ONGOING DATA ANALYSIS PROJECTS

 

This page provides an insight into projects that various EGRG members work on.

 

All projects are carried out under the supervision of one, two, or more working groups of the LIGO Scientific Collaboration (LSC). Projects must be in line with collaborational activities described in LSC white papers. For an example of such an LSC white paper, click HERE.

 

Additionally, results of the projects must be reported annually to the LSC, where they undergo a review process carried out by a dedicated LSC committee. The annual review of group activities must be satisfactory in order to have a member group's memorandum of understanding for the following work year signed by LSC representatives.

 

EGRG has been a member of the LSC since 2007. For an overview of a few selected projects from the past, please click HERE.

 

 

 

 

Responding to gravitational-wave candidate alerts during LIGO observing runs

 

Akos Szolgyen in coordination with the EM Follow-up Working Group

 

The Advanced LIGO Hanford and Livingston detectors had their first observing run (O1) between September 18, 2015, and January 12, 2016, during which they gathered coincident data approximately fifty percent of the time. During such periods, LIGO scientists select potentially interesting events and send them with low latency to electromagnetic (EM) partners to allow them carrying out searches for possible EM counterparts. 

 

The selection process involves qualified volunteers of the LIGO-Virgo Collaboration (LVC), named as ‘follow-up advocates’, who rapidly respond to gravitational-wave (GW) candidate alerts sent out by the detectors, and carry out tests and data quality checks before deciding to send out follow-up requests to EM partners on behalf of the LVC. Follow-up advocates are arranged into groups of 3-4 people, and must be ready 24/7 to respond within minutes to GW candidate alerts. They must also take part in composing a report describing the GW candidate to EM observers, monitor the findings of the site rapid response teams, and follow internal discussions of the LIGO-Virgo analysis of the event as well as follow-up observations by partners.

 

Representing the Eotvos Gravity Research Group in this important project, Akos Szolgyen has undergone training sessions where volunteers were prepared for the participation in O1 data and alert monitoring, and was among the follow-up advocates responding to GW candidate alerts during the O1 run. 

 

As a dedication to the success of the LIGO project, EGRG is committed to continue delegating members to the EM follow-up advocate program in the upcoming observing runs as well.

 

szolgyen_EMadvocate.png 

 

Akos Szolgyen monitoring aLIGO data quality and alerts during the first observing run  

 

 

 

 

GLADE: a value-added full-sky catalog of galaxies for the advanced detector era

 

Dalya, Raffai, and Frei, in coordination with the Burst, EM Follow-up, and GRB Working Groups 

 

EGRG has created a value-added full-sky catalog of galaxies, named as Galaxy List for the Advanced Detector Era, or GLADE. The purpose of this project is (i) to identify host galaxy candidates for gravitational-wave (GW) sources detected and localized by advanced GW detectors, (ii) to support target selections for electromagnetic (EM) follow-up observations of GW candidates, and (iii) to identify host galaxy candidates for poorly localized EM transients, such as gamma-ray bursts observed by the InterPlanetary Network. The catalog is already being used by the LIGO-Virgo Collaboration and by external collaborators in all three areas.

 

GLADE has nearly 2 million galaxies (about 40 times more than the catalog used by the LIGO-Virgo Collaboration in the initial detector era), and a high level of completeness (100% within 73 Mpc and 53% within 300 Mpc) within the Advanced LIGO range for binary neutron stars, which are the main targets for joint GW-EM observations. GLADE was originally constructed by combining and matching sources from four previously existing galaxy catalogs: GWGC, 2MPZ, 2MASS XSC, and HyperLEDA. As according to theoretical models, B-band magnitudes of galaxies can be used as a proxy for binary neutron star formation, EGRG considered it as a crucial requirement to have B-band magnitude data for all galaxies listed in GLADE. Providing distance estimates for all galaxies is also necessary in determining the likelihood of a galaxy being the host of an event, as well as in calculating the completeness of GLADE as a function of distance. Therefore we used a quantile regression forest machine learning technique to calculate B-band magnitudes and distances for a total of ~550,000 galaxies where these parameters were previously unavailable.

