Benjamin Dean public lecture in the Morrison Planetarium on SN 2011fe, the nearest Type Ia supernova in decades, caught hours after explosion by the Palomar Transient Factory's machine-learning pipeline: what it revealed about the exploding white dwarf and its companion, and why Type Ia supernovae matter for cosmology.
His second Benjamin Dean lecture in the Morrison Planetarium (the first was in 2010), introduced by Ryan Wyatt, the planetarium's director, who also moderated the questions. The Academy's audio recording (1 hr 24 min, from its former iTunes U collection) is linked above; the slides’ title page misdates the lecture to September 9.
“You are made up of the stuff that was expelled from stars. You are nuclear waste. I mean that in a good way, by the way.” – Joshua Bloom
“Some people have invoked this magical, mystical stuff called dark energy, but because it's sort of the opposite of gravity, maybe we should call it levity.” – Joshua Bloom
“There's no Edwin Hubble here. He's been completely replaced by robots.” – Joshua Bloom
“The way in which astronomers usually go about their business is they have a whole bunch of ideas and then go about falsifying them one by one, and the only thing that's left standing is the one that people wind up believing.” – Joshua Bloom
“One of the wonderful things about nature is that she really, really likes to mess with us.” – Joshua Bloom
QUOCK: Welcome to the Dean Lecture Podcast. I'm Bing Quock, Assistant Director of the Morrison Planetarium at the California Academy of Sciences. The following lecture was recorded on September 10th, 2012. Here is Ryan Wyatt, Director of the Morrison Planetarium and Visualization Studio, to introduce our speaker.
WYATT: Tonight it's my pleasure to introduce Dr. Joshua Bloom from Berkeley, just across the bay, to talk to us a little bit about a really exciting event that took place last year, Supernova 2011fe, since it occurred last year. The topic is one of the — actually, we call it once-in-a-generation supernova, so I thought it would be great to have Dr. Bloom come and talk about this. After his undergraduate career at Harvard, he actually went across the ocean to Cambridge. So I hadn't really thought about it, but we have a nice little Oxbridge combo between this month and next month for our presenters. And then to Caltech for his PhD.
And he actually got into astronomy during his undergraduate career when he worked at Los Alamos National Lab and started working on gamma-ray bursts, which are a phenomenon that, over the decade after he started working on them, went through an increasing amount of data and actually went from a complete cosmic mystery to a well-resolved astronomical phenomenon. That's actually not, as I said, what he's talking about tonight. If you want to learn more about his views, you can actually read his book on the topic published by Princeton Press. But tonight he's going to be talking about supernova, in spite of that great work on gamma-ray bursts, which, I should note, actually garnered him the Pierce Prize from the American Astronomical Society for outstanding work in observational astronomy, and also a Sloan Research Fellowship. So tonight it's my pleasure to introduce Dr. Joshua Bloom to talk about Supernova 2011fe. Thank you.
BLOOM: Hi, everybody. I'm sure all of you have read Hamlet, right? Yes, good. How many of you knew that you could read Hamlet as an allegory for the Copernican revolution? It's true, yes. And in fact, in this first scene, we hear one of the guards looking out at this ominous new star that's appeared in the sky. He says, “Last night of all, when yond same star that's westward from the pole had made his course to illume that part of heaven where now it burns.” That's the beginning of a number of very interesting connections to a very important time in the history of man, where we were learning that the Earth was not indeed at the center of the universe.
And in fact, there were some of the cracks that had already started to show up in this idea over time, but some believe that Tycho's supernova in 1572 — at the time it was just called a nova, or a new star; now we know it's a very good star, it's a supernova — really was one of those nails in the coffin against the Ptolemaic model of the universe. The idea that the heavens were fixed was a pretty important tenet of the old guard. And the idea that a new star could show up and then actually move on the sky with all the other fixed stars was very, very troublesome. One of the reasons to show you this, of course, is that supernovae, through the history of mankind, have been unbelievably important, not just for poetry and plays, but for our understanding of mankind and our place in this vast universe.
So I'm happy to talk to you today about a very new supernova — Supernova, as it's called, 2011fe — that was discovered in August of last year, basically just a year ago. And this is more or less the anniversary to the date of when we submitted our first papers on this. So it's still very fresh in many of our memories. But I hope to connect this event back through history, looking at previous supernovae, and try to show you the importance of this event in context. So thank you very much for the introduction. And I'm very, very pleased that all of you made it out tonight. I hope we can enjoy the show.
Let me tell you about supernovae. Here is a picture of a galaxy. It's not such a different galaxy than our own. It has dust lanes here, just like in the Milky Way. And it's got a whole bunch of stars that make up the bright light of that galaxy. And somewhere, as you'll see as I play this movie, there's going to be a new thing that appears right there. And look at this on this side, but at the same time, look at what's happening over here. What is it that we see as this star brightens and then fades away? We'll see this pop up again. This is what astronomers call a light curve. This is basically just brightness as a function of time. And I haven't given you the time scale here, but this is a few weeks. And what you see here on the bottom is what we call a spectrum, and this is basically just brightness as a function of wavelength.
