: Maria Rosselló-Sastre, Sascha Husa, Yumeng Xu, Jorge Valencia, Joan Llobera-Querol, Antoni Ramos-Buades

Published 2026-07-07

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Maria Rosselló-Sastre, Sascha Husa, Yumeng Xu, Jorge Valencia, Joan Llobera-Querol, Antoni Ramos-Buades

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the universe is a giant, silent ocean. When two massive black holes crash into each other, they don't just make a splash; they send ripples across the fabric of space-time itself. These ripples are called gravitational waves.

For a long time, scientists have been listening to these waves with giant "ears" (detectors like LIGO and Virgo). They've mostly been listening to the loud, rhythmic "chirp" of the black holes spiraling together and merging. This is like hearing the main melody of a song.

However, the paper you're asking about is trying to hear something much quieter and stranger: the aftermath of that song.

The Two Parts of the "Sound"

The authors are looking for a specific part of the gravitational wave signal called the (2, 0) mode. To understand this, imagine a bell being struck:

The Ringing (Oscillatory Part): When you hit a bell, it vibrates and makes a sound that goes up and down in pitch. This is the "ringing" part of the black hole merger. It's loud and rhythmic.
The Permanent Dent (Memory Part): Imagine that after the bell stops ringing, the metal is slightly bent or dented. It doesn't go back to its original shape. In the universe, when black holes merge, they don't just vibrate; they permanently shift the space around them. This is called "displacement memory." It's a permanent, one-way "dent" in space-time that stays there forever.

The (2, 0) mode is the specific mathematical "note" that contains both the ringing and this permanent dent.

The Big Challenge: Hearing a Whisper

The problem is that the "permanent dent" (memory) is incredibly faint. It's like trying to hear a whisper in a hurricane. In fact, for any single black hole crash, the detectors are not sensitive enough to say, "Yes, we definitely heard that permanent dent." The signal is too weak, and the "noise" of the universe is too loud.

The Strategy: The "Crowd" Approach

Since they can't hear the whisper in one event, the authors decided to listen to many events at once.

Think of it like a choir. If one person hums a very quiet note, you might not hear it. But if 100 people hum that same quiet note at the same time, the sound adds up, and suddenly you can hear it clearly.

The authors took data from 84 black hole mergers (from a catalog called GWTC-4.0) and a few high-quality ones from a newer list (GWTC-5.0). They used a sophisticated computer model (a "waveform") that predicts what the sound should look like if the "permanent dent" exists. They then compared this model against the actual data.

What Did They Find?

No Single "Smoking Gun": Looking at each black hole crash individually, they couldn't find strong proof that the "permanent dent" was there. The evidence was too weak to be certain.
The "Crowd" is Getting Louder: When they added up the evidence from all 84 events together, the signal started to emerge. It wasn't a shout yet, but it was a clear "murmur."
- They calculated a "score" (called a Bayes factor) to see how much the data supported the idea that the "dent" exists. The score was positive, meaning the data does lean toward the idea that the dent is there, but it's not high enough to be considered a definitive discovery yet.
The "Ringing" Helps: They found that including the "ringing" part of the note (the oscillatory component) actually helped them hear the "dent" part better. It's like if the choir hummed a slightly louder note first, it helped the audience tune in to the quiet whisper that followed.

The Future: How Many More Do We Need?

The authors ran simulations to guess how many more black hole crashes they need to hear before they can say, "We definitely found the permanent dent!"

Current Status: They are close, but not there yet.
The Goal: They estimate they need about 166 events (give or take) to get a "decisive" proof. Since they have about 84 now, they are roughly halfway there.
The Catch: They warn that their computer models might still be missing some tiny details (like other weird vibrations in the black holes). Before they can claim a victory, they need to make sure their "ears" aren't tricked by these missing details.

The Bottom Line

This paper is a stepping stone. The authors haven't shouted "Eureka!" yet, but they have successfully built a ladder. They've shown that by listening to many black holes together, we are starting to hear the faint, permanent "dent" left behind by these cosmic collisions. It's a crucial step toward proving that space-time can be permanently warped by the violent dance of black holes, a prediction of Einstein's theory of General Relativity that has been very hard to catch in the act.

Technical Summary: A Cumulative Analysis of the (ℓ= 2, m = 0) Spherical Harmonic Mode in Gravitational Wave Data

Problem Statement
Accurate inference of source parameters in gravitational wave (GW) data requires waveform models that capture all measurable signal contributions. While the dominant (ℓ= 2, m = ±2) modes are well-characterized, subdominant modes play a critical role in breaking degeneracies for asymmetric or highly inclined systems. This paper focuses specifically on the (ℓ= 2, m = 0) mode, which contains two distinct physical contributions: an oscillatory ringdown component associated with the quasi-normal modes of the remnant black hole, and the non-oscillatory displacement memory. The displacement memory arises from the cumulative emission of gravitational radiation, manifesting as a secular, permanent change in spacetime. Detecting this memory is a significant challenge; previous observational analyses of catalogs such as GWTC-1 through GWTC-3 yielded no conclusive detections. Theoretical forecasts suggest that detection via individual events is unlikely with current second-generation detectors, requiring instead the stacking of hundreds to thousands of events. This work aims to assess the observational support for the full (2, 0) mode (including both oscillatory and memory components) in current GW data, serving as a stepping stone toward the eventual detection of gravitational wave memory.

