GROWTH Global Relay of Observatories Watching Transients Happen

It is almost unbelievable how much we have learnt from GW170817. Now we use GW170817 to derive conclusions in current studies and to make predictions for future scientific projects. It is like a legendary creature whose existence was uncertain awhile ago, but is now a subject of a full documentary. However, are all the legendary creatures exactly the same? Each new event will likely have an exquisitely unique story to tell, that may challenge what we have learnt so far. Besides, only a large number (a "population") of such events can allow us to measure precisely some key pieces of the puzzle, including for example a new method of measuring the ratio at which our universe is expanding. Expect surprises in the next few months!

Everyone needs to first imagine a robotic voice waking you up in the middle of the night with, "Attention, humans. This is the GROWTH Target of Opportunity Marshal (a system that GROWTH astronomers use to keep track of triggers and follow up activities)." I haven't figured out how to get it to be Arnold Schwarzenegger's voice yet, as that might be more inspiring to get out of bed after watching Terminator one too many times. Starting the marshal on your laptop , you see a new event at the top that has a Binary Neutron Star (BNS) or Neutron Star-Black Hole (NSBH) tag, and you are a bit more willing to get out of bed (I still prefer to observe from bed though!). The GROWTH team has worked hard to automatize the observation plan generation, scheduling, and retrieval of observation information from the telescopes, so it is ideally just a click of the button to make the observations you want happen. In the case of the S190425z LIGO trigger (190425 indicates the date of the detection), the possible area of the sky where the collision could have happened was in the northern sky, which is visible to the Zwicky Transient Facility (ZTF) at the Palomar Observatory. ZTF is a wide-field camera that can rapidly scan large swaths of the sky to look for transients associated with the gravitational wave triggers. So, ZTF was already naturally doing what we would wanted. Two other facilities we use, Palomar Gattini-IR and the Kitt Peak EMCCD demonstrator (doing targeted galaxy observations) were triggered based on the schedule in the Marshal. At that point, it was mostly a matter of waiting for possible transient candidates to begin to flow in.

I did observe from bed as soon as the S190510g LIGO trigger arrived. I was in England asleep in a hotel room when the trigger went off. I think it was about 4 in the morning. I do not have an international phone plan, but somehow Igor managed to make my phone ring via Wi-Fi and wake me up (I think it took about 5 tries). I picked up and Igor told me there was a BNS merger. We called Steve Heathcote, the director of the National Optical Astronomy Observatory - South in Chile, woke him up and got the OK to trigger DECam - the Dark Energy Camera . Five minutes later I was sitting in bed, with my laptop open, observing. In another window on my laptop, I was processing all of the data in real time at a supercomputing center in Berkeley. I was hoping the internet connection in my hotel would not give out. I think about 15 people are mobilized every time we trigger DECam, across at least two (this time three) continents. It is very chaotic, but always exciting and fun. Things are changing by the minute as we all learn new information, and we are all connected on Skype, talking to each other from different corners of the world. Even over the phone, you can feel the excitement of discovery in the air. It all makes gravitational wave hunting unlike anything else I have experienced in science.

When LIGO/Virgo provides an area where the possible merger has happened which is not in the Northern hemisphere we would not be able to search for it with ZTF alone. What we need is a valuable discovery telescope located in the Southern hemisphere. Since 2018, Danny I have had the privilege to lead the follow-up observations of gravitational waves with a powerful instrument called the Dark Energy Camera, also known as DECam, located at Cerro Tololo in Chile. One nice aspect of our DECam observations is that we make the data immediately public to the community. DECam is mounted on a telescope with a mirror more than 157 inches in diameter, which allows us to collect much more light than ZTF, which is mounted on a 48-inch telescope. Because it is bigger, DECam can observe a sky area about 16 times smaller than ZTF but at the same time it allows us to look deeper in the sky, capturing very distant and faint sources, which would be impossible to catch with most survey telescopes. We used DECam to follow up on the S190426c and S190510g LIGO/Virgo triggers, using optimized observing strategies that we keep on perfecting. The latest LIGO/Virgo observing run (O3) is a very different game than O2. It is more challenging because mergers are detected at larger distance and are more coarsely localized, so the design of the observing schedule is particularly tricky.

In a typical night of DECam observations we will find about a hundred thousand things that appear to be changing in brightness. Maybe a hundred of these (1 in 1000) will be real astrophysical sources of variability, and of those, we will have resources to get follow up observations on bigger and more sensitive telescopes for maybe two or three. So it is a problem of picking one or two of the best objects out of 100,000. And usually we need to do this within minutes. It is a huge task.

The way we deal with this is using artificial intelligence (AI). We have trained computers to 'look' at our data and distinguish the real, interesting objects from the bogus ones. Our algorithm assigns to each candidate object a number representing how likely the object is to be real. It can do this for all 100,000 objects in just a few seconds. This way we can rapidly know which objects we should be looking at more closely.

This AI classifier is just one small part of a larger pipeline our team has written to process data from DECam. We run it on a supercomputer in Berkeley, and we have optimized it for speed. At peak, it can process 1152 DECam images simultaneously, allowing us to find the best candidates super fast. For the most part, all of this happens automatically. When things are running smoothly, we can sit back and let the pipeline do the hard work.

