Optimal Strategies for Gravitational-Wave Memory Detection with Noise Marginalization in Pulsar Timing Arrays

arXiv:2610.20772 · gr-qc, astro-ph.IM · Submitted 2026-10-14 · Read on arXiv

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: We're looking at the paper: "Optimal Strategies for Gravitational-Wave Memory Detection with Noise Marginalization in Pulsar Timing Arrays".

Mira: Here's the gist of the paper. Current pulsar timing arrays could detect gravitational wave memory, which would provide a unique test of general relativity The optimal statistic for the GWB was the testbed for developing noise marginalized techniques In this work, it was shown that noise marginalization of the optimal statistic reduces the signal recovery bias due to covariance between the intrinsic red noise in the individual pulsars and the GWB As a result, noise marginalization has become the standard for GWB frequentist work Methodological Comparison The paper investigates whether noise-marginalization (NM) is necessary when searching for gravitational wave memory signals in pulsar timing arrays (PTAs) The authors compare the NM method to the simpler maximum posterior (MP) method to determine its effectiveness They test these methods on simulated PTA datasets with red and white noise content, as well as the NANOGrav 15-year data set They find no significant variations in their detection statistics between the noise-marginalized and the maximum-posterior methods for the simulated datasets, and only small variations in the NANOGrav 15-year comparisons The conclusion is that for realistic PTAs, noise marginalization could be useful for assigning a false alarm probability to Bayesian results, though the maximum posterior method is sufficient for identifying potentially detectable signals Signal Modeling and Statistics The analysis begins by describing the form of a GW memory signal in a PTA and giving an overview of the frequentist Fc and Fi statistics derived in

33: The full time-series residuals for a single pulsar in a PTA are modeled as delta t = delta tmem + Mϵ + n + Fa Red noise processes are modeled using a power law PRN (f) = A squared RN f(-γ) with an amplitude ARN at the reference frequency f1yr and spectral index γ The Fi statistic is derived by maximizing the log-likelihood ratio, which involves defining the likelihood p(δtδtmem) The incoherent Fi statistic treats each pulsar’s TOAs as independent, yielding a total Fi = PN Fα where Fα is the value for each pulsar The average over many noise realizations of 2Fi follows a χ2 distribution with N degrees of freedom and non-centrality parameter ρ squared = (˜δtmem˜δtmem) Simulated Data Analysis The authors test the Fi statistic across three simulated data sets, varying the GW memory signal amplitude and red noise content The low amplitude simulation results show no significant bumps in the distribution of Fi and FAP across the t0 range, which is expected with the strength of GW memory present The mid amplitude signal shows a peak where there had been a valley around the expected epoch t0 = 58 109 MJD, but the FAP values are still around the 0.1% level, indicating it is not enough to claim a detection The large GW memory injection in Figure 3 shows an unmistakable peak centered around the correct t0 value with an FAP value around 10-13 for both detection methods Realistic Data Set Evaluation When analyzing the NANOGrav 15-year data set, both MP and NM methods show non-detection results consistent with previous Bayesian findings In the realistic results, there is more variation between methods, though both show non-detection results that are consistent with the Bayesian findings On closer inspection of the burst epoch preferred by the Bayesian results, NM results show more significance at that time than the MP method The NM method seems to still more closely align with the Bayesian results and could be useful for assigning FAP to such Bayesian peaks in future analysis The MP method is sufficient as a detection statistic without noise marginalization Across all simulated data sets, with different GW memory significances and different red noise contents, the MP and NM methods produce equivalent results Both methods produce FAPs that effectively indicate the level of confidence expected for signal detection This gives a baseline for how a GW memory signal would appear in future PTA analysis, indicating that the MP method is sufficient as a detection statistic without noise marginalization The paper concludes that the MP method is sufficient for GW memory detection analysis using the Fi statistic The authors also note that in simulated results, when comparing the two methods of estimating red noise parameters, the MP method values almost completely overlap with the mean values of the NM method In each simulated data set, the resulting inference possible from the Fi statistic does not change when using either method The paper ultimately finds that for GW memory detection statistics, noise marginalization is not strictly required for signal identification The authors also note that in realistic results, there is more variation between methods, though both methods show non-detection results consistent with the Bayesian findings This suggests that the MP method is sufficient for GW memory detection analysis using the Fi statistic The paper states that the NM method seems to still more closely align with the Bayesian results and could be useful for assigning FAP to such Bayesian peaks in future analysis Acknowledgments The NANOGrav collaboration receives support from National Science Foundation (NSF) Physics Frontiers Center award 2020265 The authors would like to thank Jerry Sun for the development of the original Fi statistic code that they built off of The paper is 2610.20772.