And you can see that the peak of where this gets bright, when this actually hits its peak brightness, is somewhere over here in the green. But you notice that we have a whole bunch of interesting features, as we'll talk about later on, in this spectrum. When astronomers follow up supernovae, what we do is we look at light curves. And we also try to take spectra as a function of time to try to understand what it is that we're seeing inside of the light that came from that event. Another thing that's pretty important to point out is that when this reaches its peak brightness, although it might be hard to see, most of the light, if you add it up — because it's coming from just one little place on this image — is comparable to the whole light that comes from that galaxy.
That's a nearby galaxy, but we also have observations of supernovae in very, very distant galaxies. Here's an artist's conception of essentially an anonymous galaxy, say, 10 million or 10 billion light-years away. We'll see a supernova that goes off in it, and we'll zoom out and start looking at real data. It'd be nice if we had some music or something. Oh, there we go. Okay, good. I won't pull these gimmicks throughout the whole lecture, I promise — just to wake you up a little bit. Think about how many galaxies are in this big picture that we have — okay, that's enough. Think about how many galaxies are in that big picture. There's something like 400 billion galaxies in the observable universe, and a supernova occurs in one of them perhaps one every hundred years. So when we can look out at the vast reaches of the universe, even though supernovae themselves are pretty rare, we can actually find them on Earth with appreciable frequency.
But what's very rare is the discovery of supernovae within our own galaxy, and that's only happened over the course of the last millennium about three or maybe four times. The first recorded supernova was very well recorded by Chinese astronomers in July — basically went off on July 5th, 1054 — and there's very good readings of how bright it got, where it is in the sky. And at the same time, the Anasazi in Chaco Canyon near Albuquerque were also observing that. And this is the so-called famous pictograph, which is thought to represent the observation of a new star which occurred, and you might be able to make it out right here — it's a little bit in the shadows. Here's the new star, and here's the crescent moon, and this is thought to be the representation of the Pleiades, and that's actually in the right place on the sky. So if you're trying to send a message to somebody and you're hoping that your email lasts a long time, it probably won't last as long as pictographs, for a thousand years. So if you really care about something and you want to message somebody in the future, you probably want to carve it in stone.
Another supernova which occurred not in our own galaxy, but in a satellite galaxy of the Milky Way. The last one visible by human eyes occurred on February 24th, 1987, in the so-called Large Magellanic Cloud. It's SN 1987A. And you notice that we've got these characters after the year in which these are discovered. A — capital A — means it was the first supernova discovered in that year, and then we go all the way up to capital Z, and then we go back down to little a, and then we add another a, and you just keep cycling through this. So that's the naming convention, for better, for worse.
And it's probably easy to see — the before and the after picture — that indeed there's something which is quite new and quite bright. And in fact, the exciting thing of this is that we actually had an image of the thing that would eventually wind up exploding before it exploded — and before, obviously, we knew that it was going to explode. So that was very, very important. 1987A is considered essentially the most modern nearby supernova, and that's only something like 170,000 light-years away, which for astronomers is really in our backyard. It's been observed from the time it exploded until today, but unfortunately it was only visible for a few weeks to the naked eye, and it was really only visible in the Southern Hemisphere. So all the Northern Hemisphere telescopes really missed out on this show.
I have a picture here of the location of different supernovae within our own galaxy. As you'll be able to see — they'll bring this up on the screen. Aha. So this is if we're in the solar system and looking out at the galaxy, and all of these different orange dots wind up depicting where nearby supernovae have occurred through the history of our own galaxy. We can see the remnants of supernovae as we now zoom out of the solar system. And we'll now get to see a 3D picture of where these supernovae have occurred.
So here we are flying above the plane. And you can see each one of these orange dots is really strewn all around us. But most of the supernovae that we know about that have occurred within our own galaxy — not the ones that happened in the last millennia, but the ones that happened, say, over the last million years or 10 million years — they're reasonably well distributed around where stars are. And that's an important part of the story of supernovae and its connection to us. What I'll do after this finishes — what I'll show you is what happens when we zoom into one of these so-called nebulae, and it's a fairly remarkable picture. I'll zoom into that now. This works? Okay, they're bringing me back to Earth. Are we going back? Well, let's just admire this for a little bit.
By the way, the representation of the Milky Way here is actually real, but the locations of each one of these things has been diagrammed fairly well in three dimensions. So here we are zooming into the so-called Crab Nebula. This is the remnant of that Chinese supernova that I told you about in 1054, so we know the date in which it happened. We can study that nebula very well in detail now, and you see these very pretty colors — I'll talk about that in just a little bit — and it allows us to study really in great detail what was strewn out from this supernova. What was part of this star, as it turns out, that wound up exploding? There's other types of supernovae. This is one sort of type, and this is one sort of remnant of a supernova. There's other types as well.
