 Hello. Good morning, everyone. It is a great pleasure for me to welcome my old friend from a long, long time ago from our college days together at the Indian Institute of Technology in Delhi, and actually even before that. So I've known him for about maybe 50 years or something, but he is now the George Ellery Hale Professor of Astronomy and Planetary Science at Caltech, where he is in two divisions, the Physics Math and Astronomy Division, as well as the Geology and Planetary Sciences Division. And he has been the Director of the Optical Labs at Caltech, which if you know Caltech's command of JPL and all the other observatories is a big deal for, he's been Director of that for many years. And I have a CD here where I can tell you at least roughly speaking that he is a member of the US National Academy of Sciences, the Indian National Academy of Sciences, the Royal Society of London, the American Academy of Arts and Sciences, and so on. He has won many prizes, including NSF's Allen Waterman Prize, Presidential Young Investigator Award, a fellowship from the David and Lucille Packard Foundation, the Helen Warner Award of the American Astronomical Society, the Jansky Prize of Associated Universities Incorporated, and the Dan David Prize from Israel. So he's a very celebrated astronomer. The thing that I remember most about him is he's, A, a lot of fun to work with, and B, he is very dynamic. So I've seen him work like crazy. And he's done a lot of things when he was, even as a grad student, he discovered the millisecond pulsar. Until then, we knew only the slower kind and these are much faster pulsars. And since then he has been interested in the study of compact astronomical objects. And he sent us a talk abstract, which was very complicated with lots of astronomy. And so we said, hmm, well, we like that. And so then he created a whole new talk to talk about the astroinformatics aspect of astronomy. So I'm very thankful to them for having done that. Hopefully that's more understandable and interesting to us here. At the same time, feel free to ask him questions on astronomy in general after the talk. So without any further ado, please welcome. Thank you. One of my Japanese colleagues, this is such a nice open, I've always shared this, I've shared from Hawaii because I spent a lot of time observing it on the big island. And of course there are similarities between the Okinawan shirts and that from Hawaii. I guess I have to buy one from Mango House from my way out. So it's a pleasure to be here. It's always nice to be in Japan. There are two things I always like about Japan. One is you don't have to do it. There's no tipping. So it takes all the stress away. Just go, you pay, there's sort of no, too much too little and so on. Other one is the love trains and they're really part of the time. But today I discovered this light side effect of that particular thing which I always like. The institute had arranged a taxi to pick me up. It said 945. I came a little early, maybe 942. And there was a taxi with the voice of the car. So I said, okay, that's me. So I got in. And the taxi driver said, well, I can't leave right now. And I said, okay. Then I said, no problem. I do all the work all the scientists do and there's a lot of time to catch up and just start getting on to the email. That's what you do, right? That's the stuff you do in between doing science now and then. And then at 9, I noticed that 951 or 52, he left. So we reached exactly 10, which is where it says to be greeted by Dr. Purwait and the auditorium. And then I remembered, yeah, this is of course the case in JR a few years ago and now I remember suddenly. They shoot an apology because one of the trains left the station one minute early. Okay, so this of course is very impressive, I must say. Okay, enough of that. So I've been making some broad statements and I expect actually for those of you, I guess most of you don't know me, but I love to be challenged. Absolutely, there's nothing as good as an intellectual fight I welcome, in fact. So I'll make some statements. I'll try to make it a little outrageous to get you to challenge. I know this is Japan but there are some non-Japanese and maybe by induction, please challenge. It's fine, completely fine. Okay, so let me start off. So I'll make an assertion that most of science is phenomenological actually. Astronomy, like biology, geology, and you can add a new subject, exoplanetology. These are phenomenological subjects. Well, what does that mean? It means you go out and observe, measure and do things because you can't imagine them yourselves. Okay, that's what phenomenology is. Which is what most of science really was. And the amazing success of physics in the last century has, I think, worked many people's idea of what science is. And I think though that these subjects, which have been doing phenomenology, will in fact now be the subjects of this century. You know, what's the big thing you've heard from fundamental physics in the last 20 years? Okay, they've got the mass of the exposition. Yeah, fine. Okay, it's one of them that we're done. It's not particularly a thriving field right now. No, it's true. It's not particularly a thriving field. It's an important field, but it's not something you get up and read every day something's happening. Okay. But these subjects are phenomenologies of what to take off. And my thesis is that because they're very driven by technology, and therefore they're about to take off. Okay, so let's understand what do phenomenologic subjects have in common. Plus you've got to go and discover. That means you have to go and map things out. So let's start with something which is usually divided, which is your stamp collecting, your butterfly collecting. So yeah, it's fine. You go out with your eyes or camera, or whatever you take pictures of things out there in the wilderness. And you realize, first the butterflies, they're winged creatures, but they're not the same, grasshopper and butterfly, dragonfly and birds, they're not the same at all. So first you have to just go discover. Okay, then you start seeing patterns. And that's when you'll say there is a family called butterflies. There's a certain mechanical way of doing things. And eventually there's a deep molecular basis of who they are. Birds are also a mechanical way of doing things. So that takes what is what I call a search for patterns. And once you have these patterns organized, that's when you come up with what you call physical models. And this is where physics enters, and it's been spectacularly successful in certain aspects of understanding, which is physics is a subject that aims to build models by first coming with some laws and eventually rules to explain this or to account for these patterns and then say, can I reduce it? So phenomenology is all about complexity and classification, and physics is all about reducing it. And the goal of physics is to reduce all that to as few rules as possible and ultimately to a few numbers. So in that sense, the success of astrophysics in the last 20 years has been spectacular. You can specify the geography and makeup of the universe in seven numbers, which is really spectacular. Of course, seven numbers, and then you have to know some rules, which are the rules of GR to explain to dynamics and certain rules of deep fundamental physics to explain the particle and composition. Those are some rules there. So this is sort of phenomenologic subjects. Now let me come to mathematics and technology. Now there are some mathematicians who are completely