Wednesday, June 15, 2016

Merry Christmas, LIGO: Another Gravitational Wave!


On the evening of Christmas day 2015, at 9:38 pm CST (3:38 am UTC) at the LIGO Livingston Observatory in Louisiana, another gravitational wave signal was recorded.  1.1 ms later, the LIGO Hanford Observatory in Washington state also picked up the same signal.  70 seconds later, the supercomputer that runs analyses on the near real-time data noticed that there was something special in the data and sent out emails and text messages that some of us affectionately call the "Bat Signal".  This goes out to scientists primarily to summon those who evaluate candidate gravitational wave events to determine if this event should be shared with traditional astronomers (i.e. ones with telescopes).  I am on the list because I am interested in keeping up on the latest results.  I remember exactly where I was: I was in my room at my mother's house outside of Pittsburgh changing clothes after getting back from visiting the in-laws (who live within a few miles of my family's home) for Christmas.  I looked at the event record and saw that this was an extraordinary candidate gravitational wave in that its statistical significance was high but the signal wasn't as obvious in graphs as the first detection in September was.

It was decided to send out the location of the possible detection to traditional astronomers and the emails started flying discussing the evidence that this was a true detection.  It was determined that the preliminary information on the signal warranted starting the detection checklist - the large-scale investigations that try to disprove that the signal is real.  Only after a candidate passes every test and has a high statistical significance is it accepted as a detection.  The same checklist that was applied to the first detection, labeled GW150914, was applied to this candidate as well.  Once this Christmas detection was verified, it was labeled GW151226 (the number reflects the UTC date that the gravitational wave was discovered) although we had nicknamed it the "Boxing Day Event" before the verification.

(Below I will often refer to GW150914 as the "first detection" and GW151226 as the "new detection".)

Read the paper on the detection here.


The signal is similar to the first detected gravitational wave (GW150914).  We call this kind of signal a "chirp" because initially it has a low frequency which increases over time as does its amplitude.  You've heard signals like this before if you've ever hear a slide whistle increasing in tone.  The increase in tone reflects the increase in frequency and the loudness of the whistle represents the amplitude.  The signal we detected starts at about 35 Hz (close to the frequency of the sound made by the second black key from the left on the piano) and reaches its highest frequency at about 450 Hz (very close to the A above middle C if you convert this signal into sound).

Graph of the 1-second signal of GW151226.  The red line is the prediction of what a gravitational wave from a 14.2 and 7.5 solar mass black hole merger would look like and the grey area around it is the signal that LIGO recovered from its data.  The zoomed in portions allow you to get a better look at hour the prediction (in red) and the actual signal (in grey) compare.  At the end of this signal, the frequency and amplitude both go up.  The two black holes merge at the point where the amplitude of the signal is the highest (seen in the zoomed data to the far right).

The plot above shows what we detected in our data compared to the predictions of a pair of black holes orbiting each other and merging into one.  So this is similar to the last detection in that this is also a pair of stellar-mass black holes (formed from the death of extremely massive stars) but different because the masses of these new black holes are less than the first detection.  Here, our newly detected black holes are 14.2 and 7.5 solar masses where our last detection was 36.2 and 29.1 solar masses.  That makes this signal weaker than the last (the peak amplitude of this new signal is about 1/3 that of the first detection) but we are able to observe more orbits of the system here.  We see about 27 orbits of these new black holes (corresponding to the 55 cycles of the gravitational wave we see in the figure) where we only saw about 5 orbits (or 10 cycles) in the first detection.  It is interesting to note that lower mass black hole pairs will merge at higher frequencies than higher mass black holes.  This means that the signal will stay in LIGO's most sensitive frequencies longer and that is reflected in what we see here.  This new detection's signal is about 1 second long while the first detection is less than a half second long.

So, what did this new detection look and sound like?  As far as what it looked like, there was no light that we are aware of that was produced from this system.  But we can visualize the black holes as they orbit around each other and track the corresponding progression through the signal to the merger.  [Credit: SXS Collaboration/]:


We can also "listen" to gravitational waves by taking the signal, and converting it into sound through your speakers.  Below is a comparison of what the new detection "sounded" like compared to the first detection.  The actual "sounds" are quite low in tone so that they sound more like thumps.  We also have shifted the sounds up to a higher tone so that you can hear more of the detail in the signals.  That will play after the original lower tones.  The background graph shows how the frequency changes on the vertical axis (you will see that it increases for both signals) as time progresses on the horizontal axis:


The next question is where in the sky are these black holes?  We primarily determine this using the delay in detection time between the two detectors.  When the delay is large, there is a smaller area in the shape of a ring on the sky where the gravitational wave could have come from.  The detection delay for the new detection is much shorter than the first detection, so our uncertainty is going to be larger.  Below is an illustration of the areas on the sky where the new detection (the area to the left) and the first detection (the area to the right) are likely to have come from.  Note that for the new detection on the left, there is another similar area on the opposite side of the sky that cannot be seen in this image.