 

EGRG is still working on extending GLADE by matching it with additional galaxy catalogs as they become available. There are also ongoing attempts to improve the precision of the machine learning technique used in estimating missing galaxy parameters. Finally, we are working on developing a technique for estimating stellar masses of GLADE galaxies, since theoretical models suggest that they can be used as an alternative proxy for binary neutron star formation and for GW events in general. These upgrades will make GLADE an even more powerful tool for the LIGO-Virgo Collaboration and for the broader astrophysical community as well.

 

This project is carried out with the participation of Gabor Galgoczi, Laszlo Dobos, and Rafael de Souza, from Eotvos University. For more information on the GLADE project, please visit our project website

 

GLADE.png

 

Galaxies per square degrees density of the GLADE catalog in azimuthal projection

 

GRB150906B_2.png 

 

GLADE galaxies around the error box of the 150906B short gamma-ray burst event

 

 

 

 

Targeted search for long-duration gravitational-wave transients

 

Raffai and Frei, in coordination with the Burst and STAMP Working Groups

 

We have participated in the development of a search pipeline (named as "Stochastic Transient Analysis Multi-detector Pipeline" or STAMP) dedicated to focus on gravitational-wave transients in the 1-1000 second time scale. STAMP was the first data processing pipeline developed by LIGO-Virgo Collaboration (LVC) members that covered this duration range of signals. STAMP uses cross-correlated spectra of multiple detectors to construct frequency-time (ft-)maps of cross-power.

 

STAMP applies a pattern-recognition algorithm on the ft-maps that incorporates an image processing method developed by members of the EGRG group. The first test of STAMP was carried out by searching for long-duration gravitational-wave transients in coincidence with long gamma-ray bursts (GRB) detected by the Swift telescope during the S5 LIGO run. In order to determine the time window in which STAMP searches are to be carried out around a GRB trigger, Raffai participated in a study aiming to explore the temporal characteristics of emission processes in GRB engines.

 

STAMP is still the long-duration search pipeline applied by the LVC in processing data of second-generation gravitational-wave detectors.

 

summary.png

Summary of GRB emission process durations around the targeted trigger times 

 

 

 

  

Parameter estimation of gravitational-wave signals

 

Becsy and Raffai, in coordination with the Burst and Parameter Estimation Working Groups

 

Now as LIGO has made the first few detections of gravitational waves (GW), we expect that it will regularly detect GW signals in the near future. GWs are not part of the electromagnetic spectrum, thus these observations will provide a novel tool for astrophysicists to study the physics and properties of GW sources. The only way to obtain information on a GW source that we can utilize in astrophysical research is giving estimations on values of its astrophysical parameters, so it is crucial to develop precise and efficient methods that can extract the values of source parameters from the detected waveforms.

 

In this project we focus on the model independent parameter estimation of GWs, which means that we do not assume any specific astrophysical source model. This approach enables us to estimate parameters of signals for which the source is either unknown or hard to model accurately. It is also reasonable to apply these methods to signals with well-known source types (e.g. binary black holes), because it gives an estimation of parameters which can be crosschecked with the estimation from model dependent parameter estimations.

 

In the first part of this project we focused on characterizing the performance of one of the most advanced parameter estimation algorithms, named as BayesWave (BW). BW uses Bayesian statistical methods and Gaussian-modulated sinusoids to reconstruct the complete waveform and to estimate model-independent parameters of the signal (e.g. central frequency). In case of coincident detection of a GW signal with multiple detectors, sky coordinates of the GW source are also estimated by BW, which can help electromagnetic follow-up observations. This project has led to publishing the paper Becsy+ 2017. In this paper we examined the performance of BW in the aforementioned aspects of parameter estimation by injecting a large number of simulated GW signals into mock aLIGO noise samples, and carrying out computationally intensive data processing with aLIGO computer clusters. By knowing the parameter estimation accuracy, future studies can identify the broadest range of astrophysical models that can be tested with BW.

 

We are continuing this projects by examining new ways of parameter estimation which can be used to estimate astrophysical parameters of the source using the previously estimated signal parameters.

