Introducing plans for Gravity Spy 2.0!

Hey GravitySpy-ers,

We are excited to announce that the Gravity Spy team was awarded a three-year research grant by the U.S. National Science Foundation (NSF) to build the next volunteer-enabled gravitational-wave search phase. The project will investigate approaches to use auxiliary channel data, the same data used by LIGO scientists when isolating other noise signals from the glitches in the main channel. Existing volunteer efforts engaging in advanced work (e.g., reading and interpreting information in alogs and providing insights into the root causes of glitches) typically performed by scientists at the detector sites inspired the research team to consider additional technical support to assist volunteers. A recent publication by the research team (Soni et al., 2021) demonstrated that volunteers are pretty good at making discoveries in large data sets even without such technical support.

The new project is complex, which is why we have built a team of researchers from across the U.S., including LIGO researchers within CIERA, LIGO researchers at Christopher Newport University, machine learning researchers at Northwestern University, and crowd-sourced science researchers at Syracuse University and the University of Wisconsin-Madison. 

One goal of the project is to create novel interfaces for Gravity Spy that visualize the auxiliary channel data and provide statistical summaries about correlations between glitches, activities performed by LIGO scientists when they investigate the source of glitches. The team plans to create Beginner, Intermediate, and Advanced workflows that support different modes of interaction at each stage. In the Beginner mode (see Figure below for an interface mockup), volunteers will be able to compare a glitch that occurred in the gravitational-wave channel with the spectrogram of correlated auxiliary channels to assess whether the correlation seems natural or coincidental. In Intermediate mode, volunteers will be able to examine and refine collections created by beginners and identify correlations between glitches and the auxiliary channel over some timeframe. Finally, work in the Advanced mode will enable volunteers to discover the root cause of glitches by examining patterns and making inferences among large sets of auxiliary channel data. 

​​A Mockup of the glitch vetting interface with an example auxiliary channel image and glitch image (in this case, a Whistle).

By building tools to support similar activities in Gravity Spy 2.0, volunteers will view new data and learn to make statistical inferences about noise signals among the thousands of auxiliary channels. Over the next year, the research team hopes to understand what background knowledge of the auxiliary data is helpful for volunteers, and what system tools and expertise will help volunteers identify connections among data sources. Through this, the research team hopes volunteers will isolate which relationships are causal vs. spurious. The research team will engage existing Gravity Spy volunteers throughout the project to obtain feedback and insights about the new interfaces (a mockup interface is shown below). We hope everyone is as excited about this project as we are. 

We look forward to continued collaboration with volunteers as Gravity Spy sets the standard for the next generation of people-powered science projects.

Cheers, 

The Gravity Spy Team

New Glitch Category Added to Gravity Spy – Pizzicato!

Hello everyone!

Pizzicato (originally known as High Frequency Burst 500 Hz) has been added as a new glitch category option in Level 5 and Level 7!

Pizzicato was brought to our attention when it was proposed as a new glitch class by @EcceruElme. She, along with the help of other users, is ultimately the reason why Pizzicato has been added as a new category. 

Pizzicato glitches typically resemble a UFO and show up within the frequency ranges of First Order (~500 Hz) and/or Second Order(~1000 Hz) Violin Modes. Pizzicato first showed up during Engineering Run 14 (ER14), and continued to the end of O3. About 80% of these glitches are found at the Livingston detector, with the other 20% occurring at Hanford. They showed up in both detectors within months of each other; however, since we have seen possible previous incarnations of Pizzicato (see Church), we cannot be certain that whatever is causing these glitches is something completely new to the detectors.

This glitch is so interesting to us primarily because it so clearly seems related to the violin modes of the test mass suspensions, but at the same time we haven’t yet been able to pinpoint an exact cause.