astonished you first, and then they'll irritate you next by telling you mathematics and technology are the opposite ends of this thing. They're the same things actually, but opposite in some ways. So mathematics is an integral part of natural sciences. Mainly because it allows the one to compute signals. It's a computational device. I know that many mathematicians are saying, what about all the pure thinking, the symmetry and all that? No, no, no. There's a lot of literature out there, the unreasonable effect in mathematics and explaining theories and all that. But actually the origin of mathematics lies in poetry and in practical results. What you call as number theory. There's a fantastic lecture by a field theorist, a field medal winner, on the origin of number theory from counting the number of long and short vowels in poetry goes to about 8th century AD. And practical results is obvious. That sort of, if you want to stack oranges and you say what is the best, what's the way I can stack oranges maximally without it falling down? Which all children like to do. Or for oranges, immediately we start to discredit the head and ships. Okay, that's sort of mathematics. That is really mathematics. So the unreasonable success of physics and mathematics in particularly where you formulate a physical law and then from that you make a prediction has been really spectacular in particularly physics. I would say it led to a wrong idea that somehow that this is the basis. You even have books like God was a mathematician and all that sort of silly things. It's not true. These are all our constructs to compute signals and sometimes you see a success. And sometimes when you go out in the wilderness you see a rare animal and you don't attribute that to God or any such thing, you attribute to luck. So astronomy has had a fantastic growth of all the phenomenological sciences which are largely count geology, astronomy and biology. Astronomy had a fantastic growth but biology has overtaken that in just in terms of acceleration. Astronomy has a very valid branch of astronomy called theoretical astrophysics and I'm sorry to say there is nothing equivalent to that in biology. There is no real fundamental theory of biology yet. It will come, it will come. It will take a while. Remember, astronomy had a very long history whereas biology in that sense is very recent. Okay, so let me give you an example of what I mean by mathematics technology because these are somewhat unusual ideas. I completely admit that what I'm telling you are really not what you'd call mainstream. Okay, so let me give you an example of my thesis. So let's start with planetary motions. Okay, centuries ago, thousands of years ago people at that time, I'm just astronomers noticed that there are two kinds of things in the sky. There's a pattern of objects that maintain distance with each other. The same pattern rises and comes back year after year and then there are a few bright objects that move with respect to this. There are the wanderers that, i.e. what you now call as the planets. Okay, so this was known and these planets are confined to a certain plane, a very important plane so much so that all the astrological signs come with the planets. This plane is where all the astrological 12th signs, animals or whatever you want to call your head. With the very first telescope, i.e. technology, when Galileo's first telescope was put together, it may look small to you today, but remember it was the first telescope. The jump in telescope diameter went from 0 to maybe 10 centimeters. Well, that is infinite jump actually in technology. Okay, so as soon as it appeared at the moon and Jupiter, it discovered the moons of Jupiter. Well, that was a major finding to realize that the bodies, that all the other bodies, in fact, Galileo had really thought through it. He very quickly understood the role of gravity, but it had to take a century later for Newton and the apple and all that sort of stuff to happen to understand that gravity is there to keep these objects going around in a circle. Okay, now Brahe made very careful observations of planetary positions. So you may not appreciate this, but he convinced the king of Denmark to invest what amounts to about 7% of GDP in a year to build the great observatory. That's a lot of money. There's probably nothing in parallel comparable to modern-day time scales. Maybe the Apollo program would have won. Well, let's say that's about 3% around for 20 years. Yeah, okay, that would be fine. That would exceed that. But nothing other than that. And the result was a cat-long half-minute position on sub-planetary positions. He was an extremely skilled guy then. I said, how did he do this guy? Because one of the things that I do in my position is, ideas, let me tell you, Fidel Castro once said ideas are cheap, and I thought that was a bit strange statement. But now, over time, I realized ideas are very cheap, actually. They're easy to come by. In fact, the end of the day, we can get together, you know, I like to sing them all scotch. I'll tell you, after that, ideas go even much faster. Okay, the main problem in life is to make or execute these ideas, which usually requires raising money. So I'm always interested in how people raise money. So it turns out that prior from the king, he was an alchemist. So at some point, he must have in his NSF proposal said, the goal, aim of our observatory is to convert whatever base metals into gold. Okay, but of course, he did exactly what we want. To be right, some aid, some nice looking BS, and then get the money into what we want anyway. Okay. So then he left this stuff. He died, and Kepler is the one who distilled this into the now, you see, he already discovered the patterns, but the patterns now become rules, which are the laws of Kepler. Okay, patterns are general statements, but rules are very exact things. There's Kepler's first law, the second law, and the third law. Okay. And those he made, which led as soon as Newton developed a theory of gravity, which said, okay, there's an inverse square law for force. Then it took 20 years for Newton, and led me to develop the mathematical tools. So this is what I'm saying. Mathematics is technology, actually. And just because they don't look like engineers, we confuse them with something else. So it took them 20 years, okay, for these two people to develop Kepler's, which is needed to solve any equation of dynamics, because force has a second derivative, and so you'll have to learn. Okay, that led to modern mechanics. Gauss, in fact, computed the trajectory, so you can see that all mathematicians here, and, in fact, he invented what he called as the least square method, and made a prediction of a minor of what would be a series of something, and said where it would be. Okay. So you can see the beautiful example of phenomenological subject and the rule of mathematics and technology, all in this order considered the most classic example. Okay. So now let's switch to astronomy. So here's my way of astronomy. Between 1930 and 1970, so modern astronomy, roughly speaking, is about maybe 120 years, or roughly speaking, if you wish. The real basis of astronomy begins, I would say, with the Sahai equation and spectroscopy, which is the ability to look at a spectrum and say something about composition and temperature and density. But now you can talk about physical units. Okay. So the basis for this progress of stars was spectroscopy from large telescopes, and the physical model of the Sahai equation. Okay. Between 1950 and 1980, the synthesis