If we could go back to the other one, I think we're going to now zoom into the remnant of Brahe's supernova. And are we waiting for it? Ah, here it comes. Now let's get the galaxy out of the way. There we go, okay. What a pretty picture this is. This is actually a composite image from the Hubble Space Telescope, the Chandra Space Telescope, and other infrared and UV imagers. And you see this one's a lot more round than the previous one, but it's pretty and colorful nonetheless. The size of this, by the way, on the sky is not nearly as it is to you right now — this is obviously just for demonstration purposes. This appears to be something about the size of the full moon on the sky. So these remnants can be fairly large on the sky. Okay, so I think — are you taking us back to Earth? Are you taking me back to my talk? One or the other, we have to get there. Ah, okay, good, there we are.
Right, so that's what you just saw in some detail. And you can notice the shapes of these things are quite different. And in fact, blue is where the hotter gas is. And here, you can see the hotter gas appears to be on the outside. Here you see that the hotter gas appears to be on the inside. What's making that light? Well, it's actually just elements. It's the fluorescence of elements. It's the heating of those elements. It's all the different processes that give rise to colors in our own universe, in our local lives. But what are the important things with supernovae, when you start thinking about those colors, is where those elements actually came from.
This is a depiction of the periodic table, and this is a bit of a simplified view, but it's pretty accurate. So here, in the top left, we have hydrogen, and far right, we have helium, and lithium. Those are really the only three elements that are made in the Big Bang. So at the time when the universe is very young, before it's had chance to expand and cool, this is the elements that are made. And those are the most prevalent and abundant elements in the universe today by far. And then after that, what you wind up seeing in orange are some of the elements that are actually made during the normal processes of stars, during the fusion processes that power stars, makes them hot, and keeps them supported against the crush of their own gravity. These are the elements that are made. These are the intermediate ones. But everything in red that you see is created during a supernova.
So during the explosions — in the few seconds during the implosion of a star and then the explosion — many of these elements are created explosively. And some of them are radioactive, so they don't last for that long, but some of them are actually stable. And they wind up making their way into nebulae and gas clouds around where the supernova occurred. And what's important about that is that some of those gas clouds, if they have the right conditions — temperature, density, pressure — will wind up condensing and form proto-solar systems and proto-stars. And around those proto-stars, you get proto-planets. And in some cases, you wind up getting proto-Earths. There were supernovae in our neck of the woods around the proto-solar system. They wind up enriching with all of these elements that you see in front of you into that proto-star and into those planets and the planetesimals and into Earth and into you.
One of the amazing things that connects you to supernovae, not just because you're sitting here and you paid your entrance fee, is that you are made up of the stuff that was expelled from stars. You are nuclear waste. I mean that in a good way, by the way. And Shakespeare, of course — we have to keep coming back to him — said this very well. He said, “What a piece of work is a man! How noble in reason, how infinite in faculty and apprehension, how like a god, the beauty of the world, the paragon of animals, and yet, to me, what is this quintessence of dust?” Very few people say things better than Shakespeare, but I think Carl Sagan had a winner here. He said, “We are all star stuff.”
Let's talk about different types of supernovae. There's supernova of so-called Type II, and a prototype of this type of supernova — the remnant, at least — is the one that we talked about from 1054, that leaves behind a rapidly spinning neutron star, a very, very heavy but very, very compact object, and this leads to so-called pulsars, for those that are familiar with the phenomenon. And it is the most common type of supernova in the universe, and when we look and take a spectrum of this light that we see here, well, we see neon and oxygen, but most of all, we wind up seeing hydrogen and helium. Remember, these two very common elements in the universe.
And in fact, most stars are made up primarily of hydrogen and helium. So it's natural that if you have a star which explodes — and we think these aren't just all stars that explode, there's a special kind of star. These are stars that start off about eight times the mass of our sun. And because they're very heavy and they get very hot in the center, they go through a different life cycle than that of our sun. They essentially wind up running out of pressure support in the very end stages of their life; they wind up imploding and then exploding.
The other type of reasonably common supernovae — the second most common supernovae — is Type Ia supernovae, which is going to become the central part of the talk for the rest of the time now. Unlike Type II supernovae, there's no central remnant which is left behind. There is no hydrogen or helium in the spectrum, which is very interesting, because if hydrogen and helium are the most prevalent elements in the universe, why is it that we don't see any of it in the spectrum that we take of this remnant? And in addition to the lack of those two elements, we also wind up seeing something that we don't see in Type II supernovae, namely the existence of silicon in the spectrum. Over the years, people have obviously been curious what makes these very two different types of supernovae, and people have more or less come to the belief that it's a thermonuclear explosion of some sort of compact object. It's a star, but it maybe is a star that's lost most of its hydrogen or helium — in fact, if not all of its hydrogen or helium — before it decides to blow up. So it's a very different beast than these Type II supernovae.
