🔭 astrophysics

Lomb-Scargle periodograms struggle with non-sinusoidal supermassive BH binary signatures in quasar lightcurves

This study demonstrates that the widely used Lomb-Scargle periodogram fails to detect the majority of supermassive black hole binaries because it is optimized for sinusoidal signals and performs poorly on the complex, non-sinusoidal pulse shapes predicted by hydrodynamical simulations, thereby highlighting the urgent need for advanced detection tools for future surveys like LSST.

Original authors: Allison Lin, Maria Charisi, Zoltan Haiman

Published 2026-01-15

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Allison Lin, Maria Charisi, Zoltan Haiman

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, noisy concert hall. In the center of this hall, two massive black holes (the "Supermassive Black Hole Binaries") are dancing around each other. As they spin, they pull gas from their surroundings, creating bright flashes of light. If we could watch this dance perfectly, we would see a rhythmic pattern of brightness, like a lighthouse beam sweeping by.

For years, astronomers have been trying to find these dancing pairs by looking at the light from distant quasars (super-bright galactic cores). They use a very popular tool called the Lomb-Scargle Periodogram (LSP). Think of the LSP as a specific type of metal detector. It is excellent at finding smooth, round metal coins (which represent smooth, sinusoidal waves, like a perfect sine wave).

However, this new paper argues that the "coins" we are looking for might not be round at all. According to computer simulations of how these black holes dance, the light they produce doesn't look like a smooth wave. Instead, it looks like a sawtooth—a sharp, jagged spike that rises quickly and then slowly fades away, like the teeth of a saw or a heartbeat on a monitor.

Here is what the authors found when they tested their "metal detector" (the LSP) against these "sawtooth" signals:

1. The Metal Detector is Tuned for the Wrong Shape

The LSP is designed to look for smooth, rolling waves. When the researchers simulated light curves with smooth waves, the detector found them about 45% of the time in realistic data (and even less in high-quality future data).

But when they simulated the "sawtooth" signals (the jagged, realistic shape predicted by physics), the detector was almost blind.

In realistic data similar to past surveys, it only found 9% of the sawtooth signals.
In ideal, high-quality data, it found only 1%.

The Analogy: Imagine trying to find a jagged, broken piece of glass using a magnet. The magnet is great at finding smooth iron nails, but it completely ignores the glass. The LSP is the magnet; the sawtooth signal is the glass.

2. Why Did Past Searches Miss Them?

The authors suggest that previous searches for these binary black holes likely missed the vast majority of them. Because the LSP is so bad at spotting jagged signals, even if a binary black hole system was right there, shining its sawtooth light, the tool would have dismissed it as random noise.

It's like trying to find a specific song on the radio by only listening for a perfect, smooth hum. If the song has a sharp, rhythmic beat (the sawtooth), your "smooth hum" filter will tell you, "No, that's just static."

3. The "Red Noise" Problem

The universe isn't quiet; it's full of "red noise." This is a type of background static that gets louder at lower frequencies, making it hard to hear the signal. The LSP assumes the background is "white noise" (like static on an old TV), which is a poor assumption for quasars.

Because the tool assumes the wrong kind of background noise, it gets confused. It tries to fit a smooth curve to a jagged signal, and the math just doesn't work out. The authors note that even when the data is perfect and clear (like what we expect from the upcoming LSST telescope), the tool still fails to find these signals because the tool itself is the weak link, not the data quality.

4. A Surprising Twist

Interestingly, the tool actually worked slightly better on the messy, sparse data from past surveys (like the Palomar Transient Factory) than on the perfect, high-quality data they simulated for the future.

Why? In the messy data, there are fewer data points. Sometimes, by pure luck, the few points we have happen to land right on the "peaks" and "valleys" of the wave, making it easier to draw a smooth line through them.
In the perfect, dense data, we see every little wiggle and jagged edge. The tool sees the jaggedness clearly and realizes, "This isn't a smooth wave," and gives up.

The Bottom Line

The paper concludes that the most popular tool astronomers use to find these dancing black holes is fundamentally flawed for the job. It is optimized for smooth waves, but the universe likely produces jagged sawtooth waves.

If we want to find these binary black holes in the future, especially with the massive amount of data coming from the Rubin Observatory (LSST), we cannot just keep using the same old "metal detector." We need to build new tools specifically designed to recognize jagged, sawtooth patterns, or we will continue to miss the vast majority of these cosmic dances.