of elements was understood. Now some of you don't think about this too much, or if you're a chemist, you assume the periodic table is granted to you. But remember when the universe began, the periodic table only had hydrogen and helium. You don't have been a fantastic time to be a chemistry major, because there would be no chemistry. So the process would be almost nothing. Okay. So the synthesis of elements was understood, you know, this is the major, the real details for all these cross-sections is really all about nuclear physics on the Manhattan Project. In fact, all the guys who did the Manhattan Project later ended up working on what it was called as a nuclear astrophysics. Okay. Neutrons start black holes. Wow. They're only found there. You can sit around and think as much as you want. Until it's there, it ain't there. This radio-engineers and physicists who understood X-ray methodologies applied to the heavens and discovered these objects. To me, the most impressive, these are all very impressive. The one impressive because I can relate to is sort of during my peak production time is cosmology understood. I find this the most fascinating thing that you sit around and contemplate, but then you go and make measurements of the small structures in the universe and they grow. And we can actually explain this quantitatively now. Okay. But this relied on mapping the sky at centimeter wave technology with very, very sensitive receivers. Otherwise, there'd be no understanding of cosmology. Otherwise, it would be stuck with the GR equations and there are many solutions to that. Okay. The growth of galaxies. Okay. So it tells you the last construction and how did matter get organized in galaxies and then star formation. And that's all due to, I would say, the fundamental technology that is necessary is optical CCDs, which is commercial, though first invented in the commercial realm but really developed by astronomers and nearly infrared detectors, which is still a very strong defense activity. Okay. Those are the technologies that draw all this progress. Okay. So now let me get into my main part of this background, which is time domain astronomy. Okay. So for a long time, the idea, and actually the first thing you do when you get into a subject, you map, you try to understand the landscape. Okay. So first thing you do in astronomy is try to understand landscape, which means how is this, what does the sky look like at near-infrared wavelengths, radio wave bands, x-rays and so on. Okay. So that's what astronomers did in the previous century. And so we had a pretty comprehensive multi-band understanding all the way from literally 10 megahertz to something of the order of 10 to the, to cosmic rays, which are like 10 to the 21 electron volts. So that's a huge range of energy scales understood. Okay. So once you map the sky, what's the next thing you do? Well, the only thing left next is to map it again and again to look for things that are changed. And that's in brief, it's time domain astronomy. So if you look up once it's astronomy if you look up more than once, it's time domain astronomy. And what do you find? You find new stars, supernovae, or moving objects, because a moving object is not in the same place when you look them twice. Which means, and moving objects are all basically solar system objects, asteroids and debris and so on, so forth. Okay. So much of my talk will focus on supernovae because time domain astronomy is a methodology. It's like saying, after you invent the electron microscope, you can't really give a token and say what it is, but you have to then choose something special and that's all. I'll talk mainly in supernovae. And so, let me go back. The time domain astronomy and the study of supernovae began with Professor Fritz Wickey. He was the first physicist appointed to do astrophysics at Caltech in 1926, 27. Caltech had to receive six million dollars from the Rockefeller Foundation for the construction of the 200-inch project by George L. Ruby Hale. And they realized they didn't have an astronomer. So, and Wickey was, if you ask me, the three great people I've known in the last century in astronomy, and Wickey was one of them. He was absolutely brilliant. He had fundamental patterns and jet engines. He's well-known in the departments of philosophy. And he was the most imaginative, sort of a very lateral thinking person. And surprisingly, is that he did everything by himself. So, Wickey, he's observing, he reduces the, develops the photographic plates, makes the measurements, and writes single-order papers. And that's what he did for almost all of his life. I respect that immensely. If you use that right here, we would hardly be anywhere. And I like Wickey. I admire him so much that I call everyone that I measure everything by this Wickey scale. Most of the time, we meet micro-Wickey. This is pretty. Once in a while, I meet a Middle East Wickey. Okay. So, he started this modern field of time in astronomy with an 18-inch Schmidt telescope at Paloma in 1936. Bada and Wickey showed their two types of explosions. Now, many of us use the word phase space. This term was not this, but people are using it in a different, they give it a different name, morphological plasm. But phase, the concept of phase space and exploration goes to Wickey. So, any time you use phase space, you have to remember it's Wickey. Okay. And there's a whole textbook on this phase morphological approach to science and so on. So, in a phase space, you try to look at phenomena with some major variables plotted. Okay. Simple now, but many things which are simple now, they're more not obvious then. So, here it's the prime scale of the event. So, you look at something in the sky, a new star or new event has come up and you say, how long can I see it? Well, there are all sorts of bias issues. Okay. There's a difference for bigger telescopes. Okay. What is sensitive? And so, don't get worried. Life is full of biases. You couldn't live with biases, actually. It's a part of evolutionary biology. So, that's not going on what time scale, what exact is it? It's one day, ten days, hundred days, and one year. That's how that phenomena is visible. And this is how bright it is. Okay. Now, astronomers use a unit called magnitude and I can tell you there are two reasons why we do that. The first one is, the first one is kind of a weird thing. It's got like minus. So, as you go up, it's brighter and in a log sense and sort of these negative numbers and so on. Reason number one is straightforward. You know, many of these cases are actually pretty bright guys and we want them to come into a song which is actually a fairly easy field. So, once you put like negative numbers meaning more negative and log and so on, that seems to get them all confused. And the ones who do persist, okay, you then tell them it's all in solvency of luminosity and then finally the ones who are still there, we just want to use CGS. That gets rid of most places that I know, immediately because we use MKS. Anyway, so there are all these units that the sum would be minus four and a half here. Okay, and two and a half units and this is ten. So, they distinguish NOVI which are ordinary surface explosions on the white walls and supernovae and we now know these are thermonuclear explosions. So, these are stars where carbon and oxygen, there's a core, let's say, about a solar mass of carbon and oxygen, okay. It's an equilibrium with heat and repression, balance and gravity and then there's a nuclear ignition that goes off and the carbon and oxygen