We can recognize, now not in our galaxy but in nearby galaxies and ones that are very far away, when we take a set of images as a function of time — this is what's called a light curve — we could recognize very characteristic light curves to tell us whether we're looking at different types of supernovae. So spectra help a lot, but images actually help us a great deal. And we can take images with different filters. You can take it with a blue filter, a red filter, maybe even a narrow-band, say, H-alpha filter. And you see these characteristic light curves.
Here's another very nearby supernova. Let's see — oh, there it is. Okay. And you can see its brightness as a function of time here, taken with a blue filter and a red filter. This was in a reasonably nearby galaxy, and you can see also that this thing gets almost as bright as the core of this galaxy here. And you can see some characteristic bumps here in the red filter, and you see it gets bright and then winds up declining. You can also see that the time scale over which it gets bright and then falls away is about two weeks from the time it goes off to the time it reaches the peak, then it falls back down, and then it's visible for months and months afterwards.
This supernova was considered very bright in terms of how bright it was on the sky for us. It was only about 300 times fainter than what the human eye can see, which is, again, not nearly as good as what Tycho and then Kepler got to see about 30 years later, and it's not nearly as bright as the supernova in the Large Magellanic Cloud, but reasonably bright, and essentially the best-studied Type Ia supernova in modern times. So this was a very important supernova, SN 1998bu.
Now, what's exciting about Type Ia supernovae — and this gets into some of the broader implications of why we actually like to study them — is that they all look basically like this light curve that you see here, with a couple of differences. Let me show you those differences. This is a plot again of light curves, and what you have is time since the peak brightness. So it's a little cut off here, but this is zero, and this is minus 20 days. And this is brightness in astronomer units. And what you see is, in different colors, different supernovae of Type Ia.
And you notice that they don't all reach exactly the same brightness. And this is using a correction for distance. So you're basically putting all these supernovae at a distance that are thought to be about the same. And you notice that some of these are fainter and some of these are brighter. And this is a scatter of a factor of maybe three or so in absolute brightness. And you say, well, that's not very good if we want to now actually make use of these things for measuring distance, as I'll tell you about in a second. But what's exciting is that if you notice, the ones that are rising quicker and fall quicker appear to be dimmer than the ones that take longer to rise and take longer to fall.
So if you create essentially one parameter that corrects these light curves based on how fast it's rising and how fast it's falling, you end up getting a so-called calibrated Type Ia supernova. Which means that if we're trying to understand the brightness of a supernova as it is, if we don't know the distance to it, we can actually just look at how fast it rises and how fast it falls and make some guess at how bright it is intrinsically. So that'd be the equivalent, for instance, of looking at a light bulb and noticing that it's a slightly different color than the light bulb next to it. And what it's really telling you is that perhaps that color is telling you that one is 98 watts, and the other one is 102 watts. And so if you wanted to make them all the same brightness, effectively calibrate them, you could use color as one parameter that allows you to do this calibration.
What's exciting about this means I can observe a distant supernova and, using the one-over-r-squared law — the idea that as you go farther and farther away, a source appears on the sky fainter and fainter, and that just falls off like one over the square of the distance from you — it means that if I make a measurement of a Type Ia supernova, I can make a measurement of its distance. Now, distance is a very, very hard thing to do for astronomers. We can measure very precisely where a star is on the sky with respect to other stars, but how do we know how far away something is? How do we know how far away the center of the galaxy is? Or how do we know how far away the nearest galaxy is? This is actually a very hard endeavor. And this is something that astronomers have been perfecting the measurement of distance in many different ways over 100 years.
But until Type Ia supernova came around, and until we were able to make this correction, which really didn't happen until about 1992, when someone realized that this was even possible, we couldn't have a precision way of making a measurement of distance of a supernova. And that, coupled with the idea that supernovae can be seen to the very farthest reaches of the observable universe, means that now we have a precision way of making measurements of distance throughout the observable universe using Type Ia supernovae. And because of that, supernovae serve as the primary tool for so-called cosmography, or measuring and mapping the relationship between the distance of galaxies from each other and the apparent expansion rate of the universe.
This is a little toy cartoon to tell you about the expansion of the universe, which I'm sure many of you have heard about. The universe has been expanding ever since the Big Bang. And one of the interesting questions we have is, what's the history of that expansion? Is it slowing up and how fast is it slowing up? So here's a measurement of distance between galaxy green and galaxy red. And then you're going to make a measurement of a distance between a farther away galaxy from red to blue. And these are about a factor of two distance from each other. And what happens in this expanding and infinite universe — so it ends here on the screen, but think of this actually going out to infinity — is that no matter what galaxy you're sitting on, if there is some expansion, what you wind up seeing over time is that this distance gets larger and larger.
And here's the expansion now over 100 million years, and now these galaxies have moved away by a factor of two from each other. So basically the universe has become less dense by a factor of two cubed, so one-eighth. And what you notice is that galaxies that were farther away to start off with obviously have to be farther away at later times. And what we can see is that there's a relationship between the distance and how far away they are in a fixed amount of time. Or if you want to think about it, there's a relationship between the apparent velocity of these galaxies away from us and their actual distance from us.