Kai: Every sentence must end with a reference token Page 1 --- Optimal Strategies for Gravitational-Wave Memory Detection with Noise Marginalization in Pulsar Timing Arrays Rand Burnette 1 and Xavier Siemens 1, 2 Department of Physics, Oregon State University, Corvallis, OR 97331, USA 2Center for Gravitation, Cosmology and Astrophysics, Department of Physics, University of Wisconsin-Milwaukee, (Dated: October 16, 2026) Current pulsar timing arrays could detect gravitational wave memory In order to optimize our searches for gravitational wave memory in pulsar timing arrays, we integrate the optimal frequentist statistics for GW memory into the enterprise code suite We additionally test a noise marginalized version of this statistic, and compare it’s effectiveness to the simpler maximum posterior method We perform these tests on simulated pulsar timing array with red and white noise content, as well as the NANOGrav 15-year data set We find no significant variations in our detection statistics between the noise-marginalized and the maximum-posterior methods for the simulated datasets, and only small variations in the NANOGrav 15-year comparisons We conclude that for realistic PTAs the noise-marginalization method could be useful for assigning a false alarm probability to Bayesian results, though the maximum posterior method is sufficient for identifying potentially detectable signals I.

Mira: INTRODUCTION Pulsar Timing Arrays (PTA) use precise timing of radio pulses from millisecond pulsars (MSPs) with ultrastable rotational periods

1: to search for gravitational waves (GW) In recent years, we have found strong evidence for the presence of a gravitational wave background (GWB)

2–8: The most likely source for the GWB is a population of supermassive binary black holes

9–11: , but it could also arise from beyond the standard model physics in the early universe

12, 13: PTA data have also been searched for continuous waves

14–17: , GW memory

18–22: , and generic bursts

23–25: as potential deterministic GW signals we could see beyond the GWB Over the past two decades, the PTA community has devoted a great deal of effort to develop and refine gravitational wave detection techniques in PTAs Bayesian analyses are robust to the broad noise posteriors typical of PTA analyses, but can be computationally expensive Frequentist analyses, on the other hand, are computationally efficient because we can analytically maximize the likelihood ratio over (at least) some of the signal parameters, and avoids the Bayesian marginalization of high dimensional integrals For fixed values of the noise parameters, however, they can be biased, and a hybrid Bayesian-frequentist technique called noise-marginalization can be used to mitigate these biases The optimal statistic for the GWB was the testbed for developing noise marginalized techniques, as the most detectable GW signal in PTAs In this work, it was shown that noise marginalization of the optimal statistic reduces the signal recovery bias due to covariance between the intrinsic red noise in the individual pulsars and the GWB As a result, noise marginalization has become the standard for GWB frequentist work As PTAs prepare to search for signals beyond the GWB, we are working to bring the frequentist methods used for various deterministic GW signals on par with the GWB optimal statistic Very recently

37: developed a noise-marginalized version of the Fp statistic introduced in

14: for continuous wave (CW) searches Surprisingly, unlike the results for the GWB optimal statistic, the implementation of noise marginalization into the Fp statistic resulted in no tangible difference in signal significance recovery Another potentially detectable signal in PTAs is GW memory, which is a nonlinear effect resulting in a constant shift in the space time metric due to violent GWemitting events, such as mergers of super massive black hole binaries (SMBHBs)

38–41: In PTA we measure the times of arrival (TOAs) of these radio pulses, and compare them to our predicted model for when we expect these pulses to arrive GW memory appears as a sudden change in the rotational frequency of the pulsar that results in a linear drift in our timing residuals which we will refer to as a ramp In this paper, we study the robustness of frequentist techniques to search for GW memory We start from the frequentist statistics developed for GW memory in

33: , and build upon this work to include noisemarginalization as was done for the GWB and continuous waves in

35, 37: We apply this method to the incoherent Fi statistic for GW memory, then compare its precision and efficacy with the standard method, which uses arXiv:2610.20772v1

gr-qc: 14 Oct 2026 Page 3 --- 3 II.