The location of both the new GW151226 detection (on the left) and the first detection, GW150914 (on the right).  These are pictured on a star map (you can see the center part of the Milky Way galaxy on the left and extending right).  There is another similar area for the new detection on the opposite side of the sky (not pictured here).  The outer purple area is where we are 90% confident where the sources are located.  The inner circles each have decreasing certainty.

We will be better able to determine the location of a gravitational wave source on the sky when we have more than two detectors in operation.  Fortunately, Advanced Virgo has completed their upgrades and is currently testing their new detector.  LIGO's next observing run is expected in the 4th quarter of this year and Advanced Virgo will likely join the search before the completion of that run.  When we detect more gravitational waves (which we expect since we will be even more sensitive than we were for the two detections we have already made and the run will be longer in duration) together with Virgo, we will know even more about what it is that we are seeing.

This is an exciting time to be a scientist!

Read the official LIGO "Science Summary" on this new detection, GW151226.

Thursday, April 7, 2016

The Source of GW150914: Stellar Mass Black Holes

On September 14th, 2015, LIGO made the first direct detection of gravitational waves.  This event is labeled GW150914 (referring to the year, month, and day of the detection).  The objects that produced the GW150914 were a pair of stellar mass black holes that orbited each other and gradually moved closer and closer together over the course of eons.  The closer together they became, the faster they orbited around each other and the stronger the gravitational waves produced.  LIGO detected the last 0.4 seconds of these stars orbiting until they became so close they merged into a single black hole.

While we saw the death of this paired (binary) system, we didn't get to observe other parts of its life.  Where did these black holes come from?  To answer this question, we need to apply what we know about stellar evolution.


There are several classes of black holes, determined by their mass and how they were formed: stellar mass black holes, intermediate mass black holes, and supermassive black holes.  For stellar mass black holes, they formed when the most massive of stars (more than 15-20 times the mass of our Sun) run out of nuclear fuel and gravity takes over and collapses the star.  For smaller stars, this collapse stops when the pressure from inside the atom (neutron pressure) equals the pressure from the gravitational collapse.  But for these more massive stars, there is no pressure that can stop the collapse and a black hole is formed.  It is in this way stellar mass black holes are the corpses of the most massive stars (but these kinds of black holes are among the least massive).  The newly merged GW150914 black hole now holds the record for the largest stellar mass black hole known.

There are several theories about how this happens... Sometimes this collapse is accompanied by an explosion called a hypernova and is believed to be the source for a kind of gamma-ray burst.  Sometimes the gravity of the collapsing star is so great that all of the matter and light gets sucked into it even if there was a hypernova-like explosion.   


But how did two stellar mass black holes come to be paired together?  The most likely explanation is that they also lived their lives together as a binary star system.  This is very common as it is estimated that about 1 out of 3 stars are in systems of 2 or more stars.  This binary system would likely have formed together and lived their entire lives paired.  The more massive of the 2 stars would have died first since the more massive the star, the faster it burns through its fuel.  Once the nuclear fuel ran out, the more massive star collapsed into a black hole making the system a star/black hole system.  Eventually, the second star would run out of fuel and collapse into a black hole as well making our stellar black hole binary system.  These black holes would orbit for eons before they were close enough to merge and produce the gravitational waves LIGO detected.

In a recent paper (see reference below or read it here), simulations of millions of stars with different material compositions (specifically metalicity which, to an astronomer, is anything that isn't hydrogen or helium; the Sun is 2% 'metal') were simulated and some produced similar outcomes to what we observed.  What was found was that there were similar characteristics for the stars the went on to resemble the GW150914 binary system and this gives us estimates on the time needed for each stage in the system's evolution from birth to the gravitational-wave-generating merger.