Does anyone know who discovered this relationship? Hubble, yes. And here's a famous picture of Hubble looking through the 48-inch telescope at Palomar. And by the way, you can tell this is a PR image. How is that? Look at the shadow here. They're just blasting the light in his eye, and somehow he's looking out at the dark night sky — yeah, I don't think so. And it also looks like he's just too relaxed having his little pipe here, although there's very few pictures of Hubble, apparently, without a pipe.
Now, Hubble discovered in 1929 this relationship between the distance that a galaxy appears to be from us and its apparent velocity away from us. And this happens naturally in the context of explosions. The things that are farthest from the center have to be moving the fastest. And now we're not talking about an infinite universe, we're talking about one explosion. That analogy really holds here. The farther away something is, if everything is expanding out, the faster it has to be appearing to move away from you. And everybody feels the same way. Many people have told us that the universe is expanding. I put a lot of credence in what Chipotle Grill said, that the universe is expanding. Although those that are holding stock in Chipotle probably don't feel like the universe is expanding, but collapsing. That's another talk, I guess.
What would it mean to be expanding? Well, if we had the ability to look back in time and there was a constant coasting — that galaxies farther and farther away from us appear to basically be moving at faster and faster velocities away from us — then we would wind up getting this white curve here. This is the apparent velocity moving away from us as a function of distance. But all these other little colored curves here is a reasonable universe where — and by the way, I should say that this relationship, velocity is equal to H0 D; H0 is called the Hubble's constant. What we would expect is that because the universe has stuff inside, it has matter. Matter is attracted to other matter, just through gravitational interaction. You would expect that in the past, we may have actually been moving faster away from everybody because we didn't have that chance to coast.
Think about it this way. If I throw up a piece of material, at early times it winds up moving fast, but eventually as it winds up really feeling the force of gravity and it starts decelerating, then it falls back down to me. If I want to think about the early universe, the way to think about the Big Bang is that everything was moving fast away from everything else. And today, well, gravity's had some time to interact so that galaxies are pulling other galaxies in. And so we should be moving slower, right? We should be expanding less fast. This is called deceleration. And these are the kinds of curves that people expect to use and find when they're studying the very distant universe, because it's very natural to think in a universe that just has gravity, it's constantly decelerating.
Now the measurements of Type Ia supernovae showed something very different than this. This is a bit of a meaty slide, so let's ignore most of it here. Just to show you that the curves that I showed you before are these red curves here. And the line connecting the accelerating universe and the decelerating universe is that white line that I had before. This is actually putting it in astronomers’ units. This is the apparent motion, apparent velocity away from us in our units. And this is some measurement of brightness over here. What you notice is the data is not very good. These are actual data points of actual supernovae. Again, because we can make a measurement of the apparent brightness of a given supernova, and actually we can measure the speed at which it appears to be moving away from us. And you notice, though, that there's this preponderance of points that appear just above this interface here. And this one up here is even worrisome.
It's as if when we look farther back in time — and we do have a time machine. Astronomers are great in that way. Because it takes a while for light to get across the universe, when we look back at a galaxy that's very far away, we're not just looking at something which is far from us. We're also looking at that galaxy as it was in an earlier time in the universe. So when we look back at galaxies, or we look back at supernovae, what we wind up seeing are these things that are obviously much fainter because they're much farther away. But we're also looking at the universe when it was younger than it is today. And so we get to look back in time as astronomers all the time.
One of the remarkable things here is that the supernovae were telling us — and it's a bit subtle, but they certainly seem to be unambiguous in some ways when people do the statistics — that the universe appeared to be moving a little bit slower in the past. So if the expansion was slower in the past, what does that mean for what it's doing today relative to the past? It means it's accelerating. The universe is not decelerating, it is accelerating, and it's taking off. This is not what's supposed to happen in a normal universe that just has gravity involved.
Many of you know, of course, that this discovery of the accelerating universe was a great triumph for physicists and astronomers, and it's what won some of my colleagues the Nobel Prize last year. Saul Perlmutter is a physics professor at Cal. Adam Riess did his Nobel Prize-winning work while he was at Cal as a postdoc, and Brian Schmidt is Adam's collaborator. They won the Nobel Prize in 2011 for the discovery of the accelerating expansion of the universe through the observations of distant supernovae. And it's not just any supernovae, it's Type Ia supernovae. So you can imagine they and the astronomy world was very pleased and very proud. But one of the things that Saul was most proud of is that he got a special parking sticker for Nobel laureates. He has a bigger smile here than he did when he got his check, I think. Those of you that walk around Cal will notice these NL parking spaces. You have to win a Nobel Prize to get one of those — to be a Nobel laureate. So he was very happy. So that's great.