Violin modes occur when the glass fibers that hold up the test masses (the mirrors that we bounce lasers between to measure gravitational waves) in the detectors experience excitations from changes in temperature. These excitations manifest in the fibers at around 500 Hz and, depending on how strong these excitations are, can cause vibrations at higher frequencies, similar to how plucking a violin string can cause reverberations at multiple frequencies. Excitations around 500 Hz are First Order Violin Modes, 1000 Hz are Second Order, and 1500 Hz are Third Order Violin Modes.

Comparing Spectra

Since these glitches are present in the same frequency ranges that Violin Mode Harmonics glitches take place in, the first thing we checked was if Pizzicato glitches were directly correlated to the violin modes being excited. We did this by first finding a time where a Pizzicato glitch was seen, and then picking a nearby time where there were no nearby glitches (Fig. 1 below). We then plotted data from these times as a spectrum (Fig. 2 below), which shows us the Amplitude Spectral Density (ASD) versus frequency. The ASD shows us the average amount of noise in the detectors at any given time.

Figure 1: Livingston BNS Range (blue line) placed over the glitchgram (dots) for this day. We tracked down the time for one of the louder Pizzicato glitches from this day, a 1023 Hz glitch from 07:46 UTC, and then found a time within the same lock stretch where there are no glitches near that frequency, 03:00 UTC.
Note: BNS Range y-axis units not shown for viewing clarity.
Figure 2: ASD vs Frequency to compare possible changes in the violin modes for the two times from above.

The blue in Figure 2 above gives us the ASD at the time of the glitch, and orange is a nearby time that didn’t contain any glitches in this frequency range. Both spectra line up perfectly, which tells us that the violin modes were present in this exact capacity during both times, but Pizzicato wasn’t. This tells us that Pizzicato could not have been caused by a change or gain in the amount of violin modes present in the detector. After comparing several Pizzicato glitch times to nearby times, it seemed like the biggest takeaway was that when the Pizzicato glitch was centered around 500 Hz, the spectra during the lock time showed the ASD as being higher around 500 Hz than 1000 Hz, and vice versa for glitches centered at 1000 Hz. 

Comparing to Violin Mode Changes in Amplitude

Below (Fig. 3) is a day where Pizzicato glitches were cropping up in the Second Violin Mode frequency range, between 999 Hz and 1038 Hz. There are multiple lines of glitches (dots) laid over the amplitude of the Second Order Violin Modes relative to the median amplitude. The confirmed Second Order Violin Mode frequencies that Pizzicato glitches occur at are 999 Hz, 1005 Hz, 1018 Hz, 1023 Hz, 1028 Hz, and 1038 Hz. We can see lines of glitches at all these frequencies. This tells us that the glitches in these lines are very likely Pizzicato.

You will also notice in Figure 3 that there seems to be a blue band between 1009 Hz and 1023 Hz and a red band between 1013 Hz to 1018 Hz where the amplitude of the violin modes seems to be heightened. We are not sure if that has anything to do with Pizzicato, but I thought I would note it anyway since we know that 1018 Hz and 1023 Hz are proven to be the peak frequencies of some Pizzicato glitches.

Figure 3: Glitchgram laid over Second Order Violin Mode Amplitudes Relative to Median for 2019/11/13. The red rectangle shows the frequency range for Pizzicato glitches and the blue brackets highlight the biggest amplitude changes.

Looking at the spectrum in Figure 4 below that gives us the ASD of the Second Violin Modes for this date, we see that the range where these higher frequency Pizzicato glitches occur also happens to encompass the majority of the range where Second Order Violin Mode Harmonics are seen ringing up. 

Figure 4: ASD vs Frequency for 2019/11/13 Second Order Violin Modes

Below (Fig. 5) is another day where we were seeing Pizzicato, but this time the glitches are seen between 492 Hz and 521 Hz, which corresponds to the First Order Violin Mode frequency range. We determined that these lower frequency Pizzicato glitches can occur at 492 Hz, 496 Hz, 510 Hz, 514 Hz, and 521 Hz. Again, these frequencies line up with multiple lines of glitches on the glitchgram, although in this example we also see lines of unknown glitches at frequencies Pizzicato is not currently known to occur at.