The two stars were born about 2 billion years after the Big Bang and were each somewhere between 40 to 100 times the mass of our Sun.  These low metalicity stars (only about 0.06% 'metal') orbit each other as stars for about 4 million years until the more massive one collapses into a black hole.  The now star-black hole system orbit each other for another 1.5 million years until the other star collapses into a black hole.  Both of these stars were massive enough that there wouldn't have been a hypernova-like explosion for either of them; any material ejected would have fallen back into the black hole.  Our new black hole binary system, which is just the corpses of once very massive stars, now go on to orbit each other for over 10 billion years - that is 1000 times longer than the either star was a alive.  At the end of that time, they merge and produce the gravitational waves that LIGO detected 1.3 billion years later when they arrived at Earth.


The short answer: nothing.  This new single black hole is spinning (it is the first detection of a Kerr rotating black hole) but its shape and center of mass are not moving in a way that will ever produce gravitational waves again.   Gravitational waves are also the only way this system would ever have been detected since there wasn't any matter (like dust or gas) to fall into the black holes and generate X-rays.  We will never be able to observe this black hole again.

Of course, there are extremely unlikely events like another black hole flying by and crashing into it...  That may make new gravitational waves for us to see (but I wouldn't hold my breath).


K. Belczynski, D. Holz, T. Bulik, R. O'Shaughnessy, "The origin and evolution of LIGO's first gravitational-wave source" arXive e-Print: 1602.04531 (2016).

Thursday, February 11, 2016

LIGO Makes the First Direct Detection of Gravitational Waves

On morning of 14 September 2015 at almost 4:51 am in Louisiana (09:50:45 UTC) the LIGO detectors in Livingston, LA and Hanford, WA detected a gravitational-wave signal we've labeled GW150914 (based on the date).  The online (near real-time) data analyses alerted scientists about 3 minutes later that there was something of substantial interest in the data.  While vetting this signal (that only lasted about a half of a second) took a substantial amount of time, it opened the new field of gravitational-wave astronomy.  We had not only made the first direct detection of gravitational waves but we also made the first direct detection of a black hole binary (pair) system and proved that these kinds of systems really do exist (it was contentious because the formation of one of the black holes was expected to have destroyed the star that would have made its partner).

At the time of the posting of this blog, the press conference making the announcement is going on and I am working the satellite event being held at the Livingston Observatory.  I will be sure to update this post with the link to the recording or the announcement later (update: see the bottom of this post).  There is too much to talk about in just this post, so I am going to keep this to the basics: what did we see and what does it mean?  I will be doing a series of posts about what we did to make sure that this is a real gravitational wave, the astrophysics of the source, how we detected it, the creation of black holes and why finding a pair like we did is important to astronomy.

Update: Read the Physical Review Letters journal article here.


This gravitational-wave detection was seen as a common signal between the two LIGO sites:

This image shows the data (top row), signal (middle row), and what's left over after the signal is subtracted from the data (bottom row).  Detailed discussion on each image is provided below.

What you see here is a series of images (above and in detail below) that picks apart the signal that was detected.  In the left column is information focusing on the Hanford Observatory and on the right the Livingston Observatory.


The vertical (Y-axis) units are strain with a scale of 10-21.

In the top row is the signal that was seen.  However, this is not the raw data as it was collected.  What you see here is data that has been filtered to 1) reduce noise and 2) to include only frequency components that are around the frequency range of the signal itself.  The red graph on the left is the signal as seen at Hanford and on the left the blue trace is as seen at Livingston.  For comparison, the light red line under the blue Livingston line is the Hanford signal that has been shifted in time to account for the travel time between detectors and flipped (multiplied by -1) to match the orientation of the arms (the arms of each site have a opposite orientation compared to each other so the positive signal in one detector will be negative in the other).  The common signal can be seen with the noise in this comparison.


The vertical (Y-axis) units are strain with a scale of 10-21.

These plots compare the signal predicted by numerical relativity (which are results of computer simulations where the predictions of general relativity cannot be solved by in explicit mathematical expressions) for a pair of black holes with one mass 36 times the mass of our Sun and the other 29 times.  (The red line in the left plot for Hanford and the blue line on the right for Livingston.)  Beneath each of these lines are grey shadowed areas that show the signal as detected from actual LIGO data with two different independent data analysis methods (wavelet and template).  Here again, we can see that the predictions and observations match well.


The vertical (Y-axis) units are strain with a scale of 10-21.

These are plots of residual signals which are the noise that this left behind when the gravitational-wave signal is removed.  Seeing that there is no pattern left in these plots supports that what was seen was a real common signal - a real gravitational wave (this is necessary for a gravitational wave detection but not sufficient - the extra investigations performed will be the subject of a future post).