Kai: METHODS We begin by describing the form of a GW memory signal in a PTA, and giving a brief overview of the form of the frequentist Fc and Fi statistics derived in

33: We then detail the noise-marginalization (NM) process, and how we compare it to the maximum-posterior (MP) method A.

Paper summary: Kai: So to wrap up this paper "Optimal Strategies for Gravitational-Wave Memory Detection with Noise Marginalization in Pulsar Timing Arrays," they've shown that noise marginalization isn't strictly required just to identify the gravitational wave memory signal itself <ref:2610.20772#pg6>

Mira: They conclude that the maximum posterior method is sufficient for GW memory detection analysis using the Fi statistic, even though they found more variation between methods on realistic data sets <ref:2610.20772#pg2>

Lev: So if we are designing our experiments for real pulsar timing arrays, we can probably stick with the maximum posterior method for finding a potential signal, but keep an eye on that noise marginalization approach when we want to assign false alarm probabilities based on Bayesian findings <ref:2610.20772#pg1>

Kai: Right. It's about having a clear baseline where the simpler method works for detection, but the more complex one offers something extra for deeper analysis <ref:2610.20772#pg2>

Conclusion: Kai: So this paper, "Optimal Strategies for Gravitational-Wave Memory Detection with Noise Marginalization in Pulsar Timing Arrays," it basically looks at how we search for gravitational wave memory signals using pulsar timing arrays <ref:2610.20772#pg3>

Mira: It tackles a problem there, trying to figure out if adding noise marginalization helps us find the signal better, specifically when comparing it to the simpler maximum posterior method <ref:2610.20772#pg4>

Lev: The core idea is that noise marginalization can reduce bias caused by how the intrinsic pulsar timing noise mixes with the gravitational wave background <ref:2610.20772#pg2>

Kai: They tested this on simulated data and real NANOGrav data, and for those simulated sets, they found no big difference between the two methods for detecting a signal <ref:2610.20772#pg5>

Mira: But when you look at realistic data, they suggest that noise marginalization might be useful if you want to assign false alarm probabilities to Bayesian results <ref:2610.20772#pg1>

Lev: So for actual hardware running these searches, the maximum posterior method seems sufficient just to find a signal, but the noise marginalization approach could be good for assigning confidence levels later <ref:2610.20772#pg4>

Kai: It seems like they've given us a baseline then, showing that the simpler method is okay for finding what's out there <ref:2610.20772#pg5>

Mira: But it opens up this path for how we interpret our findings in future PTA studies, especially when you look at the confidence levels of those detections <ref:2610.20772#pg1>

Lev: So where does this leave us? We have a solid frequentist method that works well, and a more complex Bayesian way to handle noise that might give us better confidence metrics down the line <ref:2610.20772#pg2>

Rand Burnette 1, Xavier Siemens 1, 2

Department of Physics, Oregon State University · Center for Gravitation, Cosmology and Astrophysics, Department of Physics, University of Wisconsin-Milwaukee

gr-qc, astro-ph.IM

Submitted: 2026-10-14

Updated: 2026-11-06

Comments: 8 page paper on methods development for gravitational wave memory searches in pulsar timing arrays, tested with both simulated and real datasets

License: http://creativecommons.org/licenses/by-sa/4.0/

Importance score: 83/100

The gist: Current pulsar timing arrays could detect gravitational wave memory, which would provide a unique test of general relativity The optimal statistic for the GWB was the testbed for developing noise…

Key concepts

Gravitational Wave Memory
This is a nonlinear effect where violent gravitational wave events, like black hole mergers, cause a permanent shift in the spacetime metric. In pulsar timing arrays (PTAs), this manifests as a sudden change in the pulsar's rotational frequency, creating a linear drift known as a 'ramp' in timing residuals.
Maximum Posterior (MP) Method
This is one frequentist method used to estimate noise parameters by finding the maximum likelihood value obtained within the Bayesian chain. It is computationally efficient and is sufficient for identifying potentially detectable signals in this study.
Noise Marginalization (NM) Method
This more intensive method involves sampling many random sets of red noise parameters from the posterior distribution and calculating the statistic for each set. It is used to reduce signal recovery bias caused by covariance between intrinsic pulsar noise and the gravitational wave background.
Fi Statistic
This is a specific frequentist statistic derived from maximizing the log-likelihood ratio to detect GW memory. The average of this statistic over many noise realizations follows a chi-squared distribution, allowing researchers to calculate false alarm probabilities (FAPs) for detection.