Like the previous example, we also have a frequency band, between 505 Hz and 515 Hz,  where the First Violin Mode amplitude is changing relative to its median, however, we do not see any similarities between the frequencies for the changes in amplitude here and the frequencies Pizzicato manifests at.

Figure 5: Glitchgram laid over First Order Violin Mode Amplitudes Relative to Median for 2020/01/08. The red rectangle shows the frequency range for Pizzicato glitches and the blue brackets highlight the biggest amplitude changes.

Figure 6 below, similar to Figure 4, shows a spectrum that tells us that the frequencies that these lower frequency Pizzicato glitches occur at spans the entirety of the frequency band for the First Order Violin Modes. Because of this, it’s hard to determine whether Pizzicato is or is not a direct outcome of the violin mode harmonics.

Figure 6: ASD vs Frequency for 2020/01/08 First Order Violin Modes

So then what causes Pizzicato? Why do they occur at these particular frequencies? The evidence doesn’t lead us in a particular direction, so we can only speculate for now. 

A theory for what may be causing Pizzicato are the damping mechanisms for the violin modes. Since violin modes generate a lot of noise in the detector, LIGO has multiple ways to dampen the fibers, and it’s very possible that one of the many systems put in place to dampen them is causing a glitch to appear in the main channel. This theory is supported by the fact that Pizzicato glitches around 500 Hz seem to be correlated with lock stretches of higher First Order Violin Modes, and glitches around 1000 Hz appear when Second Order Violin Modes are particularly high.

On the other hand, it is also possible that Pizzicato glitches are indeed directly caused by violin modes. If they are a direct result, it would be very interesting to compare them to the Violin Mode Harmonics glitches (Fig. 7) that we already keep track of in Gravity Spy, especially since they have similarities in frequency but are otherwise extremely different in terms of morphology, duration, variation over time, and in how they appear in the detectors – Violin Mode Harmonics like to string together whereas Pizzicato glitches don’t.

Figure 7: Violin Mode Harmonics Glitch (top) versus Pizzicato Glitch (bottom)

We will be continuing to explore possibilities and other avenues of looking through the data and hope we can update you all soon!

We want to thank everyone who interacts on Talk, proposes categories, and creates collections. It really helps the gravitational wave community! If you haven’t already, check out our recent announcement of the discovery of two separate Neutron Star – Black Hole Mergers! Discoveries like these would not be happening nearly as often without the dedication of citizen scientists like you!

Oli Patane and the Gravity Spy Team

New O3 Glitch Options

Hi all,

Backed by popular demand, we have added a few new classes of glitches that are prevalent in the O3 data. You will see these options available if you are classifying on the “Binary Black Hole Merger” or “Inflationary Gravitational Waves” workflows.

 

70 Hz Line: 

These are strong, monochromatic lines that started appearing in O3 and sit at around 70 Hz. 

 

Crown: 

These are repeating, arching glitches at frequencies of ~20-50 Hz and appear in groups of ~3-5.

 

High Frequency Burst: 

These are short-duration, blip-like glitches that occur at frequencies greater than ~1000 Hz. 

 

 

Happy classifying!

Mike

Diagnosing the Tomte glitch

Hi everyone!

Recently, some of LIGO’s detchar experts investigated the connection between Tomte glitches from Gravity Spy and the bias voltage on a particular circuit. You can read the full alog of this investigation here. Below is a summary written by TJ Massinger, one of the detchar experts who does a lot of work with Gravity Spy data!


During recent commissioning work at the LIGO Livingston Observatory, it was noticed that glitches were occurring while the bias voltage on a circuit (an electrostatic actuator) was adjusted. Inspection of these glitches using the same time-frequency visualization that Gravity Spy uses showed that they looked qualitatively similar to glitches classified by users as the “tomte” class.

