Catching the waves

By Daniel Holz | July 5, 2009 3:44 pm

I spent last week at Columbia University attending the Edoardo Amaldi Conference, the largest annual international meeting on gravitational waves. Short synopsis: GWs have not been found.

I was reminded that I still owe CV readers a discussion of gravitational waves, following-up from an ancient post on their convoluted theoretical history. While the theoretical community was arguing about the existence of gravitational waves, the observational community was essentially non-existent. The wave strengths expected at Earth are extraordinarily weak, with the most promising sources being the inspiral and merger of stellar-mass compact binary systems (e.g., neutron stars and/or black holes). It took great courage and vision to propose instruments to detect the waves at all. One of the pioneers of gravitational wave detection was Joe Weber, who first invented and built so-called Weber bar detectors. As the gravitational waves pass through a large cylindrical bar of material, they excite resonant modes of the bar, which might then be detectable. In 1969 Weber published an article in Physical Review Letters announcing the detection of gravitational waves by noting coincidences between two bar detectors separated by a thousand kilometers. These observations generated tremendous excitement, especially given that they suggested wave strengths greatly in excess of what was expected. Unfortunately, as other experimenters built ever more sophisticated follow-up detectors, they were unable to reproduce his findings. Over three decades later, and with many orders-of-magnitude improvement in detector sensitivity, gravitational waves have yet to be directly detected. An unfortunate side-note is that Weber continued to claim that he was seeing gravitational waves, even in the face of compelling counter evidence, right up until his death in 2000. Weber was an experimental master (having independently invented the maser and the laser), and is widely credited as the father of gravitational wave astronomy, but in his last decades he was an outcast of the very community he helped found.

A profound development in gravitational waves detection came in 1974 with the discovery by Hulse and Taylor of a binary pulsar. This system consists of a 59 millisecond pulsar in orbit with another star, with a period of 7.75 hours. The pulsar provides an exceedingly precise clock, allowing us to measure the spin-down of the binary system due to the emission of gravitational waves (using the same quadrupole formula mentioned in the previous post). Theory and observation agree spectacularly well, and Hulse and Taylor were awarded the Nobel prize for this indirect detection of gravitational waves. The community is now breathlessly awaiting the first direct detection of gravitational waves. So why am I nattering on about all this? Because of this:
LIGO noise curve

This is a sensitivity plot of LIGO, the Laser Interferometer Gravitational Wave Observatory. LIGO is composed of two power-recycled Michelson interferometers with 4-km long Fabry-Perot arms, located in Hanford, Washington and Livingston, Louisiana. Each curve shows the noise floor, with LIGO sensitive to sources falling above the curve. The curves represent official science runs, and the steady progression shows the improvement since 2002 (1st science run). LIGO has been in development for decades, and has in the last few years (the red curve) reached its original design sensitivity (the solid black curve). LIGO has finished a year-long science run (S5), which constitutes by far the deepest look we’ve ever had at the gravitational wave Universe. It hasn’t seen anything yet; only upper limits. (At least, that is the public stance. Given the convoluted history detailed above, there will be a long period of double and triple checking before any sort of public announcement is made. Although it’s hard to imagine rumors of a first detection won’t leak out, and my rumor well is dry.) At the moment LIGO is being “Enhanced” (a factor of 2 improvement), with the installation of more powerful lasers. It should be back up and running within the year, will run for a year, and then will undergo a major upgrade. By 2014 LIGO will come back online at “Advanced” sensitivity (another factor of 5, which translates into a factor of 1000 in volume compared to today), at which point the first direct detection of gravitational waves is widely anticipated.

There is lots to say about LIGO, but I’d like to focus on one point: the scale on the y-axis of the plot above (which represents strain, the fractional change in the length of the LIGO arms). LIGO is sensitive, over a wide range of frequency, to a strain of better than 1 part in 1022. In other words, it measures changes in the relative length of its 4km arms to better than a thousandth of the size of a proton. This plot should absolutely blow your mind. If not, perhaps I’m being too abstract? This is the equivalent of monitoring changes in the distance between New York and San Francisco to better than one ten billionth the width of a human hair. LIGO is a technical tour-de-force. It is one of the most amazing instruments humankind has ever built.