Technical Summary: Lomb-Scargle Periodograms Struggle with Non-Sinusoidal Supermassive BH Binary Signatures in Quasar Lightcurves

Problem Statement
Supermassive black hole binaries (SMBHBs), formed following galaxy mergers, are expected to exhibit periodic electromagnetic variability due to mechanisms such as periodic accretion, Doppler boosting, and self-lensing. While hydrodynamical simulations predict these binaries produce complex, non-sinusoidal pulse shapes (e.g., sawtooth patterns), previous observational searches for SMBHB candidates in quasar lightcurves have predominantly relied on detecting sinusoidal periodicity. The standard tool for these searches, the Lomb-Scargle periodogram (LSP), is optimized for sinusoidal signals and assumes white noise, whereas quasar variability is characterized by red noise (stochastic variability). This paper investigates the efficacy of the LSP in detecting non-sinusoidal periodicities, specifically sawtooth signals, within realistic quasar lightcurves, questioning whether previous searches have missed the majority of binary systems.

Methodology
The authors conducted a comprehensive simulation study to assess the recovery rates of periodic signals under varying survey conditions.

Sample: The study utilized a subset of 12,400 quasars from the Palomar Transient Factory (PTF) catalog (originally analyzed in Charisi et al. 2016), for which Supermassive Black Hole (SMBH) masses and absolute magnitudes were available.
Lightcurve Simulation: For each quasar, six mock lightcurves were generated by combining:
1. Intrinsic Noise: Modeled using a Damped Random Walk (DRW) process, with parameters ( $\sigma$ and $\tau$ ) estimated via empirical scaling relations based on SMBH mass and redshift.
2. Periodic Signals: Two types were injected: (a) sinusoidal and (b) sawtooth (asymmetric sharp rise, linear decrease), with amplitudes, periods, and phases drawn from uniform distributions.
3. Observing Scenarios: Three distinct sampling patterns were simulated:
  - Idealized: Daily sampling with Gaussian photometric errors.
  - PTF-like: Downsampling of the idealized curves to match actual PTF observation times and error characteristics.
  - LSST-like: 10-year baselines with 5-day cadence (perturbed by Gaussian noise), annual gaps, and photometric errors corresponding to magnitude 21.
Detection Pipeline: The generalized Lomb-Scargle periodogram was applied to detect periodicity. Statistical significance was assessed by comparing observed peak heights against a null distribution generated from 100,000 DRW-only simulations per frequency bin to account for the "look-elsewhere effect" and red noise properties.

Key Results
The study quantified the fraction of injected signals successfully recovered as statistically significant:

Sinusoidal Recovery: The LSP detected only a minority of sinusoidal signals, with recovery rates of $\sim45\%$ for PTF-like lightcurves, $\sim24\%$ for idealized lightcurves, and $\sim23\%$ for LSST-like lightcurves.
Sawtooth Recovery: The detection efficiency dropped precipitously for non-sinusoidal signals. Recovery rates were $\sim9\%$ for PTF-like, $\sim1\%$ for idealized, and $\sim1\%$ for LSST-like lightcurves.
Period Accuracy: For detected signals, the recovered periods were generally accurate (within 15% of the injected value) for sinusoids ( $\sim97-99\%$ ) but less so for sawtooth signals ( $\sim80-91\%$ ), partly due to secondary Fourier components in the sawtooth power spectrum.
Parameter Dependence: Recovery rates increased monotonically with signal-to-noise ratio (amplitude relative to DRW noise) and decreased with period length. Long-period sawtooth signals ( $>250$ days) were effectively undetectable in idealized and LSST-like scenarios.
PTF Anomaly: Counter-intuitively, the PTF-like sparse sampling yielded higher recovery rates than the higher-quality, longer-baseline LSST-like or idealized data. The authors attribute this to the sparse sampling occasionally aligning fortuitously with the extrema of the red-noise-dominated lightcurves, allowing a sinusoidal fit to pass through the data points more easily than in densely sampled, noise-rich datasets.

Significance and Conclusions
The paper concludes that the Lomb-Scargle periodogram is a sub-optimal statistic for detecting SMBHBs in quasar lightcurves, particularly when the underlying signal is non-sinusoidal.

Missed Candidates: The extremely low recovery rates for sawtooth signals ( $\sim1-9\%$ ) imply that previous periodicity searches, including those in PTF, CRTS, and ZTF, have likely missed the vast majority of SMBHB candidates if they produce the complex pulse shapes predicted by hydrodynamical simulations.
Limitations of Current Methods: Even for sinusoidal signals, the LSP fails to recover the majority of injected periodicities due to the mismatch between the white-noise assumption of the LSP and the red-noise nature of quasar variability.
Implications for LSST: As the Rubin Observatory's Legacy Survey of Space and Time (LSST) prepares to provide millions of high-quality quasar lightcurves, the authors argue that relying on standard LSP searches will result in a highly incomplete sample of binaries.
Future Directions: The paper advocates for the development of advanced periodicity detection tools that explicitly account for red noise (e.g., using Bayes factors or matched filtering) and utilize templates that match non-sinusoidal pulse shapes to improve detection sensitivity and reduce false positives in upcoming time-domain surveys.

1. The Metal Detector is Tuned for the Wrong Shape

2. Why Did Past Searches Miss Them?

3. The "Red Noise" Problem

4. A Surprising Twist

The Bottom Line

More like this