42851_20190118161507_ESD_glitch_scan.png

Spectrogram of an ESD bias ramping glitch. 

Using the GravitySpy classifier, it was found that these recent glitches are also classified as tomte glitches. To quantify the similarity of the bias voltage glitches to the tomte glitches seen in the second observing run (O2), GravitySpy was used to gather a collection of previously identified tomte glitches in O2 with parameters similar to those seen when adjusting the bias voltage. Upon comparing these populations, their time- and frequency-domain morphology was found to be nearly identical, suggesting that the population of tomte glitches in O2 might be understood by continuing to investigate the glitches that occur when adjusting bias voltages.

42851_20190118162015_O2_rate_tomte.png

Tomte rate in O2 based on Gravity Spy classifications with greater than 90% confidence. 

Engineering Run 13

Hey GravitySpy-ers,

We are excited to bring you new data from our Engineering Run 13! Data taken during Engineering Runs are meant to test not only some of our new upgrades to the detectors but some of our software (like Gravity Spy).

When

Dates: 10 am Central Time Dec 14 to 8 am Central Time Dec 18 (N.B. Due to some issues at the sites science ready data was not available until Saturday December 15)

What Detectors Are Running

Originally, we anticipated only having Hanford and Virgo due to critical repairs at Livingston. These repairs, however, completed yesterday and after a short delay Livingston has joined ER13. At first, we will be streaming in the data live for Hanford and Livingston over the weekend, and at a later date will add Virgo ER13 data to the Virgo only workflow.

What’s New

The sensitivity of the LHO detector has increased its range to detect binary neutron stars from 80Mpc to 90Mpc, LLO has increased to 100Mpc and Virgo has nearly doubled its range from 25 to 43Mpc. A number of different glitch classes have arisen and the engineering run is a golden opportunity to identify and eliminate these so we can be rid of them for the year long O3 run which is anticipated to start in March 2019.

Example: https://alog.ligo-la.caltech.edu/aLOG/index.php?callRep=41765L1_mWetkqJlpa_spectrogram_4.0

Some of the changes at both LHO and LLO that have led to this improvement include squeezed light and a new 70W laser amplifier that will improve LIGO’s quantum noise limit. In addition, Acoustic Mode Dampers will damp internal modes of the test masses to reduce parametric instability (light interacting with mirrors as positive feedback). Also, there was a change of several test masses to improve their coatings (especially for green light) and to remove a point absorber at LHO.

We look forward to your collections of interesting new glitches and for determining the cause of the new excess noise sources!

The Gravity Spy Team.

Introducing Virgo and a new workflow structure!

Hey Gravity Spiers,

We are really excited to finally introduce a new detector, workflow structure, and tool this week. First, we present a new workflow containing glitches from the Virgo detector in Pisa, Italy. Second, we are changing the level structure to speed up our user training. Finally, we are bringing you an auxiliary web tool to help the search for unique and novel glitches.

Virgo

46e30172-d5e8-4d83-b338-e428138b7839

Virgo differs from Hanford and Livingston in a few ways including the length of the arms, the apparatus holding up the test masses, and the suspensions. We anticipate there will be a number of interesting new glitches in Virgo. For some glitches, such as Scattered Light, they will appear different but have the same cause. For other glitches, such as the Violin Mode Harmonics, they will be the same source but at different frequencies due to the different suspension system. Below we demonstrate a few novel glitches you may find along the way while classifying Virgo glitches, including a new class we have called Fireball (bottom right).

 

New Workflow Structure

flowchart

For those of you familiar with Gravity Spy’s training method, we intend to utilize pre-labelled images to help train new users in the classification task. In addition, in order to facilitate training, we introduce a different number of new families of glitches in different levels, culminating in Level 4 where all 22 classes are introduced. However, after some feedback, as well as looking at the data, we learned that getting from level 4 to level 5 was taking longer than anticipated. We decided that this was due to too many new classes being introduced between level 3 and 4. Specifically, the amount of pre-labelled images users were seeing was spread out across too many new classes causing the number of classifications a user must complete before seeing pre-labelled data to sky rocket. Therefore, we are adding another intermediate level that has 15 classes. We believe this will cause users to see pre-labelled images of the new classes faster and, in turn, move through the levels faster. In total, with the addition of Virgo, there are now 7 levels in Gravity Spy (see image above).

