Thursday, February 18, 2016

Once upon a wave ...

The beginning
Mihai at Caltech
I first learned about gravitational waves from my brother, Mihai. It was 2001. He was a PhD student at Caltech, and had just started working with Kip Thorne. He was so excited about the topic that he convinced Kip to let him record his gravitational wave course. It felt so timely because LIGO was running and this was one of the first courses taught on the subject. These were the movies that Mihai and Kip produced before Kip became executive producer of Interstellar. Mihai had to buy a camera, which was eventually reimbursed, and learn how to use the equipment to turn the lectures into the almost-timeless movies that are available on youtube today. The movies were burned on DVDs and each DVD would take hours to burn, and many days to produce. The Interstellar movie was selected by Steven Spielberg, whom Kip called the King of Hollywood, about two years before I received my PhD from Cornell (around 2006).

 I was an undergraduate student at the University of Illinois under the supervision of Edward Seidel. I worked with Gregory Daues, Michael Russell (now Head Engineer for Curiosity) and Jason Novotny on one of the first grid portals. It proposed ideas that are now applied in Cloud computing (our project won two HPC challenge awards and the Bandwidth challenge at SC2002; the demonstration is described in our paper). Physics-wise, in collaboration with Jayashree Balakrishna, Greg and Ed, I started a project on computing gravitational waves from simulations of perturbed boson stars and soliton stars, which are objects made from dark matter particles that could mimic black holes.

I was 20 when I graduated from college, and started a PhD at Cornell. In vacations, I kept visiting Mihai at Caltech. We'd go to Kip's group meeting together. The Caltech group was so confident that LIGO and LISA were already 'done' that they started thinking about missions beyond LISA. Sterl Phinney was talking about the Big Bang Observer - a mission that would map the gravitational wave sky and would be sensitive to both LIGO and LISA sources, and to gravitational waves produced in the early universe. The technology is still not yet there, but perhaps worth re-investigating.

I started graduate school in 2003. Then numerical relativists could not yet solve Einstein's equations on the computer to simulate black hole orbiting each other. Their solutions would blow up. The first simulation of a black hole binary through plunge, merger and ring-down was performed in 2004 by Frans Pretorius, who had been a postdoc at Caltech. Later the following year, the development of new coordinate conditions also known as the moving puncture method allowed for accurate, long term evolution of binary black holes. We were all surprised by the simplicity of the merger part of the waveform. The movie of the two coalescing black holes showed in LIGO's press release was simulated with the Spectral Einstein Code - my PhD advisor, Saul Teukosky, led the SpEC effort. I remember him telling us that in 20 years everyone will use Spectral methods because they are the most rapid and the most accurate, and provide exponential convergence. Almost 20 years have passed. Results from SpEC have been used by Hollywood in the Interstellar movie and in the LIGO press release where they present the first gravitational wave detection.  Other methods continue to be used as well because they are more amenable to more general, non-spherical situations.

Tanja Hinderer in one of our hiking trips
I first visited LIGO in 2003/2004 when Mihai and I taught the first gravitational wave course from Louisiana State University. The course was based on Kip's Caltech lectures. It was Christmas, which was the only time we could leave our respective universities for such a venture. We were sponsored by Edward Seidel and Gabrielle Allen. Most of our students were LIGO members, and so they invited us to see the detector.  A detection was expected any time. They were so enthusiastic that they came to class on Christmas Eve and New Years Day. The student who had asked Mihai to give the lecture series in the first place was Tiffany Findley. She was in her early twenties, had two young children, and yet she served as the driving force for this class that must have been put above the needs of her family. I then organized a similar course at Cornell. We'd alternate between watching gravitational wave lectures and going hiking (see picture with Tanja above).

Mihai fixing the glass of my 1991 car
Dave, me and Andy on top of Cornell
As his thesis took shape, Mihai's interest lay with improving the sensitivity of advanced LIGO. He eventually found that the mirrors that optimally reduce the coating thermal noise, which dominates other noises at LIGO's best sensitivity, are conical.  This was contrary to the prior expectation that the beam shape should be as flat as possible to average over the bumps and valleys of the mirror.  He convinced Andrew Lundgren (Andy), David Tsang (Dave) and me to work on an extension of his project, where we showed that finite mirror effects are important. They create some resonances that perhaps could be taken advantage of in future detectors. Had Mihai not been as enthusiastic as he is, neither Andy nor I would have joined LIGO.


Sensitivity of Initial vs Advanced LIGO
In 2008, after finishing my thesis on R-modes - a kind of oscillations in neutron stars that are driven unstable by gravitational wave emission - I joined the Lee Samuel Finn's Penn State group and the LIGO collaboration. Andy's first postdoctoral position was at Syracuse University in the group of Prof. Duncan Brown. He was fascinated by the detector, and delved into his work body and soul.  

Long before he started working with me, Sam had developed the noise model that generates the LIGO sensitivity curves like the one above.  With me, he worked on a pipeline called MaxEnt. We aimed to infer gravitational waves for unmodelled sources. I remember giving a talk about 'detecting things that go bump in the night'. The burst search could, in principle, detect any signal that appeared in both detectors, and was not identified as noise or some kind of instrumental effect.