N.B.: In the comments, Brian137 points out that LIGO will be back on starting tomorrow (7/7/09) for a month-long run! (From the LIGO blog.) And nicolas points out that I was remiss in neglecting to mention Virgo, which is a French/Italian GW detector currently operating in Italy. It is easily as impressive as LIGO, since it achieves similar sensitivity with 3km arms (instead of 4km for LIGO).

CATEGORIZED UNDER: Science, Technology
  • Brian137

    Thanks Daniel. Good post and nice colors.

    According to this
    LIGO is scheduled to start a science run in two days. I think I read that VIRGO would run simultaneously, but now I cannot find that link.

    “LIGO is a technical tour-de-force. It is one of the most amazing instruments humankind has ever built.”


  • Jimbo

    You must be bored Dan, to leave LANL to attend this funeral dirge that is LIGO. All of the science runs have been posted on the arxiv for several years (5?), with the same result: Nada, zilch, zip. No reporting required.
    LIGO was sold to congress for ~ 280M$ ~ 13 yrs ago, with the explicitly stated intent of direct detection(dd)of gravitational waves.
    Then, somehow in the interim, they got approval and $$ for `Advanced LIGO’, and have given the impression ever since that the original LIGO never had a chance at dd.
    So why was it sold as such, rather than as a test bed ?
    The program has stunk from the git-go, compounded ever more by the ridiculous choices of seismically unstable WA state to politically unstable LA state.
    None of the other international attempts at dd, such as VIRGO, been successful either. In a decade, the final verdict on LISA will be in, but I predict the result in advance: Nada, zilch, zip.
    Not to be negative. I would LOVE to see GR’s smoking gun vindicated irrefutably. Yes, LIGO/VIRGO are technical virtuosities, but measuring metric perturbations to parts per 10^22, propagated to earth from 1000’s of light-years distant is grossly improbable & probably technically impossible.
    But because it had the seal of `Einstein’ upon it, the chutes were greased, similar to their grandaddy GP-B, which has not delivered the smoking gun on GR either.
    We have to face the reality that the ultimate verdict on GR may never be delivered. The ancillary evidence, tho is so overwhelming, that its status as the effective theory of gravitation will never be questioned until an unambigous discrepancy with experiment, not unlike the Lamb shift, is established.

  • hanfordgradstudent

    As a first note Dan, I thought your talk about using gravitational wave signals as standard sirens for studying cosmology was very good. That talk alone could provide blog fodder for more than a couple of posts.

    If one goes back to look at the initial proposal for funding the LIGO project, you will see that a second generation of detector technology (ala Advanced LIGO) was part of the plan from the beginning. No one had ever built a large scale facility of this nature before, and so the funding agencies provide the initial money with requirement for strain noise at 10^22 over a few hundred Hz band. Having reached that design sensitivity, denoted by the black line is Dan’s strain curve above, start-up funding for Advanced LIGO was granted. So Initial LIGO was sold as a test bed with the possibility of direct detection.

    Regarding the sites, I think Jimbo has things a bit confused in that Washington is the seismically quieter of the two sites. And the additional isolation systems at the Louisiana site through the dedication and perseverance of the staff down there have allowed LLO to operate at the same sensitivity as its sister facility in the PNW.

    The difficult with direct detection of gravitational waves beyond pushing our noise floor down far enough is that the populations of prospective sources are not well constrained. We are forced to run the detectors as long and as quiet as possible, waiting for strong enough (due to mass or relative distance) signals to pass by Earth.

  • hanfordgradstudent

    Oh forgot to mention, the LIGO observatory electronic logs are available for public viewing (read only). Just google it and the Livingston, Lousiana site comes up with links to the Hanford, WA log.

  • miller

    My understanding is that the first few runs were just preparation for later runs. You don’t build this hugely complex machine and expect it to run at maximum sensitivity right away. Likewise, enhanced LIGO is just preparation for advanced LIGO in 2013 or so. For example, compact binary coalescences (CBCs) are expected to be observed by enhanced LIGO at a rate of once every ten years. CBCs are expected to be observed by advanced LIGO 40 times a year! (see source)

    Now, that’s just CBCs of course–I’m not sure about other gravitational wave sources. We can always hope that enhanced LIGO will find something, but advanced LIGO is when we really expect to find something. Meaning, if advanced LIGO doesn’t see anything at max sensitivity, it might require some serious revisions of theory. And if it does see something, then we got a whole new kind of astronomy in our hands.