This restructuring of the workflows may cause some users to start on levels lower then they may expect. This can be due to a number of factors, and we encourage all users to simply charge ahead with classifying on whichever level they find themselves on. You should experience a fairly rapid promotion through the levels.

Thank You!

We want to thank all of our volunteers for their continued efforts on Gravity Spy and we appreciate all the feedback we have received. We look forward to seeing what novel Virgo glitches you are able to find! As always please reach out to me with all leveling issues. We hope this restructuring proves an effective method to boost training.

 

 

Introducing Gravity Spy Tools

Gravity Spy Tools

trainingset_feature_space

With the introduction of the new Virgo workflow, we anticipate there being a number of novel glitches, some that will look like what you may have seen in Hanford and Livingston, and some very different. In an effort to help facilitate the generation of large collections of novel glitches, especially when we are not sure what to expect with Virgo, we are introducing a new supplementary tool for Gravity Spy, gravityspytools. For an idea of how to use this tool please watch the linked video. The goal of this tool is to maximize the impact of a new machine learning algorithm that the Gravity Spy team has developed called DIRECT. This algorithm utilizes transfer learning in order to learn what makes gravity spy images similar and dissimilar from each other. This allows every Gravity Spy image to be abstracted into a feature space containing 200 points. It is in this feature space that we calculate distances from one images to another. An interface to do this is provided on gravityspytools called the “Similarity Search.” It takes as input one sample from Gravity Spy and as output the closest samples in the feature space based on distance. An attempt to visualize in three dimensions what the set of known images (such as blip, whistle, etc) looks like in this 200 dimensional feature space is shown above.

This slideshow requires JavaScript.

Thank You!

We want to thank all of our volunteers for their continued efforts on Gravity Spy and we appreciate all the feedback we have received. Please let us know how you find using the gravityspytools! As always please reach out to me with features you would like to see!

Introducing the Sounds of Glitches

Hello Gravity Spiers!

At long last and after much demand, we have added audio examples of what the Gravity Spy glitches sound like to our field guide! As many of you may know, the frequencies of gravitational waves (and frequencies of glitches) detectable by LIGO are similar to the frequencies of sound (i.e. the pitches) that humans can hear. Therefore, LIGO scientists oftentimes convert our signals to sound!* The MP3s are embedded directly into the field guide, so you should be able to play them straight from there.

Some glitch categories may be hard to distinguish above the background noise, whereas others you should hear quite distinctly. Either way, having a good set of headphones will help hear the subtle features of the glitches better.

Big thanks to our LIGO collaborator Derek Davis for putting together these glitch sounds! Head on over to the Gravity Spy field guide to take a listen to the sounds of glitches!

-Mike / the GSpy team

 

 

*There are a few changes done to the data that make the sounds easier to hear. Other than the standard whitening and band-passing, for the glitches in our field guide we also linearly shift the frequency up by 60Hz to make the sounds (especially the low frequency glitches) more in our audible range. Also, a filter that can be described as an “inverse A-weighting” filter is utilized. The basic idea of this filter is to account for the fact that our ears are less sensitive to particularly low and high frequencies. Since the drop off starts around 200 Hz, this affects a decent number of glitches. By increasing the loudness of these lower frequencies, we make it so that features of similar intensity in an omega scan are ideally heard equally loud, no matter their frequency.

Which glitches are most common at LIGO Hanford and Livingston?

