LIGO-Virgo-KAGRA O4 Run: Gravitational-Wave Advances

Updated 5 September 2025

LIGO-Virgo-KAGRA O4 Run is a coordinated gravitational-wave campaign featuring a four-detector network and major hardware upgrades that substantially enhance sensitivity.
Real-time detection pipelines and low-latency alert systems, bolstered by advanced algorithms and AI-driven glitch classifiers, enable prompt multi-messenger follow-ups.
Enhanced calibration, rigorous data-quality controls, and modern computing infrastructure underpin O4’s significant increase in detected events and more precise source localization.

The LIGO-Virgo-KAGRA O4 Run refers to the fourth full-scale coordinated science data-taking period of the ground-based gravitational-wave detector network comprising Advanced LIGO (Hanford and Livingston, USA), Advanced Virgo (Italy), and KAGRA (Japan). Spanning from May 2023 and scheduled to extend into October 2025, O4 constitutes the longest and most sensitive observing run to date for terrestrial kilometer-scale interferometric observatories, marking a critical juncture in the search for gravitational-wave (GW) events from compact binaries, stochastic backgrounds, and a wide array of astrophysical transients. Significant hardware upgrades, refined calibration, substantial improvements in detector characterization and low-latency alert infrastructure, and an expanded network layout have enabled order-of-magnitude enhancements in GW science yield, source characterization, sky localization, and multi-messenger astrophysics across the O4 campaign.

1. Detector Sensitivity Evolution and Network Configuration

The O4 run features substantial instrument upgrades and a four-detector configuration (HLVK: Hanford, Livingston, Virgo, KAGRA) (Collaboration et al., 2013, Capote et al., 21 Nov 2024). Upgraded Advanced LIGO detectors attained median binary neutron star (BNS) horizons of 152 Mpc (Hanford) and 160 Mpc (Livingston), with peak values up to 165 Mpc and 177 Mpc respectively in O4b (Capote et al., 21 Nov 2024). Advanced Virgo improved its BNS range to 90–120 Mpc; KAGRA’s commissioning trajectory projected a BNS range potentially up to 130 Mpc, although significant progress is implementation-dependent (Collaboration et al., 2013).

Enhancements were enabled by the A+ upgrade program: increased intracavity power (110 W input lasers, arm powers of 285–364 kW), deployment of a 300 m filter cavity for frequency-dependent squeezing yielding up to 6.1 dB broadband quantum noise reduction, mirror replacements to eliminate point absorbers, improved photodetector chains, and advanced suspension and control systems (Capote et al., 21 Nov 2024). KAGRA introduced the first fully functional photon calibration (Pcal) system in a cryogenic GW telescope, achieving ~0.79% amplitude calibration uncertainty—less than a third of O3 values (Chen et al., 17 Apr 2025).

The observable spacetime volume increases as $V \propto R^3$ , such that a moderate $50\%$ range improvement leads to a tripling of sensitive volume (Collaboration et al., 2013). These sensitivity gains underpin a sharp escalation in the expected number of detected GW events, with the network BNS event rate rising proportionally (Collaboration et al., 2013, López-Idiáquez et al., 2022, Colombo et al., 2022).

2. Real-Time Detection Pipelines and Low-Latency Alerts

O4 detection pipelines integrate both online (low-latency) and offline analyses to handle high event rates, robustly classify candidates, and facilitate prompt multi-messenger follow-up (Ewing et al., 2023, Alléné et al., 8 Jan 2025, Chaudhary et al., 2023). Key pipelines include GstLAL, MBTA, PyCBC Live, SPIIR, and for unmodeled transients, coherent WaveBurst (cWB) and PySTAMPAS (Ewing et al., 2023, Alléné et al., 8 Jan 2025, Collaboration et al., 16 Jul 2025).

GstLAL gained improved detection efficiency through advanced likelihood-ratio ranking leveraging an analytic matched-filter SNR–χ² (ξ²) signal model, signal removal from background histograms, and frequent (4 hr) updating of FAR background estimates (Ewing et al., 2023). MBTA extended its template bank to higher masses (up to 500 M $_\odot$ ), introduced advanced glitch rejection (SNR Excess method), and for the first time, incorporated single–detector and early–warning BNS searches (Alléné et al., 8 Jan 2025).

The O4 alert infrastructure dispatches public alerts (median ~30 s latency) via GraceDB, with classification probabilities for source types (BNS, NSBH, BBH), robust “astro” rankings, and Bayesian-derived sky maps (typically using BAYESTAR) (Chaudhary et al., 2023). “Early warning” triggers, particularly for BNS, may be issued several seconds before merger (e.g., median –3.1 s), facilitating rapid EM counterpart searches (Chaudhary et al., 2023).

3. Detector Characterization, Calibration, and Data Quality

Detector characterization (“DetChar”) and calibration have been substantially overhauled for O4, ensuring that data integrity and systematic uncertainties do not degrade science results (Arnaud, 2022, Capote et al., 21 Nov 2024, Chen et al., 17 Apr 2025). Automated rapid data-quality vetting (using Data Quality Reports, DQRs) accelerates alert confirmation/retraction and minimizes noise contamination (Arnaud, 2022). KAGRA pioneered cryogenic-compatible photon calibration, achieving an amplitude uncertainty of 0.79%, dominated by laser power calibration and beam-position estimation (verified with telephoto cameras) (Chen et al., 17 Apr 2025). LIGO achieved up to 6.1 dB squeezing, with quantum noise limited by optical loss and residual phase noise (Capote et al., 21 Nov 2024). New online noise-subtraction pipelines (e.g., NonSENS) and the use of AI-driven glitch classifiers (e.g., GravitySpy, iDQ) further improved data quality and event validation (Cesare, 24 May 2025).