All members of the collaboration had to go to one of the sites for science monitoring. I went to the Handford LIGO detector for the shift starting at midnight on Halloween 2010.  Although we did not find gravitational waves, I learned about the instrument and about the vitrification plant (turns radioactive material into glass) that was near there. Workers at the vitrification plant retire in their early 50s with pensions; also, cancer is not the most prevalent disease when the poisoned while working in a toxic medium - immune system disorders like multiple sclerosis are more frequent.

Test waveform from my work with Sam
My job at Penn State was to combine the polarization of the gravitational waves via Stokes parameters into circular, linear and elliptical polarization and use these to infer properties of the source. We spent a lot of time injecting fake waveforms from 20+ solar mass black hole binaries first into white noise, and later into real LIGO noise. Some were from the Georgia Tech numerical relativity group, and others we generated from post-Newtonian approximations glued upon some damped sinusoidal for the ring-down part of the waveform.  They looked eerily similar to what was finally detected. We thought LIGO would first detect mergers of 20-30 solar mass black holes because they would be the loudest, but we only guessed that this population of black holes existed and that enough such binaries were close-enough to be seen by our detectors on Earth.

The Spinning Template Bank
Ohme, Lundgren et al. PRD 88 (2013)
When I got pregnant with Edward (the youngest author from the books above), I convinced Andy to move to Penn State. He moved the day Edward was born, and worked with Ben Owen, who had been the primary developer of LIGO's original template bank. The primary method of searching for gravitational waves is through matching templates that are derived from Post Newtonian theory. These templates are correlated with data to identify the waveforms from black hole or neutron star binaries. Andy understood how the template bank worked form Ben. But just like most stars and planets spin around their own axis, so do black holes. This spin was not taken into account. The collaboration had tried to sort this problem since LIGO started running, and failed to find a solution. Andy came up with a geometric trick that treats each post-Newtonian order as a coordinate. It makes up a high-dimensional flat space, whose points can be easily computed (see picture on the right).  Andy was so preoccupied by this problem that he thought up this solution in a dream and it happened to work. The papers and the code that implemented this idea took years to finish and involved a number of collaborators.

LIGO's Science Run (O1), which ended in January 2016, is the first to search for spinning black holes. While Andy continued his core-LIGO work at the Albert Einstein Institute in Hanover, Germany, I moved to Switzerland and started work related to atomic clocks.

The detection
There are number of very good news articles out there about the detection itself in addition to the technical papers, and about how LIGO began. In a nutshell,  two black holes of 36 and 28 solar masses collided and formed one big black hole. They radiated 3 solar masses in gravitational waves. About 1 billion years after it happened, the two LIGO detectors were able to detect this cosmic storm on Earth!  Gravitational waves are how black holes talk to us. Most scientists thought that advanced LIGO would find gravitational waves, but just not in one of the Engineering runs.

The first emails after the detection - many sent by Andy.
Just before the detection,  Andy was appointed detector characterization chair with David Shoemaker.   So, more than 10 years after sitting in Kip Thorne's group meetings, and watching him lecture on gravitational waves, he became the second person to see the first gravitational wave. Marco Drago was the first to find the event, and report it to Andy as a potential injection. Andy checked all the channels and found that it was real. He talked to his other colleagues in nearby offices, and then encouraged Marco to send the first email that woke everyone up.

His first job as Detector Characterization Chair was to lead the first phone meeting after the event. In that meeting, Gabriela Gonzales and Michael Landry confirmed that it was not a blind injection, and Andy stared shaking. He regained his composure enough to continue running the discussion, and although it ran overtime, they went through the agenda as planned.  The last time he shook like that was when he drove the space shuttle at the age of 10. No, it was not the real space shuttle, it was just a very realistic simulation built by NASA. The gravitational wave was real.

Disclaimer: This detection is the work of 1000+ people over 40 years. I could only present it from my perspective. It is not meant to disparage the contribution of other people. I am writing this down for my family because one day my children will be old enough to understand this history, and perhaps by then I will no longer remember it.

The 40 years. People are surprised by the 40+ years it took to find gravitational waves. However, this is often the type of dedication that is needed for success, and LIGO is extremely successful. It's mind boggling that they can measure the stretching and squeezing of space-time at 1/1000 of the diameter of a proton.   Science is not just ideas - it is work - even theoretical papers take years to complete. Space missions take tens of years. Movies take years, too. The script for Interstellar was selected by Spielberg some 10 years before the movie was completed. So, while this detection is something I've been waiting for my whole career, I am not surprised it took this long, and I am thankful that it did not take longer.
For more about the universe and about gravitational waves, read Edward (age 5) and David (age 8)'s books from the side bar above. Andy is Edward's father, but played no role in this book because the detection was secret. There are also a number of relevant technical articles that describe the detection.

Press I got for being the only scientist at the University of Zurich with some LIGO experience: SDA news (,, UZH news (no English version; there are plenty of good articles and videos with more prominent people, though).  

1 comment:

  1. Congrats! It would be cool if you could post more often.