Methodology
The authors perform a systematic reanalysis of binary black hole (BBH) events from the GWTC-4.0 catalog (84 events) and a selected subset of high-significance events from GWTC-5.0 (6 events). The analysis utilizes the IMRPhenomTHM_20 waveform model, a time-domain phenomenological model restricted to the quasi-circular, non-precessing sector, which includes the (2, 0) mode with both its oscillatory ringdown and displacement memory contributions. For the precessing candidate GW241127_061008, the precessing model IMRPhenomTPHM_20 is employed.

The core of the analysis involves two complementary approaches:

Bayesian Model Selection: For each event, a paired comparison is performed between two hypotheses: $H_{20}$ (IMRPhenomTHM_20 with the (2, 0) mode) and $H_{\emptyset}$ (IMRPhenomTHM without the (2, 0) mode). The Bayes factor ( $B$ ) is computed to quantify the relative evidence. Individual log-Bayes factors are stacked across the population assuming statistical independence to derive a cumulative evidence metric.
Signal-to-Noise Ratio (SNR) Stacking: The authors compute the SNR for the (2, 0) mode, separating it into its full, oscillatory-only, and memory-only contributions. These SNRs are stacked in quadrature across the event population to assess cumulative detectability.

The analysis incorporates measured power spectral densities (PSDs) and calibration uncertainties. To handle the non-periodic nature of displacement memory signals, the Fourier transform method gw-foutstep is used to avoid spurious artifacts. The study also includes injection studies (using both numerical relativity and phenomenological waveforms) to validate the scaling of Bayes factors with SNR and to check for biases in parameter estimation.

Key Contributions and Results

Individual Event Evidence: For the vast majority of events in GWTC-4.0 and the selected GWTC-5.0 events, the data do not provide statistically significant evidence for the inclusion of the (2, 0) mode. Most Bayes factors cluster around zero, with uncertainties overlapping the null value. The strongest individual preference is found for GW230814_230901, with a log-Bayes factor of $0.63 \pm 0.09$ , which constitutes "substantial" evidence according to the Jeffreys scale but remains below the threshold for "decisive" evidence.
Cumulative Evidence: The cumulative log-Bayes factor for the GWTC-4.0 catalog is calculated as $\log_{10} B = 1.38 \pm 0.79$ . While this indicates a mild trend toward the presence of the mode, the large uncertainty means the result does not constitute decisive evidence.
SNR Stacking: The cumulative SNR for the full (2, 0) mode reaches $3.53^{+0.94}_{-0.41}$ for the GWTC-4.0 sample, exceeding the commonly adopted detection threshold of SNR = 3. In contrast, the memory-only contribution reaches only $0.89^{+0.29}_{-0.11}$ after approximately 7.5 months of O4a observations. This suggests that including the oscillatory component significantly enhances the detectability of the mode compared to searching for memory alone.
Precessing Systems: The analysis of the precessing candidate GW241127_061008 using IMRPhenomTPHM_20 yields a log-Bayes factor of $-0.01 \pm 0.13$ , indicating no additional evidence for the (2, 0) mode when precession is accounted for.
Parameter Estimation: The inclusion of the (2, 0) mode does not produce appreciable shifts in the inferred source parameters (masses, spins, distances) for the studied events, suggesting that omitting this mode is not currently a dominant source of systematic bias in parameter estimation.
Detection Prospects: Based on fitting parametric distributions to the observed Bayes factors and uncertainties, the authors project that decisive statistical evidence ( $\log_{10} B - 1\sigma > 2$ ) would require a catalog of approximately $N_{events} = 166^{+82}_{-55}$ events, assuming current detector sensitivities and the specific waveform models used.

Significance and Claims
The paper positions this analysis as a "natural stepping stone" toward detecting gravitational wave memory. The authors argue that because the oscillatory component of the (2, 0) mode increases the signal-to-noise ratio, detecting the full mode is likely to precede the detection of the memory component in isolation. The results provide a realistic assessment of the current observational support for the (2, 0) mode, confirming that while the signal is present in the loudest events, it remains subdominant and requires population-level stacking for robust detection.

The authors explicitly note limitations and modest claims:

The study does not account for waveform systematics, such as the absence of the (3, ±2) mode in the phenomenological models used, which may affect Bayes factor estimates (as seen in the case of GW230814_230901).
The projection for the number of events required ( $N \approx 166$ ) is considered optimistic, as it relies on the specific assumptions of the current analysis and does not yet incorporate extensive systematics studies.
The paper concludes that while the (2, 0) mode is beginning to contribute measurably to the loudest events, robust event-level detection remains out of reach for present ground-based sensitivities, and future detection will likely require larger catalogs and more extensive waveform systematics studies.