The NANOGrav 15 yr Data Set: Targeted Searches for Supermassive Black Hole Binaries

Using the NANOGrav 15-year data set, this study presents the first targeted searches for continuous gravitational waves from 114 active galactic nuclei, finding no significant evidence for supermassive black hole binaries but demonstrating improved sensitivity through electromagnetic priors and establishing a roadmap for future multimessenger detections.

The Gist

Imagine the universe is a giant, cosmic concert hall. For years, astronomers have been listening to the "background hum" of this hall—a low, continuous rumble caused by thousands of pairs of supermassive black holes dancing around each other across the cosmos. This is the Gravitational Wave Background (GWB), and the NANOGrav collaboration recently confirmed its existence.

But today, the team isn't just listening to the crowd noise; they are trying to find specific soloists. They are hunting for individual pairs of black holes that are so loud and close that we can hear their specific "song" (a continuous gravitational wave) above the background hum.

Here is the story of their latest hunt, the NANOGrav 15-year Data Set, explained simply.

1. The Search Strategy: From "All-Sky" to "Targeted"

Imagine you are looking for a specific friend in a crowded stadium.

• The Old Way (All-Sky Search): You scan the entire stadium, looking at every seat. Because you don't know where your friend is sitting, you have to look everywhere, which makes it hard to spot them clearly.
• The New Way (Targeted Search): You have a ticket that says your friend is sitting in "Section 10, Row 5." You go straight there. You don't have to look at the whole stadium; you just focus on that one spot. This makes you much more likely to see them.

In this paper, the astronomers did exactly that. They took a list of 114 active galaxies (the "tickets") that looked suspicious. These galaxies showed strange, rhythmic flickering in their light (like a lighthouse blinking). Astronomers thought, "Maybe that flickering is caused by two black holes orbiting each other!"

They used the NANOGrav data (which listens to 68 ultra-precise cosmic clocks called pulsars) to specifically check those 114 spots for the gravitational wave "song" of a binary black hole.

2. The Results: A Quiet Night at the Concert

After checking all 114 targets, the result was: No soloists were found.

• The Good News: The search worked perfectly. By focusing on specific targets, they improved their sensitivity by a factor of 2.2 on average. It's like turning up the volume on a specific instrument so you can hear it better. They also set new, stricter limits on how heavy these black holes could be (ruling out some theories about a famous galaxy called 3C 66B).
• The "Almost" Moments: Two targets, nicknamed "Rohan" and "Gondor," showed a tiny, faint signal. It was like hearing a whisper that might be your friend, but it could also just be the wind.
  • When they ran the numbers, the signal was slightly louder than the background noise, but not loud enough to be sure.
  • When they accounted for the fact that they checked 114 different places (the "multiple comparisons" problem), those whispers turned out to be consistent with random noise. It's like finding a pattern in static on a radio; if you listen to enough stations, you'll eventually hear something that sounds like music, but it's just coincidence.

3. The Detective Work: Why They Doubt the "Whispers"

Even though the signals were weak, the team didn't just give up. They put "Rohan" and "Gondor" through a battery of detective tests to see if they were real or fake:

• The "Coherence" Test: A real black hole binary should make a specific pattern of ripples across the entire array of pulsars, like a wave hitting a beach in a synchronized way. The signals for these two candidates were a bit "out of sync," suggesting they might be noise.
• The "Dropout" Test: They asked, "If we remove one pulsar from the data, does the signal disappear?" For "Gondor," the signal relied heavily on a few pulsars that are known to be "noisy" (like a radio with bad reception). This suggests the signal might be an artifact of bad data, not a real black hole.
• The "Light Curve" Check: They looked at the light from these galaxies again. "Rohan" still flickered rhythmically, which is promising. But "Gondor" stopped flickering and started flaring randomly, which makes it look less like a stable binary system and more like a chaotic, single black hole.

4. Why This Matters (Even Without a Discovery)

You might ask, "If they didn't find a black hole, why write a paper?"

Think of this paper as building the roadmap for the future.