You think we're done, but here's some of the interesting news: we still do not know why the universe is accelerating. Some people have invoked this magical, mystical stuff called dark energy, but because it's sort of the opposite of gravity, maybe we should call it levity. I don't know. And perhaps even more remarkable is we still don't know what makes Type Ia supernovae. I said, oh, it's some thermonuclear explosion of some compact object. And theorists have developed some theories, been able to simulate these things on supercomputers, but we really haven't been able to get our teeth into what actually makes a Type Ia supernova. And that's kind of worrisome in some ways, because we're hanging our hat on our understanding of the accelerating expansion of the universe on objects that we really don't understand. So we have to really sharpen our cosmographic tools.
So this is where some of the new work comes in. And I've been involved since 2008 in the so-called Palomar Transient Factory, which, by the way, is making use of a telescope I showed you before. There's no Edwin Hubble here. He's been completely replaced by robots.
And we and our collaborators down at Caltech and at Lawrence Berkeley National Lab and other places have been able to roboticize this telescope and fill its focal plane with, now, not photographic plates, as Edwin Hubble used, but CCDs — essentially glorified versions of the types of detectors that you have in your iPhones. And what we have here is a 92-megapixel camera that has exquisite resolution on the sky, and because it's a modern detector, can take images that are much deeper, that allows us to see much fainter objects than Edwin Hubble could have with the same telescope. It's a pretty small telescope by modern standards. It's only about a meter in diameter, but it's been very powerful, and you can see the depiction of the full moon here on the 11 CCDs that we have. So we get a lot of information about a lot of different stars and galaxies.
And I don't want to bore you with data flow. I'm sure you see enough of that in your daily lives. But one of the things I wanted to point out about this is that the data are taken on Palomar Mountain, which is near Temecula in Southern California, and while the data is being read out, it's being streamed through a high-speed line back to Lawrence Berkeley Labs, where we're engaged in a real-time processing pipeline, so that by the time the next image is taken down in Southern California — and we're talking about gigabytes of imaging that's transferred every night — we have already reduced that data, and we've looked at it. We've started trying to figure out what's in those images.
Because we have such a large grasp of the sky, we've been able to really tile — this is a bit of an older plot, but we've been able to tile almost the entire sky, the northern sky, over the last several years. And we've got something like 1,300 supernovae that we don't just have pictures as a function of time, or light curves — we actually have spectra to confirm that they're real. Something like 100,000 galactic transients. And we just keep on taking more and more data every night. When you take that much data, there's a significant problem here. The data's a pretty high quality, but it's not great. And what we have to do is sift through that data to find interesting objects. And let me give you just a little indication of the drill-down that we have to do for this needle-in-the-haystack work.
The Palomar Transient Factory, or PTF, produces 1.5 million candidates a night. These are candidates of things that possibly are changing or are new over a large swath of the sky. Most of those turn out not to be real. We have image artifacts and misalignments and saturation from bad stars. A whole bunch of things can go wrong, and they do go wrong. But of those 1.5 million, maybe 50,000 of those are real in the sense that they're actually asteroids. Maybe a few thousand of those are variable stars. A hundred of those are some supernova that's occurred at some point in the last couple of months. And maybe one or two of those are actually new, young, and perhaps rare types of events like supernovae. And those are the ones that we're really trying to get at as part of our collaboration. So we go from 1.5 million to one. That's a pretty big needle-in-the-haystack problem.
And so one of the roles that I've played in this collaboration is to try to help us sift through all that data more or less in real time. And the obvious way to do this is you just get a million undergrads just to look at this data. Turns out that's not very feasible. We actually have tried to use crowdsourcing to help us a bit with this. But what's turning out to emerge is an automated discovery routine that allows us to look at these images. And this is actually real data here. We have a picture of what the sky looked like, say, last year. And this is what we call the reference image. And then we have a new part of the sky, the same slice right here. And we take more or less a subtraction. We take this one, and we subtract off this one, and we're left with this. And this is a negative, by the way. So more white is actually brighter. Or no, that's not a negative. That's positive. And here you can see this really nice new object.
Okay, so that should be pretty easy. You just train machines to just find new things. Here's another one where you have a little galaxy here, and there's another picture of that same galaxy, and then when you subtract the two off, you get a nice little new object right there. But you also get all this other stuff. And it turns out that other stuff, these bad subtractions, wind up dominating this 1.5 million number. And so we had to basically build some sort of machine intelligence that could look at these images as if they were very well-trained undergraduates, and make statements about whether this is real or not real. And it's kind of hocus-pocus in some ways, because we're asking machines to look at lots of images and very rapidly, more or less, make statements in the ways that humans would make statements. And this is a whole realm of machine intelligence called machine learning.
And it was really the subject of my Dean lecture that I gave in 2010, when I was — for those that were here — telling you about this ramp-up that we were doing to be able to sift through this needle in the haystack. And it may seem a little bit strange to be thinking about machines basically acting like people, but I'll remind you, in this 100th anniversary of Alan Turing's birth, that his most famous paper from 1950 basically envisioned machines — digital machines — as more or less emulating human intelligence. So this idea of machine learning, not just on images, but now we can do it on text, and now we can do it on massive amounts of data, has been around for quite a long time.