 

skymap_becsy_H1L1.png

 

Position reconstruction for an artificially simulated binary black hole signal

waveform_becsy.png 

 

Injected and reconstructed waveform of a simulated binary black hole signal  

 

 

 

 

Parameter distributions and evolutions of eccentric binary black holes in galactic nuclei

 

Gondan, Raffai, and Frei, in coordination with the Burst and EBBH Working Groups

 

Theoretical models suggest that eccentric binary black holes (EBBHs) forming in galactic nuclei through gravitational capture are among the most common sources of gravitational-waves (GWs) that optimal algorithms could find in data streams of second-generation GW detectors. Finding the GW signals of EBBHs would not only confirm these predictions, but it would also allow studying the formation rate and mechanism of these systems, as well as properties of the dense stellar environments where they form.

 

In this study, we use analytical calculations and numerical simulations to study the parameter distributions of such EBBHs at various stages of their evolution. The aim of this project is to identify the maximum volume and parameter distributions in parameter space that need to be covered with a search for EBBH signals, and to characterize the detection efficiency of currently available search algorithms applying circular waveform templates.

 

In our study, we take into account how much the ranges of parameters are bound by the fact that stable binaries must form in the presence of many other objects that can disrupt the formation process. This effect has so far remained uninvestigated in previous studies focusing on EBBHs. Recent mass segregation studies show that properties of the black hole population in galactic nuclei (e.g. their mass function or radial density distribution), are still highly uncertain. Thus, we also investigate how these properties affect the distributions of EBBHs in order to see how much we can constrain various models of galactic nuclei with future GW detections of EBBHs.

 

Our first results have shown that EBBHs form with highly eccentric orbits (e>=0.95) in galactic nuclei hosts, and that interactions between the inspiraling EBBH and other objects in the environment are negligible. By estimating the number of EBBHs that reach the Advanced LIGO band with e>=0.1 eccentricity, we have found that there should be a significant number of such EBBHs in both single-mass and multi-mass BH populations, with the exact number highly depending on parameters of the multi-mass black hole populations. This means that search algorithms using circular templates will be sub-optimal in finding GW signals of EBBHs, and thus, dedicated algorithms should be developed that can effectively cover the parameter ranges and distributions we predict.

 

This project is carried out with the involvement of Laszlo Gondan, Peter Raffai, Zsolt Frei, and Bence Kocsis, an assistant professor at Eötvös University.  

 

eccbin_formation.png 

 

 Cartoon of the formation process of an eccentric binary black hole

 

 

 

 

Estimating parameter reconstruction accuracies for gravitational-wave signals of eccentric binary black holes

 

Gondan, Raffai, and Frei in coordination with the Burst and EBBH Working Groups

 

Eccentric binary black holes (EBBHs) are among the most promising targets for ground-based gravitational-wave (GW) detectors. These sources may commonly form through triple interactions, or through GW captures in dense star clusters such as galactic nuclei and globular clusters. GWs of EBBHs carry information about the location and distance of the source as well as about its physical parameters, including eccentricity, binary orientation, masses, and spins.

 

In this project we determine the accuracy with which a GW detector network will measure the physical parameters of the source. This accuracy depends on the sky location of the source with respect to the detectors, and on the relative orientation of the binary in the sky. We therefore generate a random Monte Carlo sample of binaries, and determine the distribution functions of measurement accuracies for various EBBH parameters, assuming detections with an advanced detector network including the two LIGO detectors, VIRGO, and KAGRA. This calculation is carried out without making any assumptions on the parameter reconstruction algorithms applied for this task, and thus we estimate measurement accuracies based purely on the planned spectral sensitivity of the detectors, and assume detections with optimal matched filtering methods.

 

An important aim of this project is to highlight the future potential in EBBH detections in constraining EBBH source and host models. This can motivate efforts to develop data analysis methods that efficiently find these types of signals in the noise present in real interferometer data.

 

We have developed the necessary numerical tools to calculate the expected parameter errors for EBBHs using a Fisher matrix approach. We are currently running our calculations on the NIIF HPC cluster in Debrecen. The calculation follows the evolution of a binary from an initial state of arbitrarily eccentric orbit up to the merger phase using leading order evolution equations of spinless binaries.

 

This project is carried out with the involvement of Laszlo Gondan, Peter Raffai, and Zsolt Frei, and is being supervised by Bence Kocsis at Eötvös University. 