Thanks to the hard work by GravitySpy Citizen Scientists, we now have more than 53,000 retired images, images that have had enough consistent classifications by citizens that we are quite sure they are correctly categorized. New images are retired every day, so this data set is always growing. Using this set of retired images, along with information from machine learning image analysis, has allowed us to get a clearer picture of which glitches appear often and rarely in LIGO Hanford (H1) and LIGO Livingston (L1). 

Since image retirement relies on classifications by citizens, the images that get retired the fastest and most often are those that are the clearest, the ones that contain only one sort of glitch, and look like the example images. Thus using only retired images may not be a good measure of the total number of glitches. Because of this caveat, we also looked at glitches that the machine learning algorithm identified as belonging to a given category with a confidence of 90% or above.

Looking at the summary information for categorization done by these two methods for LIGO’s two Observing Runs, O1 and O2, we found that there are a handful of glitch categories that are never or almost never found in H1 and L1.

Screen Shot 2018-02-20 at 12.05.18 PM

Above is a summary of all the retired glitch categories at Hanford in O1 that had at least one glitch in them. As one can see, this is not every glitch category that we have; 1080 Line, 1400 Ripple, Chirp, Helix, None of the Above, Tomte, Violin Mode Harmonic, and Whistle don’t show up at all! However, None of the Above likely does not show up because it is such a diverse category that it is challenging to have enough consistency to retire such glitches or reach high machine learning confidence. Also, chirps are found infrequently, but when they are, they are very interesting! Paired Doves and Wandering Line show up only twice each out of the 4,448 retired glitches. Here are the pair of Paired Doves:

ZVbX8Asiva

7R07EaOw5Z

For Hanford, we have determined that Paired Doves and Helix happen infrequently. Similarly for Livingston, 1080 Line, 1400 Ripple, Air Compressor, Helix, Light Modulation, Paired Doves, Tomte, and Wandering Line are very rarely used.

For the interested reader, below are links to screenshots of the full summary of retired and machine learning categorizations.

Retired

H1 O2

L1 O1

L1 O2

ML Confidence 0.9

H1 O1

H1 O2

L1 O1

L1 O2

Again, thank you all for your continued hard work in classifying these glitches. Without you all, this wouldn’t have been possible!

The GravitySpy team

Are blips and koi fish related?

There is some debate on whether Koi fish and blip glitches are part of the same morphology distinguished mainly by loudness or amplitude of the signal. Andy Lundgren has shown that removing calibration lines as part of “data cleaning” makes them look much more similar. A possible explanation of this is that for loud enough glitches, the calibration lines may be causing the Q-transform’s whitening filter to ring at those frequencies, creating the fins on the koi fish. Whitening is a process that removes stationary differences in the loudness of Individual frequencies in the signal.

koifish

Fig 1 – Q-transform of typical Koi fish glitch

Blip

Fig 2 – Q-transform plot of the same data as above, but with calibration lines removed prior to calculating the Q-transform.

Beverly Berger has suggested that the amplitude of the glitches may be an effective way of discriminating between Koi fish and blips, with Koi fish being louder. To dig into that a little further we plotted the frequency and amplitude of the two glitch types during the 02 run.

blip-freq

Fig 3 – Frequency distribution of blip glitches during 02

koi-freq

Fig 4 – Frequency distribution of Koi fish glitches during 02

The plots of frequency distributions show significant overlap between the two glitch types. Koi fish have a lower peak frequency due to the shape but not enough of a difference to help in classification.

blip-ampl

Fig 5 – Amplitude distribution of blip glitches during 02

koi-ampl

Fig 6 – Amplitude distribution of Koi fish blitzes during 02

When we look at the amplitude differences between Koi fish and blips we see a pretty sharp dividing line around 10-21, especially during the first 21 weeks. The gap starting at week 23 is a time when there was significant commissioning. There was also a significant increase in range at Livingston around this time.

It is still unclear whether Koi fish are the same as blip glitches, only louder. We also have not been able to identify what exactly causes either glitch. This is just an interesting observation that we thought we would share with our colleagues at Gravity Spy.

–Joe Areeda