So where does this supernova, the title of this talk, come in? Well, that happened in August of last year. Here is our image a day before. Here is a new object that appeared right here. And here is that object getting brighter as time goes on. It was a discovery of a new object in the Pinwheel Galaxy, so-called M101, in a very, very nearby galaxy, only 21 million light-years away. And machines — the machine-learning intelligence that we built — basically promoted this object to the top of the list for us to look at. Now, this is a black-and-white image. I think it would be reasonable to take a look at a nice all-dome color image of that beautiful galaxy. It is one of the most beautiful and dramatic galaxies that we know about, well studied by amateurs and professionals alike.
And there is Supernova 2011fe, as it wound up being called. We had a PTF name for it for a while, and then it wound up getting designated by the IAU, the International Astronomy Union, with its official designate of 2011fe. And this is a composite image of multiple filters. It's as if we're basically looking at the Milky Way face-on. And it's obviously difficult — if I was sitting, say, right here, and I was a great observer looking out at the whole galaxy, I might even miss 2011fe if there was enough dust between me and that object, or certainly if I was on this other side of the galaxy looking through the center, I wouldn't even know that it exists.
And in fact, there should be a supernova in our own galaxy every hundred years, and yet we haven't seen one since 1602 with our own eyes. And that's because they're popping off all over, but we really only get to see them because of the effects of dust in the local part of our galaxy. But now we're looking face-on to this galaxy, so we don't have any of these dust problems. And here is this bright new star, which is appearing and getting brighter and brighter.
Okay, let me transition back to the talk.
That's a bit of a zoom-in on the composite that I just showed you. What's remarkable about this is because we had machines that were sifting through this data more or less in real time, we were able to, within a few hours of its discovery, get a spectrum as the Earth was rotating around in the Canary Islands, because by the time we had found it, it had already gotten to be daytime here. But it was then nighttime in Europe and throughout Asia. And what's remarkable is that when we actually do the math and try to figure out when our first observation was from the time of the original explosion, it was only 11 hours after that explosion, plus or minus 20 minutes or so. And that's the earliest observation of any type of supernova by at least a day or two.
I'm giving a bunch of superlatives here — let me just go through it. Nearest Type Ia supernova in more than three decades, and the nearest Type Ia supernova in the digital era, when we were able to actually capture the evolution of the light curve as a function of time with digital imagers and not photographic plates. And it is the fifth-brightest supernova to appear on the sky in 100 years. So three decades, that's sort of a generation — that allows us to say supernova generation — there's the talk.
This obviously got a lot of play in the press. My colleague Peter Nugent at LBL was on PBS NewsHour talking about where you could find this in the sky. I was very happy that I was quoted in the Huffington Post, calling this a “supernova of a generation.” I'm more proud when I get in the Huffington Post than in the Drudge Report. The National Science Foundation was very excited about this because their investment in machine learning basically paid off. We've been able to sift through the needle in the haystack and find these candidates. And the Cal Academy of the East, the Hayden Planetarium, had a show on this as well. They called it “supernova of a lifetime.” I think that's maybe a little bit much.
Because it was so nearby, and amateurs have been studying this galaxy for years, everyone really knew what to do. And the amateurs got on board very quickly, within about 24 hours after the announcement of the discovery, and took multiple observations. These are thousands of data points produced by amateur astronomers submitting their data to the AAVSO, the central clearinghouse for astronomers to publish their data on supernovae and variable stars. And it really was in our backyard. You can see, for those that actually are amateur astronomers, just how bright this is — it got to a little brighter than 10th magnitude. And our eyes in a dark sky can see to about 7th magnitude. So we're within maybe a factor of about 5 to 10 fainter than what the naked eye could see. So with a good set of binoculars around the time of peak brightness, you could have gone out, and you could have seen this.
Now, if you have a reasonably sized telescope and you get very enthusiastic about this event, you go home and at 9 o'clock 16 seconds and you're standing in San Francisco, this is the location of where the supernova actually is. And it might be a little bit hard to make out. Here are the tree line. And it also may be a little bit hard to make out here, but here's the constellation Ursa Major. We like this at Cal because it's a bear, blah, blah, blah.
It's actually still reasonably bright. This is the observations from the AAVSO now proceeding over the course of an entire year, and this is, if you do the math, about 365 days now. These are the observations as it was basically last night. So it's about 18th magnitude for those that know about. Again, in a dark site in Marin or, say, Sonoma County, you might be able to catch this one with one of your amateur scopes. It's a very beautiful movie here of the supernova getting bright and then fading away, and you can see that it's basically getting too faint for this telescope, which I think is several tens of centimeter telescope.