fuse all the way to iron and you notice something very special about iron because iron is the most stable nucleus. You can't fuse iron to get energy. That's why fusion reactors will all involve elements below the ion peak and fission will be above the ion peak. Okay, I'm sure all of you remember this from 7th grade or something like that. Okay, it's a joke. I didn't learn this from 7th grade. Okay, so these are the fusion reactors thermonuclear explosions and okay, so what's the supernova? Our supernova is a star that comes up. It wasn't there a few days ago or last night and then it's there. So, this is a galaxy called M-51, M-51 also called the whirlpool galaxy. It's a beautiful galaxy. If you have an amateur storm or an amateur storm, you should look at M-51 and it produces supernovae once about every 10 years. Our galaxy, we believe is about once every 100 years. The last visible one was about 400 years ago. Okay, so it is this that new elements are made. So elements are made when stars undergo fusion reactions throughout their life which is the super-lozzy elements like carbon, nitrogen, oxygen, the stuff that is important. Organic chemistry largely is made when stars are living, but the high-z elements are made when the star dies. And the very high-z elements are made when the star dies twice. Okay, so very quickly, so this is how it began. So what it is, I've very clearly removed it twice. So what you're seeing here is a redshift and it's showing you the structure on very large scales, on hundreds of mega-bar six scales here and it's showing how matter is getting organized. So let me, this beautiful structure is evolving. By the way, if you're a biologist, you may even think this is something I'm showing you in biology, it's not it's really the structure of the universe. So it's still hard for me to believe this. When the universe began, the energy and matter density at this place compared to any other place was the same to one part in 10 to the 5. It's that homogeneous. That seems to me almost constant but it's not zero. It's one part in 10 to the 5. If you give enough time, which we have because the universe has a long time, which is 14 billion years. And if you give me a force which is divergent, which it is because the force 1 over r square is divergent which means it can act on infinitely long scales, then you can organize these very small fluctuations into large structures like the ones I showed you. So these are small structures that over long enough time with a long range of gravity get organized into big structures and now the universe doesn't have one part in 10 to 5 variations. There's enormous variations. That's why we're here in some strange way. The universe is not only simple geographically but also simple and the elements that were present then was hybrid and helium. That's it. It's hard to make just without a third body. As you well know, if you're a chemist just from molecular hydrogen without other species, it's hard to make that reaction proceed and helium is noble gas. So the outstanding question output is very simple. We do know that stars are very stable, the rest of it. But I want to know can I explain what is a fraction of manganese, we always know less to hydrogen. Can I explain the abundance of these elements in the universe quantitatively? Can I say manganese is more than for example, I will tell you right away that there will be more iron than any other elements of the medium Z things because there's something very special about nuclear chemists, reactions and iron. So I will say the grand quest for me is can we give a quantitative explanation for the periodic table? Can we explain the cosmic abundances of elements in the periodic table? And for this, we must study how stars live and particularly as we go over here, how stars die. And that's a grand quest. So if you say what am I doing because I'm a mechanist I'm trying to explain the periodic table for you in a quantitative fashion. Okay, so one of the things that I have a personal choice of it's very hard for me to work in an area for more than over five years. I consider this a bit of a failing because if you want to really accomplish something I think you should stick around for the rest of your life. It's very difficult for me to work more than five years. I have published a few papers and I understood I just can't read anything more than that. So about 2005 I finished the eight years of working in chemo reverse. I said okay what should I do next? And so I started thinking and one of the things that I absolutely think is important personally is I love teaching. I think I can tell you for sure I'm the best student in my class and I'm not a good teacher but I love teaching because that's how I get new ideas. So I always teach in fact I sometimes over to teach more and I also teach a subject probably three times and then to move on because this is the only way by which you challenge yourself to understand things. So I said okay I wanted I started teaching something else I started teaching high energy astrophysics and then I said okay this is going to probably look interesting but you know as it really you can always come up with ideas I never have any impression that I have a great idea. Many times ideas are very common they realize simultaneously by many people they tend to be a rather generic. The trick is what do which idea is feasible how do you execute can I do it timely it's all the practicalities and then also you want to choose an idea as a scientist something which is interesting but not obviously important because then it will be like doing one of these gazillion collaboration things where you remember some 5000 people doing something I never understand I would never be in a collaboration of that because even if I'm dead no one would know for a year. So I said okay I always like to understand which technology can I use to further my own science and I read widely so one of the things I noticed was that the pixel of CCD was really falling more slow and so I said okay maybe we can build a camera which would be the worst biggest camera at an affordable price which I thought would be about maybe 5 to 10 million dollars but it has to be like gigantic so my projection was by about 2008 the prices would fall so I started this project and I said okay I want to build a factory to discuss supernovae until then what's the song students postdocs faculty advice you go take pictures you come back and you analyze that compare and find these things sort of what I call a cordage industry that is fine but I wonder something where you just discuss supernovae in an industrial scale things would happen and for that you need a different approach which is a factory approach because you don't make cause by a small group anymore that you have to learn factory is more about scaling of operations so I always said it's a joke I know Japanese don't feel too bad but I would say if you come to Japan the first thing you have to do is you have to go to the Toyota factory in Nagoya I mean it's like fantastic after that if you have time maybe you can go to Kyoto but okay so the idea of this factory then was I said okay let's take this camera this telescope it's a special kind of telescope which has a large field of view okay so it is extremely good for making images of the sky so specialization and we'll take images of the sky run real-time pipelines discover stuff which are probably candidates then you need to confirm because if you have a false positive in this game it matters because if you're asking someone else to