 

Fisher_Gondan.png

Chirp mass reconstruction accuracy with an advanced detector network

assuming 30-30 solar mass binary black holes at 100 Mpc distance 

 

 

 

 

Testing globular cluster models with gravitational-wave detections of eccentric binary black holes

 

Becsy, Raffai, and Gondan, in coordination with the Burst and EBBH Working Groups

 

A globular cluster (GC) is a tightly bound spherical collection of hundreds of thousands of old  stars. The proper modelling of GCs can be difficult due to the high number of constituents, each interacting with one another. Accordingly, there are many different analytic and numeric GC models competing against each other. Thus, an efficient observational method is required to test these models, and to find the most realistic one. Electromagnetic observations are limited in this regard, because they cannot provide us information about the deeper structure of GCs. However, as we point out in this new study, detections of gravitational-waves (GWs) from eccentric binary black holes (EBBHs) could serve as a tool for testing and constraining these GC models.

 

EBBHs are expected to form in dense stellar systems, such as GCs. Properties of GCs affect the formation of EBBHs within them, and consequently, we may gain information about properties of GCs by detecting EBBHs and reconstructing their parameters. Our goal is to determine the minimal number of EBBH detections with GW detectors that allows testing implications of different GC models on the observable distribution of EBBH parameters, such as orbital eccentricity and pericenter distance at the time EBBH signals enter the sensitive band of Advanced LIGO (aLIGO). Our study is the first one that aims testing possible models of dense stellar environments using GW signals of EBBHs.

 

EBBHs are proposed to be sources of GWs detectable by aLIGO. These binaries are expected to be detectable from great distances due to the strength and spectral richness of their GW signal. When aLIGO reaches its full sensitivity, which is proposed to happen in 2019, the expected rate of EBBH observations will be 5-20/year. This detection rate can already provide enough data to carry out an actual test of GC models on a reasonable timescale.

 

In 2015, an undergraduate student at Eotvos University and external contributor to this project, Janos Takatsy, was awarded with the 3rd prize of the Conference of Scientific Students' Associations (the most famous and recognized competition in Hungary for undergraduate researchers) for his leading contributions to this project.

 

EBBH-k gömbhalmazban 

 

Eccentricity distribution of EBBHs in a modeled globular cluster 

 

 

 

 

Optimal networks of future gravitational-wave detectors

 

Raffai, Szolgyen, Dalya, Gondan, in coordination with the Burst Working Group

 

We study ways of characterizing and optimizing the efficiency of detection and signal reconstruction of networks of multiple gravitational-wave (GW) detectors. We do this work in strong collaboration with the LIGO Columbia and Glasgow groups.

 

As part of this project, we have applied different N-detector figures-of-merit (FoMs) suggested by several authors, including FoMs characterizing the transient source localization accuracy of networks of detectors, in a complex optimization process aiming to maximize the scientific output of GW detector networks operating in the near or far future. The project has lead to publishing the paper Raffai+ 2013. In this paper (among other published results) we have suggested an optimal location and orientation for the proposed LIGO-India detector in terms of N-detector FoMs, considering the future five-detector network of LIGO-India, aLIGO Hanford, aLIGO Livingston, AdvVirgo, and KAGRA.

 

Further studies in this project resulted with another paper published in CQG in 2015, with a lead author from the University of Glasgow group (Hu, Y.). This paper was among the few ones selected for being highlighted in the CQG+ website. In this work, we used a Markov Chain Monte Carlo method to optimize the locations of future generations of GW detectors in order to maximize their scientific output. By using public databases, we have developed a new perspective on the world map, presenting the allowable regions for building new GW detectors based on expected construction limitations and local noise levels.

 

Within the framework of this project, we have developed a time-dependent optimization tool for target-based gravitational wave searches of advanced detector networks. If we assume that we know the location of a source or host in the sky (e.g. known pulsars, or the Galactic Center region), we can quantify how sensitive a detector network is towards that direction at different times of a day. Comparing these source-specific FoMs for different networks of detectors, we can calculate which sub-network of detectors is the most sensitive among all towards a specific source or host both in general, and at different times of a day.

 

Optimal_Networks.png

 

A suggested network of advanced detectors including LIGO-India [R40]

 

site_map.png 

 

 Our map of allowable sites for building future generation GW detectors (black),

and regions excluded for various reasons (colored)

 

 

   
  (c) Eötvös Gravity Research Group 2007