  • nicolas

    Very good post, indeed. Although I am a little bit saddened that, along with praise for LIGO, the existence of the VIRGO detector (that is currently more sensitive than LIGO in low frequencies, btw) is not even mentioned. I guess no bad intentions are at work there, but this omission looks quite systematic to me these days. I can’t find a logical, non-twisted explanation for that…

  • Sili

    What does it cost to build a Ligo? And where should we put the next? Australia? Antarctica?

  • TimG

    May I ask a question about your earlier post about gravitational waves? You wrote:

    One way to think about the necessary existence of gravitational-waves is through the following thought experiment. Imagine that you are on one side of a room, and that on the other side is a very massive object (e.g., a plutonium bowling ball). The lights are off in the room, but fortunately you are carrying a very sensitive gravitometer, so that you notice the force of gravity due to the massive ball: your gravitometer points right at the bowling ball. Now let us assume that Mark sneaks in, and gives the ball a good (preferably relativistic) kick. Since the ball has moved, the gravitational field should register the change. Newton tells us that the gravitometer would instantly adjust. But from relativity we know that nothing travels faster than the speed of light (including information), and therefore we need ’something’ to go from the accelerating bowling ball to our gravitometer, to tell it that the bowling ball has indeed moved. This is a gravitational wave!

    But surely the gravitometer also adjusts when the bowling ball is just rolling along at a constant speed, rather than being accelerated. However I was under the impression that an object won’t emit gravitational radiation due to constant velocity motion. So what tells the gravitometer to adjust in that case?

  • Chris W.


    Try thinking about the analogous question in electrodynamics. The same issue arises there. The underlying theories encompass fields associated with a static configuration of sources, a slowly changing configuration, and rapidly changing configurations (with acceleration). Of course the detailed geometry and distribution of the sources is relevant, in general.

    (PS: If the bowling ball is rolling, then most of it is being accelerated around its axis of rotation, although this is beside the point.)

  • Jimbo

    HanfordGradStudent: That’s comforting to know that Livingston is worse than Hanford, which is anything but geologically benign….
    A Richter 6.8 quake there in 2001 caused serious misalignment to many optical systems, requiring 3 months to repair. This March, over 280 small quakes have peppered the entire region, causing glitches in electronic systems. What if a similar mini-disaster should occur during an actual science run, attempting to capture a rare event ? Meanwhile volcanoes percolate…
    As mindboggling is a recent study by a high school student to model the impact on LIGO-Hanford performance during seismic disturbances. Worse still, the report acknowledges there were NO other prior studies of such ! Simply incredible: No one bothered studying this critical issue prior to site selection/construction, and then they bring a HS student in to do it !??!!
    Sounds like a joke, but its for real.
    This is in keeping with last years big press release,in a vacuum of direct detection data, that LIGO had been able to constrain sphericity limits on neutron stars. Translation: We cannot detect grav waves, but IF they emanated from sizeable surface topographies on neutron stars, we’d have seen it…? Oh Yeah ?
    If you cannot see ANY signals, how can you claim to set a limit deduced from data only obtainable from such signals ? Sheer feel-good tomfoolery to obtain good press coverage & placate the doubters.

  • Arrow

    TimG is right in that the thought experiment mentioned does not demonstrate gravitational waves have to exist. The detector will detect movement even for constant velocity motion which does not create gravitational waves so the fact that something informs it about the movement is not a proof that gravitational radiation exists. Rather it shows that spacetime curvature changes (which don’t have to have the form of waves) can propagate from the source to the detector.

  • Oded

    A thousand’s of the size of a proton.
    That did of course, blow my mind. More than the New York-San Francisco thing.

    At that accuracy, how do you rule out noise? Even wind on the building or heating by the sun could completely rule this domain, so how do you get by?

  • hanfordgradstudent

    What you see above in Dan’s post is the noise of the detector read from the channel sensitive to gravitational waves.

    The chambers in which the instrument sits are resting on a foundation of several feet of concrete separate from the walls of the building. One also has a complicated passive (and active in Louisiana) seismic isolation system made up of primarily springs and pendula. The beam tube in which the laser is enclosed is itself enclosed within a few inches of concrete. The chambers that hold the optics are placed in buildings in which the temperatures are regulated to within a single degree.