follow up your observations and they're going to use super telescope where the cost per night is 100,000 dollars and if they screw up which may be 5k it's a lot of money so this is for confirmation and ultimately you have to get a spectral chemical signature of that explosion which means spectroscopy so this sort of idea of changing observatories our telescope sounds pretty obvious simple and it is actually but it has not been done in astronomy until then okay so that's what let go of the program which I called as the Palomar Transit Factory and we started in 2009 and we had to phase three of that project called as wiki transit visibility so we have a rough idea now and the goal of this factory eventually is we now know the sources of where these elements come from in the periodic table okay there are various sources here where the big bang small stars, supernovae large stars and a very small number of elements like lithium or maybe beryllium is made by spellation, cosmic rays otherwise it's not a very important process okay so that's the overall theme it will take many years to complete this grand project but along the way a few other things we could do is to address some of the frontier areas of physics and that's the physics so one of the more interesting things is energetic particles so physicists have discovered this amazing cosmic rays that have energy of over 10 to 20 electron volts but it's not a big number unless you relate to it it's very hard to understand the fastest thrown ball happens to be a crooked ball you do half a mv square for that that's about 10 to 20 electron volts now of course you know that when these particles do come you don't see people getting knocked down by cosmic rays that's of course because a very little momentum corresponding to the energy but they are detected about one per year per square kilometer collecting area and here's the ice cube thing it's amazing here's the ice layer in the south pole there's a long long steel ropes which you have PMPs look at I feel touched so anyway so the origin of these is hardly debated in fact unless if there are any physicists I'll take a bad thing some physicists they want some exotic this model or that model obviously it's very simple it's made by some natural phenomena some star or maybe an explosion it's nothing no very deep I had a complete thing of physics that needed okay and then you have a gravitational waves which has now become I would say in all my life perhaps the biggest most amazing thing as I said I'm not even in water that is the discovery of gravitational waves something like this happened maybe 136 years ago I don't think that was a discovery what was that I don't think that was a discovery detection it was detection I'm not going to lie to that they would predict oh this is your physicist okay many people predict many things and they only tell you when they succeed so predictions will only have a value when you put real money down so if you want to make a prediction the only person who will listen to predictions is you say I predict this I'm predicting one year of my salary I'll listen to that otherwise predictions are cheap okay so I said how do I do this actually well I have a friend that is the director of the Institute for Theatrical Physics KITP I was explaining to him he said let me introduce to a friend and I had a lunch with this friend of last in the War for Santa Barbara it was supposed to be from 12 to 1 it lasted about 5 hours finished at 5pm next day what can I help you I said why is it simple give me one and a half million dollars a six month option period and then I will come up with a plan and a team and at the end of that if I'm not double the money and there's no plan you take your money back because I strongly believe that if you have a plan you've got to execute otherwise it's worthless and it has to be done in some fixed time anyway so yeah he sent me a cheque and this project so we went from this lunch the first time in 26 months which is I can tell it's quite good okay so we used a state of art but used system from Canada, France, Hawaii it's a mosaic of very high quality science grade CCDs we refurbished that whole cryostat and this is the camera and the moon has been photoshopped just to give a sense of this camera these sorts of projects people confuse the hardware is important and now people appreciate the software because now you know you can always buy a computer cheaply but the only reason you don't want to change your computer is all the software to reinstall so everyone appreciates it but I'll tell you there's something in more the third stage is what I call grayware if you want to make these projects happen you need some very smart people so I went around the world and they joined me in this exciting project either you have to pay money to join this, or you have to give your brains if I value it so I have all these young people so Nick Boyd finishes Ph.D in lucky imaging it's the same kind of very high frame rate imaging so this other thing is I like to work with young people and I believe if you have Ph.D is a license to solve a problem you are a license problem solver if you have a Ph.D you should be able to solve any problem now we are limited so maybe if you have Ph.D most of us can maybe solve in a strong way but in a strong way you should be able to solve most problems so I think it's always that young people say yeah you are in charge and that's it so there's no management meetings and so on we have this group at all so here's a project scientist which means he led the whole world with that effort when we had finished his Ph.D from Texas and he was a software lead which is the worst possible job in this project because that means you have to take all this software coming from various people and make sure they work and in this sort of project you don't set up and have interface document specification and so on that sort of stuff is really only important for a thousand people collaborations Josh Blue had finished his Ph.D and he wanted to work on machine learning and at that point I said well does this stuff really work or just some standard computer science BS that I keep hearing and he convinced me I was a little frankly I thought it was just a lot of words you hear and like AI deep learning all that stuff but I'm not saying I was mistaken it really worked amazingly and I run, of I can finish his Ph.D Tel Aviv and he's one of these he's in charge of all the robotics and the sequencing so the project took off and this is what we do it's actually very straightforward to tell you what we do we have we image this car with these large homemade detectors and over time we built up what's called the reference image which is basically you make this come many times you add it up in some very optimal way remove all the blemishes and we call that the reference we take an image now and you compare against that and it's called image difference and it's not matrix A minus B so if you have an eye in the shop you can see this slightly brighter here compared to that's a supernova but when you build that to this here you see slightly fuzzy here and then there's maybe then you know you can't have two supernovae next to each other and this fuzzy and this is what the human eye is very good at so after an hour of training you can start actually saying what's a good subtraction and what are artifacts and this is where machine learning so we have the first to apply machine learning and that's the way because in the past when people used to come and they're finding maybe one