Terminology

Summary

Current pulsar timing arrays could detect gravitational wave memory, which would provide a unique test of general relativity The optimal statistic for the GWB was the testbed for developing noise marginalized techniques In this work, it was shown that noise marginalization of the optimal statistic reduces the signal recovery bias due to covariance between the intrinsic red noise in the individual pulsars and the GWB As a result, noise marginalization has become the standard for GWB frequentist work

Methodological Comparison

The paper investigates whether noise-marginalization (NM) is necessary when searching for gravitational wave memory signals in pulsar timing arrays (PTAs) The authors compare the NM method to the simpler maximum posterior (MP) method to determine its effectiveness They test these methods on simulated PTA datasets with red and white noise content, as well as the NANOGrav 15-year data set They find no significant variations in their detection statistics between the noise-marginalized and the maximum-posterior methods for the simulated datasets, and only small variations in the NANOGrav 15-year comparisons The conclusion is that for realistic PTAs, noise marginalization could be useful for assigning a false alarm probability to Bayesian results, though the maximum posterior method is sufficient for identifying potentially detectable signals

Signal Modeling and Statistics

The analysis begins by describing the form of a GW memory signal in a PTA and giving an overview of the frequentist Fc and Fi statistics derived in [33] The full time-series residuals for a single pulsar in a PTA are modeled as delta t = delta tmem + Mϵ + n + Fa Red noise processes are modeled using a power law PRN (f) = A squared RN f(-γ) with an amplitude ARN at the reference frequency f1yr and spectral index γ The Fi statistic is derived by maximizing the log-likelihood ratio, which involves defining the likelihood p(δtδtmem) The incoherent Fi statistic treats each pulsar’s TOAs as independent, yielding a total Fi = PN Fα where Fα is the value for each pulsar The average over many noise realizations of 2Fi follows a χ2 distribution with N degrees of freedom and non-centrality parameter ρ squared = (˜δtmem˜δtmem)

Simulated Data Analysis

The authors test the Fi statistic across three simulated data sets, varying the GW memory signal amplitude and red noise content The low amplitude simulation results show no significant bumps in the distribution of Fi and FAP across the t0 range, which is expected with the strength of GW memory present The mid amplitude signal shows a peak where there had been a valley around the expected epoch t0 = 58 109 MJD, but the FAP values are still around the 0.1% level, indicating it is not enough to claim a detection The large GW memory injection in Figure 3 shows an unmistakable peak centered around the correct t0 value with an FAP value around 10-13 for both detection methods

Realistic Data Set Evaluation

When analyzing the NANOGrav 15-year data set, both MP and NM methods show non-detection results consistent with previous Bayesian findings In the realistic results, there is more variation between methods, though both show non-detection results that are consistent with the Bayesian findings On closer inspection of the burst epoch preferred by the Bayesian results, NM results show more significance at that time than the MP method The NM method seems to still more closely align with the Bayesian results and could be useful for assigning FAP to such Bayesian peaks in future analysis The MP method is sufficient as a detection statistic without noise marginalization

Conclusion on Statistical Sufficiency

Across all simulated data sets, with different GW memory significances and different red noise contents, the MP and NM methods produce equivalent results Both methods produce FAPs that effectively indicate the level of confidence expected for signal detection This gives a baseline for how a GW memory signal would appear in future PTA analysis, indicating that the MP method is sufficient as a detection statistic without noise marginalization The paper concludes that the MP method is sufficient for GW memory detection analysis using the Fi statistic The authors also note that in simulated results, when comparing the two methods of estimating red noise parameters, the MP method values almost completely overlap with the mean values of the NM method In each simulated data set, the resulting inference possible from the Fi statistic does not change when using either method The paper ultimately finds that for GW memory detection statistics, noise marginalization is not strictly required for signal identification The authors also note that in realistic results, there is more variation between methods, though both methods show non-detection results consistent with the Bayesian findings This suggests that the MP method is sufficient for GW memory detection analysis using the Fi statistic The paper states that the NM method seems to still more closely align with the Bayesian results and could be useful for assigning FAP to such Bayesian peaks in future analysis

Acknowledgments

The NANOGrav collaboration receives support from National Science Foundation (NSF) Physics Frontiers Center award 2020265 The authors would like to thank Jerry Sun for the development of the original Fi statistic code that they built off of The paper is 2610.20772. Every sentence must end with a reference token