• The Map: They proved that "targeted searches" work. They showed that if we know where to look, we can find these signals much faster than scanning the whole sky.
• The Toolkit: They created a checklist of tests (coherence, dropout, light curves) that any future candidate must pass to be considered real.
• The Promise: As we get better telescopes and more pulsars (more "ears" to listen with), we will be able to hear these soloists clearly. The fact that they found almost signals means we are getting closer.

The Bottom Line

The NANOGrav team took a giant step forward by hunting for specific black hole pairs in 114 galaxies. While they didn't catch a "smoking gun" this time, they proved their new hunting strategy works. They found two "suspects" that were very close to being real, but after a rigorous investigation, they concluded those suspects were likely just innocent bystanders (noise).

This paper is a promise: We have the map, we have the tools, and we are getting louder. The first soloist is out there, and we will find it soon.

Technical Summary

1. Problem Statement

The primary goal of Pulsar Timing Array (PTA) experiments is to detect continuous gravitational waves (CWs) from individual Supermassive Black Hole Binaries (SMBHBs). While recent evidence for a stochastic Gravitational Wave Background (GWB) has strengthened, identifying specific individual sources remains a major challenge.

• The Localization Problem: All-sky CW searches suffer from poor spatial resolution (currently $\sim29$ deg$^2$), making it difficult to associate a GW signal with a specific host galaxy.
• The False Positive Problem: Electromagnetic (EM) surveys (e.g., CRTS) have identified hundreds of active galactic nuclei (AGN) with periodic or quasi-periodic variability, hypothesized to be SMBHBs. However, many of these are likely false positives caused by stochastic accretion noise or hydrodynamic variability rather than orbital motion.
• The Objective: This work aims to perform the first large-scale, targeted search for CWs using the NANOGrav 15-year data set. By leveraging EM priors (sky location, distance, redshift, and frequency) for 114 candidate AGN, the authors seek to:
  1. Improve sensitivity to CW strain compared to all-sky searches.
  2. Establish a rigorous statistical framework for validating low-significance candidates.
  3. Determine if any of the 114 candidates are genuine SMBHBs.

2. Methodology

The analysis utilizes the NANOGrav 15-year data set, comprising timing data for 68 millisecond pulsars with baselines up to 16 years.

Target Selection

The study focuses on 114 AGN candidates:

• CRTS Sample: 111 sources from the Catalina Real-Time Transient Survey exhibiting periodic optical variability.
• OVRO Sample: Two long-baseline periodic blazars (PKS 2131−021 and PKS J0805−0111).
• 3C 66B: A well-studied radio galaxy previously suggested as an SMBHB host.
• Assumption: The observed EM period is assumed to correspond directly to the binary orbital period ($P_{orb} = P_{obs}$), with the GW frequency set to $f_{GW} = 2/P_{orb}$.

Signal Model and Priors

• Fixed Parameters: Unlike all-sky searches, the targeted analysis fixes the sky position ($\hat{\Omega}$), luminosity distance ($D_L$), and GW frequency ($f_{GW}$) using EM data. This reduces the model dimensionality by four parameters, significantly enhancing sensitivity to the strain amplitude ($h_0$).
• Noise Modeling: The analysis models timing residuals as a sum of white noise, red noise, and a GW signal.
  • CURN Model: Common Uncorrelated Red Noise (autocorrelated background).
  • HD Model: Hellings–Downs correlated background (physically accurate for a GWB).
  • Strategy: Analyses are primarily run under the CURN model and then reweighted to the HD model using importance reweighting to account for cross-correlations between pulsars.

Validation and Robustness Tests

To distinguish genuine signals from noise artifacts, the authors employ a suite of tests (summarized in Table 2 of the paper):

1. Bayesian Model Comparison: Comparing three hypotheses: Noise-only, Coherent CW, and Incoherent CW (where each pulsar has an independent phase).
2. Coherence Tests:
   • Model Scrambling: Randomizing sky positions (sky shuffling) or phases (phase shifting) to create a null distribution of Bayes factors.
   • Three-Model Framework: Quantifying evidence for phase coherence across the array.
3. Dropout Analysis: Assigning a binary switch to each pulsar to determine if the signal is supported by the entire array or driven by a few noisy pulsars.
4. Random Targeting: Shuffling the parameters of the 114 targets to quantify the "trials factor" (probability of finding a false positive by chance when searching many targets).
5. GWB Consistency: Checking if candidate masses and frequencies align with population synthesis models and GWB anisotropy constraints.