every few nights it's okay you can go through it at the time but our goal is to find maybe half a dozen per night which means the amount of false positives so it means you have to examine something like 600 to 1000 it's not feasible to keep this up okay so there are all sorts of data flows the usual sort of things so we have to build really robust pipelines because the data is obtained in Powell and Martin you're streaming through microwave links to Berkeley, to Caltech subtractions to happen sometimes are glitches you can't have your pipelines to be very tolerant and to continue that we generate this in real time pass it on to other telescopes so you have to then rank them in some order or priority so at this point then I realize the way out of this project is to completely automate this and so I said a statement to the group I said the best way to do astronomy is to get all these loops which annoyed some older astronomers not in the group they thought I was talking about them which was true actually okay so that's what happened and we really got on to this pipeline mainly because of all these young people okay another innovation we have to do is in order to spectroscopy which is with this the demand on this was so high these two we took over this we couldn't the other astronomers using it too so we developed a knowledge spectrometer here in order to spectroscopy in that case you get an image of the piece of sky where you want to go acquisition image first look and then you take this object and put it into a very fine slit and the precision of all these things is really enough of a modern telescope to do what sort of blindly pulling it that is just a drive start integration it doesn't have that these are very large structures to get on our second is to really control this structure at something like 15 microns not easy so we adopted and devised a new type of spectrograph it's called an IFU spectrograph so in this spectrograph this is the acquisition camera which is about 30 by 30 arc seconds if you are a physicist sorry we use arc seconds the Sumerian system not red gradients but it's big so anything that goes in 30 by 30 absence gets a spectrum automatically now that's big enough that you can slew the telescope and start integrating without human in the loop so for example here is Saturn in an acquisition thing so if you put it in that thing you would actually get a spectrum of Saturn every point of Saturn and these are small spectra and from that you can see the spectrum of Saturn's body and the rings that are actually very different anyway so with this we could now do robotic rapid spectroscopy so we now have these machines working with a new contextual analysis because we are getting a lot of events we don't know which few we want to pursue so one of the outstanding discoveries we had was we got a spectrum of an object it looked like a supernova it was very bright in some sense relative to its spectrum so a a collaborator in the team the team stretches all the way from Taiwan to the west coast which means we have all this time zone so when we are observing at night time it is morning in Sweden he realized that this is something odd and so we got a detailed image of this system I am not going to explain in detail but let me just say what is happening here is there is a supernova here there is another galaxy here and let's appear that you are the observer supernova is redshift 0.4 this intermediate galaxy or it is called dancing galaxy is redshift 0.2 they are so aligned they are so aligned that the rays that leave the supernova get bent and therefore you get a magnification of 56 so the whole thing is called the gravitational lens but in order to get this alignment which happens only maybe for one in a thousand supernovae but only when you have these sort of large case factories with informatics that you can start optimizing it and then we got this adaptive optical image from the observatory and later we also had images from the Hubble Space Telescope okay so I am not really going into all the details but I can assure you that this operation doesn't have people running around actually we don't say through the night at all it's literally you come in if you are somewhere in a different time zone you can see things are happening I don't say awake in morning morning I can say okay it's like saying okay boys and girls what have you done today let's take a look what's interesting of course the ultimate phase I wanted something where I come in and the paper is written up let me take some time okay so very quick summary for those of you like these sorts of engineering details so first one I think we are the first to apply machine learning on a massive scale and then we did some like spectral classification which really required us to decrease the latency of all these processes and these are very complex pipeline that are running in different parts of the system in fact parts of the world and these are some technical things I talked about the robotic spectroscopy and then we did a demonstration of needle in haystack which is go to an arbitrary piece of sky 100 square degrees and find an article of a gamma ray glass which we know has happened and the reason we did this exercise was for the LIGO project where LIGO finds something the localization is known to help people count apart so we demonstrated these are all technically very hard projects because the data flows are extremely large and there's not much time everything has to be done quickly and reliably okay so that was phase one and two and then I decided I was myself surprised and you know for me I was about to wrap up because it's a long time here but it became so successful so proposed the Zwicky Friends facility and that's a precursor to what's called the large synoptic survey telescope so this is a big US investment in ground-based astronomy so every 10 years astronomers in the US get together as they are now this year and they said okay what shall we do for the next 10 years and so in 10 years ago they said we're going to build the large synoptic survey telescope it's a billion dollar project it's going to be one of these very large projects that I think many people will be talking about you'll hear more and more talks about this okay so there are a whole bunch of innovations we did which I have gone through but a few of your friends there are also some new fun things that are coming up marshals and brokers decisions based on data and contextual data so if I convert this language of astronomy into data analysis you'll start realizing these are similar problems so every year a few of us computer scientists engineers astronomers they have meaning an ISU where people just talk about what they're doing we don't even understand exactly the details of their field but we understand the nature of what people are doing okay so astronomers are like mathematicians mathematicians would love if they write a paper and it has absolutely no utility at all this is the ultimate hallmark of a great mathematician unfortunately it's so useful that almost all these 10 years later this is fine so use for that but we also saw that Josh Bloom after the phase one of PTF he went and founded a company called Weisert IO and his slogan was solutions for real life problems and then in his brochure he showed that how his machine learning had to actually work through the mind and execute it and test it so it's not like a Netflix challenge where I can issue a solution but it takes a long time to know if my solution of what you like in Netflix is better than