--- Page 1 ---

Optimal Strategies for Gravitational-Wave Memory Detection with Noise

Marginalization in Pulsar Timing Arrays

Rand Burnette 1 and Xavier Siemens 1, 2

Department of Physics, Oregon State University, Corvallis, OR 97331, USA

2Center for Gravitation, Cosmology and Astrophysics, Department of Physics, University of Wisconsin-Milwaukee, (Dated: October 16, 2026) Current pulsar timing arrays could detect gravitational wave memory In order to optimize our searches for gravitational wave memory in pulsar timing arrays, we integrate the optimal frequentist statistics for GW memory into the enterprise code suite We additionally test a noise marginalized version of this statistic, and compare it’s effectiveness to the simpler maximum posterior method We perform these tests on simulated pulsar timing array with red and white noise content, as well as the NANOGrav 15-year data set We find no significant variations in our detection statistics between the noise-marginalized and the maximum-posterior methods for the simulated datasets, and only small variations in the NANOGrav 15-year comparisons We conclude that for realistic PTAs the noise-marginalization method could be useful for assigning a false alarm probability to Bayesian results, though the maximum posterior method is sufficient for identifying potentially detectable signals I. INTRODUCTION Pulsar Timing Arrays (PTA) use precise timing of radio pulses from millisecond pulsars (MSPs) with ultrastable rotational periods [1] to search for gravitational waves (GW) In recent years, we have found strong evidence for the presence of a gravitational wave background (GWB) [2–8] The most likely source for the GWB is a population of supermassive binary black holes [9–11], but it could also arise from beyond the standard model physics in the early universe [12, 13] PTA data have also been searched for continuous waves [14–17], GW memory [18–22], and generic bursts [23–25] as potential deterministic GW signals we could see beyond the GWB Over the past two decades, the PTA community has devoted a great deal of effort to develop and refine gravitational wave detection techniques in PTAs Bayesian analyses are robust to the broad noise posteriors typical of PTA analyses, but can be computationally expensive Frequentist analyses, on the other hand, are computationally efficient because we can analytically maximize the likelihood ratio over (at least) some of the signal parameters, and avoids the Bayesian marginalization of high dimensional integrals For fixed values of the noise parameters, however, they can be biased, and a hybrid Bayesian-frequentist technique called noise-marginalization can be used to mitigate these biases The optimal statistic for the GWB was the testbed for developing noise marginalized techniques, as the most detectable GW signal in PTAs In this work, it was shown that noise marginalization of the optimal statistic reduces the signal recovery bias due to covariance between the intrinsic red noise in the individual pulsars and the GWB As a result, noise marginalization has become the standard for GWB frequentist work As PTAs prepare to search for signals beyond the GWB, we are working to bring the frequentist methods used for various deterministic GW signals on par with the GWB optimal statistic Very recently [37] developed a noise-marginalized version of the Fp statistic introduced in [14] for continuous wave (CW) searches Surprisingly, unlike the results for the GWB optimal statistic, the implementation of noise marginalization into the Fp statistic resulted in no tangible difference in signal significance recovery Another potentially detectable signal in PTAs is GW memory, which is a nonlinear effect resulting in a constant shift in the space time metric due to violent GWemitting events, such as mergers of super massive black hole binaries (SMBHBs) [38–41] In PTA we measure the times of arrival (TOAs) of these radio pulses, and compare them to our predicted model for when we expect these pulses to arrive GW memory appears as a sudden change in the rotational frequency of the pulsar that results in a linear drift in our timing residuals which we will refer to as a ramp In this paper, we study the robustness of frequentist techniques to search for GW memory We start from the frequentist statistics developed for GW memory in [33], and build upon this work to include noisemarginalization as was done for the GWB and continuous waves in [35, 37] We apply this method to the incoherent Fi statistic for GW memory, then compare its precision and efficacy with the standard method, which uses arXiv:2610.20772v1 [gr-qc] 14 Oct 2026

--- Page 3 ---

3 II. METHODS We begin by describing the form of a GW memory signal in a PTA, and giving a brief overview of the form of the frequentist Fc and Fi statistics derived in [33] We then detail the noise-marginalization (NM) process, and how we compare it to the maximum-posterior (MP) method A. p. 1

Where the paper says it

METHODS We begin by describing the form of a GW memory signal in a PTA, and giving a brief overview of the form of the frequentist Fc and Fi statistics derived in [33]. We then detail the noise-marginalization (NM) process, and how we compare it to the maximum-posterior (MP) method.