4. Source Localization and Implications for Multi-Messenger Astronomy

A critical O4 advance is refined source sky localization. Simulations forecast a median 90% credible area as small as 33 deg² for BNS and 41 deg² for BBH (HLVK network), an order-of-magnitude improvement over O3 (which featured multi-hundred square degree localizations with three detectors) (Collaboration et al., 2013). The accuracy of three-dimensional localizations is enhanced due to improved amplitude and timing consistency, with constraints scaling as $\Delta\theta \propto 1/$ (SNR × Δf) (Collaboration et al., 2013).

These gains are key for multi-messenger campaigns. The higher rate and improved sky localization enable faster and more efficient electromagnetic and neutrino follow-up, increasing the likelihood of identifying kilonovae, GRBs, or other EM counterparts (Collaboration et al., 2013, Colombo et al., 2022). For BNS events, ~78% are anticipated to produce kilonovae, but detection is typically feasible only in the first 1–2 nights post-merger due to rapid optical fading (Colombo et al., 2022). Relativistic jet (GRB) emission remains difficult to observe (<2% of GW-detected events in prompt gamma rays; up to ~10% in deep radio afterglow searches), primarily due to jet beaming (Colombo et al., 2022).

5. Sensitivity Estimates, Detection Rates, and Population Impact

Formal O4 sensitivity quantification utilizes large-scale Monte Carlo injection campaigns, with >4.3×10⁸ signals during O4a alone, supporting precise measurement of detection efficiency, sensitive volume, and population rates as a function of source parameters and observing conditions (Essick et al., 14 Aug 2025). Detection probability for a population is computed via weighted importance sampling, enabling robust reweighting to alternate astrophysical models (Essick et al., 14 Aug 2025).

Updated catalog and population analyses (GWTC-4.0) report 218 CBC candidates (p_ast ≥ 0.5), more than doubling the event census relative to O3 (Collaboration et al., 25 Aug 2025). Updated local merger rates (90% C.I.) are $R_{\mathrm{BNS}} = 56^{+99}_{-40}$  Gpc⁻³ yr⁻¹, $R_{\mathrm{NSBH}}=36^{+32}_{-20}$  Gpc⁻³ yr⁻¹, and $R_{\mathrm{BBH}}=19^{+4}_{-2}$  Gpc⁻³ yr⁻¹ (Akyüz et al., 11 Jul 2025). High-SNR events (e.g., SNR ~80) in O4 support precision measurement of source properties and waveform effects, e.g., higher-order modes and precession (Akyüz et al., 11 Jul 2025).

6. Non-CBC Searches: Bursts, Continuous Waves, and the Stochastic Background

O4 extended search coverage to unmodeled (burst) and continuous GW sources. All-sky short-duration burst and long-duration transient searches utilized cWB, PySTAMPAS, and ML-based classifiers, achieving 2–10× better strain sensitivity (as low as $h_\mathrm{rss}$ ) vs. O3, but reported no statistically significant non-CBC candidates (Collaboration et al., 16 Jul 2025, Collaboration et al., 16 Jul 2025). For modeled core-collapse supernovae and Galactic neutron star glitches, O4 sensitivity now covers the full Milky Way for high-energy SN models and Vela-like f-mode glitch signals with Δν/ν as low as $2$– $6\times10^{-5}$ (Collaboration et al., 16 Jul 2025).

Stochastic background searches achieved a factor of up to two improvement, constraining $\Omega_\mathrm{GW}(25\,\mathrm{Hz})\leq 2.0\times10^{-9}$ (95% credibility) for CBC-like spectra, now close to the expected CBC background, and confirmed no evidence for nonstandard-polarization or terrestrial noise backgrounds (Collaboration et al., 28 Aug 2025).

7. Data Products, Open Science, and Computing Infrastructure

The O4a data release (May 2023–January 2024) includes calibrated strain h(t), derived from differential arm motion $\Delta L(t)$ as $h(t) = \Delta L(t)/L$ , along with alternate, noise-subtracted channels and quality masks. Auxiliary detector and environmental data and a curated GWTC-4.0 candidate catalog are publicly disseminated via the GW Open Science Center (GWOSC) (Collaboration et al., 25 Aug 2025).

To support this, the LVK network deployed a modern distributed computing infrastructure, leveraging HTCondor (workflow management), Rucio (data management), CVMFS (code/data distribution), and OSDF/xrootd federation (Bagnasco, 2023). Real-time and offline computing at multi-institution clusters enables low-latency alert issuance (often within one minute, or early warning before merger for long BNS events), rapid parameter estimation, and catalog production (Bagnasco, 2023, Chaudhary et al., 2023). Pipeline and data infrastructure are designed for scaling to higher event rates in future runs (O5 and beyond).

Overall, the O4 observing run constitutes a major advance in terrestrial gravitational-wave astronomy, combining high-sensitivity, dense multi-detector coverage, low-latency response, and enabling systematic, statistically robust GW signal catalogs, improved source parameter inference, and multi-messenger scientific opportunities. Ongoing and future data releases from O4 will continue to refine the astrophysical inference landscape, testing formation-channel models, cosmological standard siren measurements, fundamental physics, and the search for nonstandard GW phenomena.