this one is simple you make a prediction if you wish and then next night we observe it in there your algorithm is no good six months later we know so we learn all these things in parallel we should challenge this and see whose algorithm is the highest performance what has come into GE two years ago when the stock was still high so I called up Josh and said you know remember your old advisor like 10% back and all that stuff he's still not done that okay so this now this has become unfortunately a large project or so precise what I don't like it has something like 40 people but fortunately we have no meetings because there's one thing I don't do in meetings anyway so NSF decided to allow us so much because we are a precursor to LSSK so they flew in about 11 million dollars and then the usual stuff you learn around the world give me money, give me brains there are all these institutions involved and this led to this facility which has just come on so I gave a talk this idea and the talk was the automated discovery of the universe at MIT in 2015 and there's a young man in the audience and he was so impressed about this automation and things that he decided he'd so he's now started a company that seeks to do software automation so I keep track of this I hope when they go public I'll call them up and say remember 10% and to do research I mean anyway so these are things you really can apply and how am I going to start up some industries which I like actually which shows a very healthy thing because my view is just because you run a PhD doesn't mean you're committed to academia for life it just sounds like a terrible thing it's a thing you do you're young you go explore you learn new things then you say what adventure can I seek so this camera we're working is a very large camera which is quite large here's a Japanese camera on Subaru in Monokeya behind 8 meter telescope and so on and here's LSSD which will come up in 20, 22 10 square degrees so we went deliberately what he called is wide and shallow so we're looking for brighter objects that's why the small telescope is perfectly fine so in some sense we want to do the easy stuff the low hanging fruit all the bright things so when these big telescopes come on they'll be forced to go to paint the objects which is a very different kind of science these are very large engineering projects making no mistake despite my unusual style of management these things work it requires dozens of engineers very smart young people to make these sorts of cameras this is the world's most compact camera it has a CCD sensor these are perfect CCDs no blemishes nothing 300 square centimeters pure silicon detectors and the work is extremely small you can't you can't miniaturize anymore which is not usually the requirement we have to rebuild the telescope which was built in the 50s I won't go in detail a lot of engineering issues here so here's our first slide tonight if the sky is clear you can go February so you should be able to see Orion in the west the big quadrilateral of Orion and then there's a belt and that's how big are cameras it's a big camera and you can zoom in picture you can see the Horset Navy I can see Orion a few other Navy velocities very much in detail a lot of data flows that happens here so data is collected I'm going to link to IPEC at Caltech real-time pipelines are kicking kicking, they start generating candidates machine ranking happens this is supposed to be an astronomer who says okay as I said we try not to have this to be some sort of emergency system the whole point is to program these and say this is the kind of event I'm looking for and routine that's fine otherwise send me a text message but it better be something unique so that's a lot of this lake where this goes here and if there's something where yeah it's a routine thing not routine it some program we're pursuing automatic this other telescope is involved to get a spectrum then the data flows to University of Washington they're a big partner in this project because they need level 1 data processing for LSSD which means they charge a real-time and that's why they signed up so then they package all these alerts so they're generating a few hundred thousand alerts when I say an alert I mean we take a picture now we compare that to a picture taken what we call the reference which is the old sky in the past if an object has moved meaning it's changed in either x, y or flux then it's an alert so we send these other packages out in x, y, there's some contextual information I'll explain and this is a very large data stream and this is what other astronomers are looking for so these alerts are sent out right now they flow to Antares in Tucson Alerze in San Diego Chile, Lassere in Edinburgh Elcio in Santa Barbara so these are what are called brokers so these are people who are these are institutions which have decided to be so if you wish, they are the retailers we just send it to these wholesalers and then if you as an astronomer let's say someone in Japan wants to look at these data streams to start finding things then you have to write all those filters and all this and they help you how to do that and then you provide you some fixed number of filters so these other packages are actually quite complex they consider data they consider oh sorry this is an alert management skip that oh these other packages have previous history of the sky they tell you what are the machine classification is it the star or galaxy these are of importance to the astronomers so there is enough information here at this point if you know python, if you know python library you should be able to then say I am now looking for this sort of phenomenon in the universe so we have created a literally virtual dynamic universe except since you are not observing but it is real because it is the real universe so these other packages can be sometimes zero if the weather is bad sometimes they can be very good if we are looking at a very dense field okay it is highly unlikely anyone at OIST really wants these alerts but there are all these places and that is exactly how it can access these alerts and there is now growing the body of astronomers mainly because this is the style of astronomy that LSD will be moving into so the astronomy is being replaced from astronomers we are successful a little bit by getting the astronomers out of the loop because now we are replacing astronomers really by people who know a lot of data management algorithms, computer science and are interested in the sky so here are some fun pictures you can have like supernovae we find so we are finding we probably have the capacity to have maybe a thousand supernova candidates but of which about a hundred is what we discover in the sense we actually take those candidates and we lose their cross-copy on a certain specific subset and so these are supernovae of different sorts that we are now finding so a hundred supernovae a month is not bad actually that is a major jump in this business we are doing okay so let me in view of time we are able to study and we can now see supernovae on like a month ago we are now reduce our data and say that we can actually see supernovae rise so we can actually see the birth of the star explosion in 20 minutes no detection 20 minutes later limb detection 40 minutes more detection exponentially rising all that sort of stuff so you know this is like really fun time there is one thing nice about astronomy you know the sky is so rich you are down on a machine like this you can write this paper so we are writing like nature