Signal Model We write the time-series for the residuals induced by a GW memory wavefront passing Earth at time t0, from the direction omega, with a polarization ˆ ψ, strain h0, for a pulsar in the direction of the unit vector ˆp as the vector δtmem = a(h0, ψ, p, ˆ omega) A(t;t0) The full effect of GW memory should also include the ”Pulsar term,” which results from the wavefront passing the pulsar However, here we consider only the Earth term because it is difficult to distinguish a single pulsar GW memory event from a transient noise effect The full time-series residuals for a single pulsar in a PTA are δt = δtmem + Mϵ + n + Fa We model each of the red noise processes using a power law PRN (f) = A squared RN f(-γ) with an amplitude ARN at the reference frequency f1yr, and γ is the spectral index B. p. 1
Where the paper says it

We then detail the noise-marginalization (NM) process, and how we compare it to the maximum-posterior (MP) method. Signal Model We write the time-series for the residuals induced by a GW memory wavefront passing Earth at time t0, from the direction Ω, with a polarization ˆ ψ, strain h0, for a pulsar in the direction of the unit vector ˆp as the vector δtmem = a(h0, ψ, p, ˆ Ω) ˆ A(t;t0).

The Fi statistic In this sub-section, we work through the derivation of the GW memory incoherent statistic Fi We follow formalism derived in [33] for the frequentist statistical methods, while updating the likelihood definition to be compatible with the enterprise [36] code base We take the inverse of this matrix using the Woodbury matrix identity to get C−1 = N−1 − N−1T(B−1 + T T N−1T) −1T TN−1 Maximizing this likelihood ratio will give us the highest likelihood GW memory signal given the data We rewrite this by defining the signal as a sum of amplitude and template vectors δtmem = X i a i A i where N is the number of pulsars Then we set the derivative of this equal to zero to maximize the likelihood ratio ∂ ln Λ / ∂a j = 0 = Ni − a i Mij, which gives us both the maximum likelihood estimators for the ”a” parameters ai = MijNj where Mij = (M−1) ij, as well as the corresponding max likelihood ratio ln Λ = 1/2 NiMijNj ≡ F The average over many noise realizations of the 2F statistic follows a χ2 distribution with N degrees of freedom, and non-centrality parameter ρ squared = (˜δtmem˜δtmem)⟨2Fi⟩ = N + ρ squared C. p. 2
Where the paper says it

Then we set the derivative of this equal to zero to maximize the likelihood ratio ∂ ln Λ ∂aj = 0 = Ni − aiMij , (20) which gives us both the maximum likelihood estimators for the ”a” parameters ai = MijNj , (21) where Mij = (M−1 ) ij , as well as the corresponding max likelihood ratio ln Λ = 1 2 NiMijNj ≡ F. (22) The average over many noise realizations of the 2F statistic follows a χ 2 distribution, where the decomposition of the signal as mentioned above can be done coherently or incoherently across pulsars.

False Alarm Probability In order to utilize the Fi statistic as a robust detection statistic, we will calculate false alarm probabilities (FAPs) for all of our simulations As the ⟨2Fi⟩ follows a χ2 distribution with N degrees of freedom, we have an analytic form of the null distribution p0(Fi) = F(N squared - 1)/((N squared - 1)! p. 2
Where the paper says it

If we take twice the Fi statistic, and average over noise realizations, we find the ⟨2Fi⟩ follows a χ 2 distribution with N degrees of freedom, and non-centrality parameter ρ 2 = ( ˜δtmem| ˜δtmem) ⟨2Fi⟩ = N + ρ 2 . False Alarm Probability In order to utilize the Fi statistic as a robust detection statistic, we will calculate false alarm probabilities (FAPs) for all of our simulations.