    Each noise source has some characteristic frequencies. Large far-away earthquakes shake the ground at frequencies between 0.03 and 0.1Hz, while the wind and traffic show up in the 1-3Hz seismic band. The isolation systems at LIGO do an incredible job reducing the ground motion’s effect on the mirrors by factors more than a million.

  • grbiersema

    Exciting stuff. The detection in the last 2 years of several new galactic SGRs (rather unexpected) with Fermi and Swift may help in increasing the odds on a giant flare from the local group – if i understand correctly, that may generate observable signals. Perhaps also good to mention that the pulsar timing array efforts are really getting very sensitive now, and may well detect gravitational waves before LIGO/VIRGO and LISA (well, the gravitational wave background at low frequencies, that is).

  • TimG

    Chris W.,

    I’m OK with the idea that the change in the field propagates at the speed of light, as in electrodynamics. Arrow’s comment states what I’m getting at more explicitly, which is this: If this propagating change requires gravitational waves (as Daniel seemed to be saying), why does it seem like it works just as well in the constant velocity case (where, correct me if I’m wrong, no gravitational waves would be produced)?

    (Good point that rolling involves acceleration. I guess I should have said “the bowling ball slides along some frictionless ice”, although that’s not a very natural image when it comes to bowling balls.)

  • shantanu

    BTW , although the speed of gravity is a very dicey concept, what Daniel has
    mentioned in the previous article that gravitational waves propagate at the speed of light is not strictly correct. Gravitational waves (like light) get slowed due to gravitational potential of all the intervening matter along the line of slight(aka Shapiro delay)
    as they propagate from the source to the Earth. This delay is
    approximately 5 months for SN-1987A distance of about 50 kpc.

    If you read I. Shapiro’s 1964 paper where Shapiro delay was introduced, the first para states “the speed of light depends on strength of gravitational potential along its path”
    Now same argument applies to gravitational waves.

  • Chris W.


    Keep it simple. Move the ball of plutonium into interstellar space, where it can just drift along (inertially!) at constant velocity with no intrinsic angular momentum.

    Here is something to think about (and again, consider electrostatics, magnetostatics, and electrodynamics, where the mass of the ball is replaced by a spherically symmetric electric charge). If the ball is moving at constant velocity, then there is an inertial reference frame in which it isn’t moving at all. In this frame the associated field is static; it changes with position (distance from the ball) but not with time. In a different frame the associated field is changing with time.

    In fact, in electrodynamics, you actually have an electric and magnetic field vector at each point. Maxwell’s equations determine what the magnitude and direction of the vectors will be. The finite speed of propagation of light is (from this perspective—more about that in a moment) a consequence of Maxwell’s equations. The magnitude and direction of the field vectors at some distance D away from the moving ball will, roughly speaking, reflect where the ball was at an earlier time (a time D / c, where c is the speed of light).

    As I indicated, from the perspective of the late 19th century the finiteness of the speed of light—ie, the speed of propagation of electromagnetic “disturbances”—was a consequence of Maxwell’s equations. Of course this is what got Einstein thinking at age 16. He took Galileo’s principle of relativity of inertial frames seriously, but he couldn’t see how Maxwell’s equations could be invariant under the coordinate transformations everyone assumed should apply to Newtonian mechanics (and Newton’s theory of gravitation). He gradually realized that the coordinate transformations had to be different for Maxwell’s equations, in such a way as to leave the speed of light both finite and invariant under those transformations. From this perspective the finite speed of propagation of electromagnetic disturbances is a consequence of the relativistic invariance of the laws governing electrodynamic fields.

    However, Einstein felt that Galileo’s relativity had to hold for all the laws of physics, and this could only be possible if Newton’s mechanics, and ultimately his theory of gravitation, were changed. And so he set about reformulating them. Elementary mechanics in the absence of gravitation was comparatively easy. Gravitation presented much deeper difficulties…

  • daniel

    Thanks for all the great comments!
    @Brian137, I didn’t realize that LIGO will be back up for a (brief) science run!
    @nicolas, Not mentioning Virgo was an oversight.
    I’ve added a note at the bottom of the post addressing both of these points.