papers to the world so ordinary paper is like one marble but nature is like three I think so that is what we prefer okay so let me end with double D jump rates which are exotic and important stars now some of you who have learned English classical way you know if you said someone what a degenerate person he was or she was you would say it is an itself which is true but in astronomy a degenerate star is like very precious and a double degenerate is like extraordinary okay so if you an astronomer tells you you are a double degenerate you should say what a compliment that is okay the reason is that as you well know the word degenerate in this context means stars are just supported by degeneracy pressure and that's what they call degenerates a double degenerate is a star with both members are supported by degenerate pressure there could be for example two neutron stars there could be two white walls or a neutron star black or a neutron star white dwarf and if I say it could be two black holes I know I get some people completely irritated because that is a very strong statement to say it's a degenerate star sorry it's a joke okay so why this degenerate is important because it's these degenerates that are really driving the forefront of astronomy so the LIGO detection was two degenerate stars coalescing two neutron star coalesce they produced so-called R process elements including gold, platinum and silver and the process of new black hole but if you take two white walls they could also do the same thing if you start with a compact amount of things so just in like in astronomy you know we have a radio astronomy which is low frequency astronomy because this may be 10 gigahertz and optical astronomy is high frequency because it's 10 to the 14 hertz so in the same way in gravitational radio astronomy there are different bands so the LIGO they look at roughly 10 hertz to over a kilohertz that's their passband okay and that's what you end up seeing but the Lisa mission which is a European mission which will come up in 2034 its passband is 1 millihertz to maybe a tenth of a 10 to the minus 4 hertz to maybe a millihertz or something like that, that's its passband and Lisa is extremely sensitive to white walls that are going around because white walls is got a larger size compared to neutron star and your detector must have you know longer wavelengths so you need a bigger detector so the Lisa mission is pretty amazing you know you'll have three stations in this equivalent for flying to 5 5 million kilometers apart in an earth trailing orbit and it will go it's really tuned to finding coalescence of supermassive black holes coalescence of white walls so we're getting ready for that that's sort of the next big frontier in gravitational radio astronomy so I spoke a lot on explosive astronomy I didn't say anything about near earth asteroids by the way we have become very good at finding near earth asteroids now and but I also didn't speak on what I call as non-transient astronomy but once you have this data which is so cadence you can start looking for periodicities you can look for small period so grandson Kevin Burge decided okay we'll do like a massive search for the whole database and look at GPUs GPUs are very good for this whole stuff so he developed the whole GPU based search for periodicities in uneven sample data that's a technical problem and he found a very interesting source you know and the part that I really like is when you can go from finding something to discover which means understanding what it is and so the same night Kaufflin we have this new facility which is kind of amazing NSF actually gave our group this telescope for 5 years for free it's a long story I like to think it's because of my charming guy but I think that's not true they gave it to us so we put on a new fast-moving camera on that so same night Kaufflin discovered that in fact this candidate is a glistening binary and this over a period is 6.9 minutes 15 not for 69 minutes but I think it's gone 9 times around in the time I've been speaking it's 2 white dwarfs it's one of these diagrams I don't have the time to explain it's eclipsing which is why the light disappears here and so it's 2 white dwarfs going on with each other in 6.9 minutes and one is hot, there's cold when the cold one comes in front of the hot one and let's say Millet is observer when the hot light is diminished and you know from the again classical application of Kepler's equations all that stuff you can compute the masses from spectroscopy and here we use the eclipse as a marker and then we went back into our old archives with the BTF data and found that object because now we knew what to look for and we found the period is different that's because it's been decaying and this is the line of GR so this is the period there's a period derivative which fits exactly what's given by GR so of course if I had done this 20 years ago it would have been a fantastic thing but now it's all very well known GR experiment with GR so let me end by saying concept for convergence so convergence is when there's progress in two somewhat unrelated areas so for example convergence is the ability to have integrated circuits that is a big deal think it really miniaturized many things okay the other one was lasers now that came later initial lab lasers with these very bulky things but then you were able to mix all these laser ever smaller things so once you can integrate lasers with ordinary electronics it led to a new field of photonics and without photonics people operate this institute or so on because you need so much optical fibers all that stuff is happening so that's an example of convergence all of a sudden it enables some new things so also there's a convergence happening in astronomy that's based on a certain brightness of stars and this is a bit for astronomers there's a major European mission called Gaia so they learn the Euro missions doing it simply well big projects that are happening were called massively multi spectroscopy which is not a spectrum at a time but 5,000 at a time then time domain which is what I talked about and then there's a whole what I'd say sort of algorithms numerical astrophysics machine learning, big data method all these are coming in so this is actually a very interesting time to go after this so if you ask me this is the thing I specialize is to actually see fields do it before maybe I'll make money and get out repeat the cycle but this as young I'll be looking at this field anyway so I organize a conference here for young people so if you go to my web page you can go and see it's a very different style of conference but there are some very interesting thoughts on what this next convergence would be in astronomy at least eight conversions so in the end I'm saying we have a precursor to this major national project from the United States probably in the world actually at this point we are scaled down so yes it will do more but we are here and today and we are writing those major papers today not for years from now so there are opportunities for institutions such as yours to get into this whole area of astroinformatics which is a very different style of astronomy than I began when I started grad student we would go to the telescope it was precious to get by and then I could also write a single of the papers too which was nice but anyway this is a new thing so let me end by saying big facilities coming on you want to do fun things so you can do it today and if you want to do plans 15 you can do it five years from now thanks very much