)exp(Fi) The FAP for a Fi value F0 is found by taking the integral over the null distribution from this value to infinity, which represents the portion of the null distribution that could produce a value of F0 or larger

--- Page 4 ---

4 D. Maximum-Posterior and Noise-Marginalized Methods In order to calculate the Fi statistic for a PTA data set, we need a set of noise model parameters to construct our covariance matrix with All modern PTA data sets contain some amount of red noise and white noise One method is to find the maximum likelihood value obtained in the Bayesian chain, and set the red noise parameters to the corresponding values, which we call this the maximum-posterior (MP) method The second method involves performing many random draws of the parameters from the chain, and calculating the Fi values for each set of red noise parameters to obtain a distribution for the Fi statistic over the red noise posterior values, which we refer to as the noise-marginalized (NM) method It should be noted that the NM method takes more than an order of magnitude longer to run, as all randomly sampled red noise parameters need a new calculation of the statistic using a different covariance matrix in the scalar products In work done with the GWB optimal statistic in [35] showed, when the signal that is being searched for is highly covariant with other red noise processes, this NM method is more robust However, in [37] it was shown that for continuous GWs the MP method is sufficient Therefore, one of the goals of this work is to determine whether the more computationally intensive NM method is needed for GW memory detection statistics

--- Page 5 ---

5 III. PTA STATISTIC ANALYSIS In this section, we describe the simulated and realistic data sets we use to test the Fi statistic All simulated data sets in this section consist of a ”Theorists PTA,” which has 20 pulsars distributed randomly across the sky with 20 years of data We will run the Fi analysis on all three data sets, using both red noise parameter estimation methods The low amplitude simulation results are plotted in Figure 1 In Figure 1, the distribution of Fi and FAP values across the t0 range shows no significant bumps, which is to be expected with the strength of GW memory present In Figure 2 are the results for the mid amplitude signal we see that in that more sensitive central region there is now a peak where there had been a valley around the expected epoch t0 = 58 109 MJD The results of the large GW memory injection in Figure 3 show an unmistakable peak present centered around the correct t0 value, we get FAP value around 10-13 for both detection methods In Figure 4, ordered in descending red noise amplitude order from top to bottom, we can see that a decrease in red noise content can change an low significance GW memory signal into a detection Across all of these simulated results we see a very consistent result when comparing the two methods of estimating the red noise parameters The MP method values almost completely overlap with the mean values of the NM method, and they all fall within the 90 percentile region In each simulated data set, the resulting inference possible from the Fi statistic does not change when using either the MP or NM method

--- Page 6 ---

6 B. Varied Red Noise Content Simulations In previous work, we have seen that the level of common red noise present in the dataset can have a significant impact on the significance of a GW memory detection [21] To test this with the Fi methods, we perform a set of tests on thre… p. 4

Where the paper says it

Varied Red Noise Content Simulations In previous work, we have seen that the level of common red noise present in the dataset can have a significant impact on the significance of a GW memory detection [21]. To test this with the Fi methods, we perform a set of tests on three datasets with a small GW memory signal of log10 h0 = −13.

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

  1. Noise marginalization can be used to mitigate biases in frequentist analyses by mitigating covariance between the intrinsic red noise in the individual pulsars and the GWB. This allows for a more robust assignment of a false alarm probability to Bayesian results.

  2. The improved system can perform signal recovery by testing if noise marginalization of the optimal statistic reduces the signal recovery bias due to covariance between the intrinsic red noise in the individual pulsars and the GWB. This is particularly useful when searching for signals that are highly covariant with other red noise processes.

  3. The system can be configured to use a maximum-posterior (MP) method as a sufficient detection statistic, as the paper concludes that the maximum posterior method is sufficient for identifying potentially detectable signals without the need for noise marginalization in all cases.

  4. The AI system can assign false alarm probabilities by calculating them using the null distribution derived from the statistic's null distribution, which is shown to be related to a chi squared distribution with N degrees of freedom.

  5. The improved system can compare different methods for estimating red noise parameters by checking if they yield equivalent results, as the paper finds that the MP method values almost completely overlap with the mean values of the NM method.

Abstract

Current pulsar timing arrays could detect gravitational wave memory, which would provide a unique test of general relativity. In order to optimize our searches for gravitational wave memory in pulsar timing arrays, we integrate the optimal frequentist statistics for GW memory into the enterprise code suite. We additionally test a noise marginalized version of this statistic, and compare it's effectiveness to the simpler maximum posterior method. We perform these tests on simulated pulsar timing array with red and white noise content, as well as the NANOGrav 15-year data set. We find no significant variations in our detection statistics between the noise-marginalized and the maximum-posterior methods for the simulated datasets, and only small variations in the NANOGrav 15-year comparisons. We conclude that for realistic PTAs the noise-marginalization method could be useful for assigning a false alarm probability to Bayesian results, though the maximum posterior method is sufficient for identifying potentially detectable signals.

Sources

Related papers