  • daniel

    @TimG, hopefully the comments from Chris W. help. Indeed, the electromagnetism example can be quite instructive. An electron in constant velocity motion doesn’t radiate, but an accelerated electron does. Purcell’s undergraduate E&M textbook has a nice discussion of this exact point (with some very helpful pictures). In the same way, the ball rolling along the floor at constant velocity, although it does move the needle on the gravitometer, is not radiating gravitational waves. If the ball has always been moving along its current path, there is no new information to transmit. We already know where it will be at the next instant. If the ball suddenly accelerates (e.g., it stops moving, or changes direction, or speeds up), then a wave has to propagate to let us know (modulo near field effects). In the same manner as the electron. To see this explicitly requires GR; Eanna and Scott have a nice review which goes through all the gory details.

  • nota bene

    So LIGO is essentially the most sensitive microphone ever invented. I never thought of it that way. Pretty cool. I wonder if some of the machine’s raw output could be pitched up into audible frequencies.

    What’s the source of the spike in the 5th run between 300 and 400 Hz?

  • Arrow

    Daniel says “If the ball has always been moving along its current path, there is no new information to transmit. We already know where it will be at the next instant.”

    Information is still transmitted – that the ball keeps moving in the same direction, information that there is no change is still information.

    “If the ball suddenly accelerates (e.g., it stops moving, or changes direction, or speeds up), then a wave has to propagate to let us know (modulo near field effects).”

    I think the point is that near field effects can also transmit information which is not in the form of waves.

  • Optickal

    The comments have made the essential point that not only do you need high sensitivity but also high isolation from noise to see these incredibly tiny distortions in spacetime. LIGO’s environmental channels, which monitor ambient noise, show that wave motion and oceanic storms a thousand miles distant can be picked up by LIGO’s sensors.

    LIGO will indeed start a new observational run this month, with a sensitivity about twice that of the previous run (which was, by the way, at better than design sensitivity!). This run will last about 1-1/2 years, so an indisputable “event” just might be seen, but that observation would still be fortuitous. We’re all banking on Advanced LIGO, which will go online in 2014-2015 and will probe a volume of space 1000 times greater than LIGO could; but even then, it will take some time for the observatories to reach optimum sensitivity.

  • hanfordgradstudent

    nota bene: The spike at ~340Hz is a collection of lines from the vibration of the wires which are used to suspend the optics like pendula. The reason that it looks like a single peak is because the frequencies of the vibrations from each optic’s wires are very similar and the resolution of this noise curve is too poor to differentiate them. What frequency those vibrations appear as noise is dependent on what material the wires are made of and how thick they are. The thinner the wire, the lower the frequency of that noise peak.

    Human hearing is an incredibly useful tool that is regularly used in LIGO. Staff at the observatories often plug the output of the interferometer into an audio-mixer and listen on a pair of headphones. One does not need to pitch up the output since human hearing ranges from 20Hz to 20kHz, which covers the entire spectrum in Dan’s plot up at the top of the post. Audio is used to diagnose problems in detector subsystems and investigate short (non-gravitational wave) blips in the data called glitches. The gravitational wave signals we are looking for, such as coalescence of a binary star system will have frequencies in the audio band. There are few websites where you can listen to simulated signals. Just check out

  • Brian137

    Apparently, Advanced LIGO will be so sensitive that quantum effects will become significant.

  • coolstar

    A sobering footnote to the discovery of the first binary pulsar is that Hulse was unable to find an academic post in physics and astronomy. He did apparently work for many years at the Princeton Plasma Physics lab as did Congresscritter Dr. Rush Holt. I’d bet hard currency that Swarthmore now boasts of turning down Dr. Holt (a good congressperson and a good person, period) for tenure.

  • Pingback: Gravitational Wave Astronomy Summer School « Brute Force Physics()

  • Matt S.

    There’s also a gravitational wave detector located in Germany, called GEO 600. The arm-length is shorter than LIGO or VIRGO, but they used some nifty tricks to get sensitivity up to almost the level of LIGO (well, a magnitude lower, but still not bad).

  • Pingback: Barely Excited | Cosmic Variance | Discover Magazine()


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Cosmic Variance

Random samplings from a universe of ideas.

About Daniel Holz


See More

Collapse bottom bar