A powerful tool in signal analysis is breaking up a signal into its frequency components in a graph called a spectrogram.  It allows us to see how much of a signal is made up different frequencies at different times.  If you can hear, then you do this everyday.  It is how you are able to pick apart the sound of a tuba from the sound of a flute when you listen to a symphony.  Both are playing at the same time, but you don't confuse their sounds as coming from anything else.

Below is the spectrogram of this gravitational wave detection:

The horizontal (X-axis) is the progression of time (like above) and the vertical (Y-axis) is showing the contribution of each possible frequency.  The more yellow at a frequency, the stronger that frequency's contribution to the signal at that time.  Our gravitational wave starts at a low frequency (about 35 Hz) and increases to higher frequency (about 250 Hz) near the end of the signal.  This is similar to a signal a slide whistle increasing tone would produce.


As I've mentioned in a previous post, the frequencies of gravitational waves that LIGO is sensitive to would be audible if they were sound waves (which they aren't).  Because of this, we can make them into sound waves by putting the signal through a speaker.  So we did!


Because the starting frequency of the gravitational wave is very low, it is difficult to hear.  The frequency is audible, but at that low of a frequency we tend to feel the sound vibration more than we hear it.  So unless you have a truly great subwoofer, you will probably only hear the end "whoop" of the signal.  In order to make the entire signal more audible, we shifted all of the frequencies up in the above sound up so you can hear the whole thing.  This is not unlike the false-color images made in astronomy for light that our eyes cannot see.


Now that you've heard the detected gravitational wave, you can see that when the tone of it becomes higher toward the end of the signal, the frequency in the spectrogram also goes up.


Because the two LIGO detectors were the only detectors operating at the time of the event (Virgo in Italy is finishing their advanced detector upgrades and KAGRA in Japan is under construction with similar advanced instrumentation) it isn't easy to state precisely where the signal came from.  We can narrow it down to an area on the sky based on how long it took the gravitational wave to travel between the two LIGO detectors, and other factors like the strength of the signal in each detector (there is a slightly different response for each detector for different sky locations).  The most probable location is in the southern hemisphere around the constellations Volans and Carina:

The colored area on this map shows the most probable source of the detected gravitational wave where red is more likely than purple.  The location is shown against a map of the night sky centered on the Milky Way galaxy with constellations outlined.
[Credits: NASA Deep Star Maps (Visualization Credits, Ernie Wright (USRA): Lead Animator, Tom Bridgman (GST): Animator) by NASA/Goddard Space Flight Center Scientific Visualization Studio with constellation figures based on those developed for the IAU by Alan MacRobert of Sky and Telescope magazine (Roger Sinnott and Rick Fienberg), and the source location based on Gravoscope screen grabs (LIGO & Nick Risinger,, all in galactic coordinates. Composition by University of Florida / S. Barke.]


Two different data analysis methods that look at the data in fundamentally different ways not only detected this event, but provided the same results for what the source of it was.  This gravitational wave was made by two stellar mass black holes (these are the remnants of extremely massive stars that have expended their fuel and collapsed under their own gravity).  As quoted above, their masses were about 29 and 36 times the mass of our Sun.  They orbited around each other for hundreds of thousands to millions of years before they come close enough together to start orbiting very quickly (much like an ice skater spins faster as they draw their arms into themselves).  LIGO was only sensitive to the very end of this process right before the two black holes merged into one black hole.  At the end, the stars had a relative velocity of about 1.8x108 m/s, or 60% the speed of light (the universe's "speed limit").  Imagine that...  Two black holes that were each the size of cities but each about 30 times as massive as our Sun whirling around each other at more than half the speed of light!  The animation below shows what it may have looked like to see these black holes merge together.  Note that since they are black holes, no light come from them directly but they do bend the light that is coming from behind them in a process called gravitational lensing:


Based on how strong we know these gravitational waves were at their source as predicted by general relativity and how strong they were once they reached Earth, we estimate that this system is located about 1.3 billion light years (~410 Mpc) away.  That distance is about 10% of the way to the edge of the observable universe!  It also means that the gravitational waves we just detected have been traveling into the universe and toward us for 1.3 billion years.  When these gravitational wave were created the Earth was in the Proterozoic eon of Precambrian time, after when multicellular life developed but before animal life.


Note:  Fast forward to 26:30.  It's just waiting before that. 

Next post: On the formation of stellar mass black hole and why this pair of them are interesting to astronomy...