3. Key Contributions

• Unified Targeted Search Framework: The paper presents a complete end-to-end pipeline for targeted CW searches, integrating EM priors with PTA data and a comprehensive suite of statistical validation tests.
• Improved Sensitivity: The study demonstrates that fixing EM priors improves strain upper limits by a median factor of 2.2 (mean 2.6) compared to all-sky searches at the same frequencies.
• Refined Constraints on 3C 66B: The analysis updates the chirp mass constraints for 3C 66B, ruling out a portion of the parameter space previously allowed by astrometric measurements.
• Roadmap for Future Detections: By analyzing two "best" candidates (J1536+0441 and J0729+4008) in extreme detail, the paper establishes a rigorous protocol for interpreting low-significance outliers in future multimessenger searches.

4. Results

Overall Search Statistics

• No Significant Detection: None of the 114 targets provided statistically significant evidence for a CW signal after accounting for the trials factor (multiple comparisons).
• Bayes Factors: The mean Bayes factor for the HD-correlated model was $0.73 \pm 0.32$. Two targets showed Bayes factors slightly above unity ($B > 1$), but these were consistent with noise fluctuations.

Case Studies: J1536+0441 and J0729+4008

These two sources were the top outliers and subjected to deep analysis:

• J1536+0441 (SDSS J153636.22+044127.0):
  • Signal: $B_{HD} \approx 1.91$. Coherence tests showed weak support ($p \approx 0.007$, $\sim 2.5\sigma$).
  • EM Context: Periodicity persists in extended light curves. However, spectral features (double-peaked H$\beta$) are stable but could be explained by a single SMBH with a structured broad-line region rather than a binary.
  • Mass: Inferred chirp mass is high ($\log_{10} M_c \approx 9.6$), placing it in a sparse, high-mass tail of population models.
• J0729+4008 (SDSS J072908.71+400836.6):
  • Signal: $B_{HD} \approx 3.7$. Coherence tests showed weak support ($p \approx 0.04-0.07$).
  • EM Context: The optical periodicity has faded/disappeared in recent data, replaced by stochastic flaring.
  • Noise: The signal is disproportionately supported by a few pulsars known for high-frequency noise (e.g., PSR J0613−0200 and J1713+0747), suggesting the signal may be a noise artifact.

Upper Limits

• 3C 66B: The 95% upper limit on the chirp mass is $M_c < 1.06(3) \times 10^9 M_\odot$ (HD model), improving upon previous limits and constraining the binary model from Iguchi et al. (2010).
• General Catalog: Strain upper limits ($h_0$) and chirp mass limits ($M_c$) are provided for all 114 targets in the appendices. The best improvement factor (15$\times$) was achieved for SDSS J141425.92+171811.2.

5. Significance and Future Outlook

• Current Sensitivity: The results demonstrate that while targeted searches significantly improve sensitivity, current PTA data is not yet sensitive enough to detect individual SMBHBs in the 114 candidate sample, even with EM priors. The lack of detection is consistent with population synthesis models predicting only a small number of genuine binaries in the CRTS sample (likely 5–8).
• Methodological Legacy: The paper defines a critical roadmap for the field. It proves that low-significance Bayes factors ($B \sim 1-4$) are not evidence of detection without rigorous coherence and dropout testing.
• Future Directions:
  • Longer Baselines: Future data releases (16+ years) will improve frequency resolution and sensitivity.
  • Multimessenger Follow-up: Continued EM monitoring (ZTF, WISE, spectroscopy) is essential to confirm or rule out binary interpretations (e.g., checking for stable periodicity vs. stochastic variability).
  • Refined Noise Models: Better modeling of pulsar-specific noise (chromatic effects, scattering) is required to reduce false positives in dropout analyses.
  • LISA Synergy: Detecting individual SMBHBs in the nHz band will provide "standard sirens" for cosmology and serve as precursors for the space-based LISA mission.

In conclusion, this work represents a mature transition from pilot studies to systematic surveys in PTA science. While no individual SMBHB was confirmed, the rigorous framework established here is essential for the eventual discovery of the first continuous gravitational wave source.