Here is where team-work shines bright, in my opinion. You can own a great telescope, the best in the world perhaps, but with that telescope alone you will never be able to collect all the multi-wavelength information that you need to truly tell the story of a BNS or a NSBH merger . Besides, plenty of observations are needed to classify the discoveries and identify them among hundreds of new transients. This can be done by monitoring the evolution of their brightness, their color, or taking spectra that carries information about the elements that constitute these new sources and their surrounding material. It is important to secure several observations every few hours, both photometric and spectroscopic. Only a network of telescopes with different characteristics and sizes (sometimes small, sometimes gigantic) can make this possible. In the case of S190426c and S190510g, the GROWTH collaboration used its global network of facilities for this task, ranging from optical to infrared and radio wavelengths. In addition, we immediately shared the coordinates of all new candidates with the community. In this way, everyone can contribute their observations, in a global effort to nail down the true counterpart.

GW170817 was indeed an incredibly lucky event because it was well localized by LIGO and it happened so close to us in the Universe that even small 1-meter telescopes could easily see it. The recent triggers were for events that were poorly localized and occurring much further than GW170817. For me, the highlights of the analyses we did was to show that we would more than likely catch events if they are within the areas we are imaging even if they happen at distances further than the more recent events. In this sense, we are taking some of the luck element out of the discovery process with the resources of GROWTH, which make it possible to vet the many candidates we have when we need to cover large amounts of the sky.

During every follow-up we serendipitously discover many sources of interest to the time-domain astronomy community. These are transients and variables both Galactic and extragalactic, such as supernova explosions so the data that we acquire can be used by our colleagues to conduct different but equally interesting science with it. I think that one of the supernovae that we discovered with DECam will even be presented in the thesis of a GROWTH PhD student.

The thing I enjoy the most about my work is the opportunity to think creatively. Most of the time we are working on things that do not have a clear solution, so making progress requires coming up with new ideas. I think the best ideas come from working on projects like this, where there is so much that is unknown that there is a lot of fertile ground for thinking differently. Nine times out of ten, my ideas are wrong, but it always a lot of fun to explore them, and they always lead down interesting new paths. And every now and then, one will stick, and it is immensely satisfying when that happens. I think this was the most important thing I learned from my advisor when I was a graduate student.

I personally like the people the most, which we will get to in a bit. Most of my time on GROWTH is spent working on the technical side, writing software to sensibly schedule the telescopes, understand how much of the skymap we have covered, and also ensuring that the Kitt Peak EMCCD Demonstrator observations get taken and reduced. Scientifically, as a member of LIGO, I think a lot about how to combine the analysis of compact binaries in gravitational waves and the electromagnetic data to understand these objects better. It turns out many of the ways to characterize them such as how massive the neutron stars are or how tidally deformable they are, are quite complementary.

What excites me the most in this job is the discovery aspect. Every new image may be hiding a secret that we are about to reveal. When the game is on and a stream of data flows into our computers, every minute could be the one that changes the history of this new research field. Also, as Michael mentioned, people's excitement becomes contagious in the hottest hours, making it possible to push our limits of endurance and our daring in decision making.

Ok, so I am the pessimist of this group, as the others will tell you, so I would put the probability as very low, given the likely low rates from LIGO-Virgo and that most of the NS-BH mergers are predicted to have mass ratios that aren't predicted to yield very bright counterparts. But obviously, I will be excited to be proven wrong.

I wouldn't bet on it. But if there is another gravitational wave detection of an NS-BH merger in O3, we will no doubt be on it as fast as possible with the powerful combo of ZTF in the North and DECam in the South. In either case, the number of NS-BH mergers detected in gravitational waves by the end of O3 should give us a better handle on the rates.

Maybe we already have one in the basket, S190426c, whose classification is still uncertain and we may never know for sure. I do believe in rates, which are pretty negative in the case of NS-BH mergers. However, O3 started well. I will go for one solid detection of NS-BH merger during O3. I am ready to bet one full week without coffee, which may sound crazy to whoever knows me a little bit.

This is one of the reasons why I chose to do this job! Of course it presents some challenges, starting from the fact that organizing meetings at friendly hours for everyone is pretty much impossible. In addition, it easily happens that collaborators from such a diverse pool of cultural backgrounds think and act in very different ways. However, facing these challenges is definitely worth it. For example, it benefits our science when we work with fresh minds in continents where it is daytime, while telescopes are observing in other parts of the world at night time. Personally, I am learning heaps from many people in the collaboration, from students as well as postdocs and faculty members. What I am learning ranges from time management to coding, to the scientific interpretation of our results. I do enjoy learning from expert astronomers leading big teams. Most of all, I think that this experience is teaching me that respect is the basis of real success.

This is all part of the fun of the experience. Coming from LIGO, where interactions with people around the world are very common, I find GROWTH to be very similar. What makes it unique though is the real excitement and chaos in the moments when there are interesting detections by LIGO that require follow-up. Luckily, astronomers, more than most people, are used to odd hours, and this 24-7 race for detection yields some pretty weird hours. Working with people around the world, with many different experience levels and wavelength expertise, means I learn something new just about every day (and then promptly forget it because it is 2 AM and discussions have been going for hours, but it sinks in eventually). So in this sense, maybe one can view LIGO as a day job, an exciting one where you are always pushing the limits of what is possible with these detectors, while GROWTH has the excitement of a start-up, where the level of coordination amongst different facilities is novel and has all out sprints when it is required.

International collaborations are pretty common in astronomy and physics these days. I have been a part of several international collaborations over the years, which have connected people across international borders to make scientific discoveries. But where GROWTH really stands out is in the level of engagement of its members. Everyone pitches in. Everyone brings something unique to the table. And I think everyone in the collaboration feels like they have a real, personal stake in the results.