Okay, so it's pretty. And I got on the Huffington Post. Okay, that's a big deal. What does it really mean for scientists? And why is it that you're going to go home and say, “Wow, this wasn't just a bunch of pretty pictures”? Why is this something that's actually pretty important? Well, we wrote a lot of papers, some of them in Science and some in Nature. So somebody thought this was interesting. The real question that we're asking now is: what makes a Type Ia supernova? That second big, huge, lingering question that we have — what is it that makes a Type Ia supernova? We don't know.
What I didn't tell you at the beginning of the lecture is that we have a huge slew of possibilities. I won't go through all the physics and all the details of these different possibilities, but there's really two dominant ideas. One is that you get so-called double-degenerate channel. That's a DD channel. And that's when you have two white dwarfs, which more or less wind up spiraling in and smashing into each other. And the lights come on at the end of this dance. Basically, one of these will wind up exploding as these things wind up merging. The combined mass is too large to withstand the crush of gravity. You get an implosion and then a very rapid explosion.
The other possibility is a single-degenerate model, where mass flows onto a white dwarf from a star. And by the way, a white dwarf — I don't think I mentioned this — is basically the similar mass to the sun, maybe 40% the mass of the sun. But it occupies a volume about that of Earth. So the densities are something like a few billion times the density of water. It's as if you took Mount Tam and you crush it all up into a one-meter cube box. That's just to give you a little bit of local flavor for what these things might feel like. Very dense, very compact objects.
And if they are in a binary star, then they may actually get to suck on matter, basically from this star, which is very large and spitting off some of its mass. It would fall back down onto this white dwarf. And if the conditions were just right, then the white dwarf would build up its mass as a function of time. And eventually it would hit this so-called Chandrasekhar mass, this magical number where we don't believe that white dwarfs can support themselves against the crush of gravity, and they wind up imploding. So there's something — either another white dwarf, or something else — that's dumping mass, either essentially very quickly or over a long period of time, that gives us the ability to blow up one of these white dwarfs.
Now, the nice thing is, in many of these single-degenerate channels, these stars are not ordinary stars, these companions. They're actually stars at the end stages of their life. They're so-called red giants. And whether the red giant is close or whether it's far from the white dwarf gives us different types of binary parameters. But the point is that red giants are red. They're very big. And they're not that cold. They're only a few thousand degrees Kelvin. So they're colder than the sun, but they're not that cold. So they emit a lot of light. And because they're so big, they actually radiate a lot more light than the sun — they can be thousands or tens of thousands of times brighter than the sun, even though they're red and they're essentially running out of fuel. And so it's really towards the end stages of their own lives.
What's so remarkable about 2011fe — and I'll get a little bit back into the observations at early times — but what's so remarkable is because it was so close by, and it's such a well-studied galaxy, we actually had images of the location of where that supernova would occur before the supernova occurred. So here's again a nice picture of the Pinwheel Galaxy. Here's a zoom-in of the location of where the supernova would go off. And these are individual pixels from the Hubble Space Telescope, from a stacked image that we created, where somebody wound up looking at this part of the sky just by happenstance and allowed us to take an image of where the supernova would go off. This is essentially the best-bet location of where this event occurred. And this is a very large distance in those units away. And the nearest stars that we were able to detect are so far away that we can very confidently rule those out as the companion to this white dwarf. Now, I should say that white dwarfs themselves are very faint, but if it had a companion, we would have been able to see it.
This is a bit of a busy slide too, but I'll just show you for those that are familiar. Yeah, sorry about that. This is data — I like it. Here's 3,000 degrees Kelvin, and astronomers are really weird here because we go from low to high, so right to left. And here we go from large number to small number. There's historical reasons for both of these things. You can blame the Babylonians, essentially, for this axis, and you can blame people from last century for this axis. But this is basically an important diagram, because it tells us something about the brightness of stars, and it tells us something about the temperature of stars. And this line here is what's called the main sequence. This is where most stars spend most of their time burning hydrogen to helium. So the sun is somewhere around — it's around here. And what you see in these little dotted curves here is the history of how we believe a star will live on the main sequence, and then it quickly evolves off this main sequence and changes its brightness very quickly. I won't go into all the details. You won't be tested on the origins of all these things.
But the important thing to show you here is that there are lots of different ideas about what could make a Type Ia supernova and what they would look like if you were able to make a measurement in temperature and brightness. And this is the prototype of a certain type of supernova progenitor called U Sco, RS Oph, and all these other guys right up here are different essential examples within our galaxy that people have pointed to and say, “Hey, I bet that's a prototype of the double-degenerate channel. That's a prototype of that channel.” And so what's exciting is that the previous limits that we've been able to place on any other Type Ia supernova comes way up here, with saying these things had to be — basically this whole region up here was excluded. And you notice that that didn't exclude any of the possibilities of what could make a Type Ia supernova.
But because 2011fe was so close by, we got limits that were about a million — or, sorry, a hundred times better, and that allowed us to essentially exclude this whole yellow region. And that's important because we ruled out this channel right here, the so-called helium star channel, and we ruled out a whole bunch of red giant channels. So for the first time we had direct constraints on the progenitor system of a Type Ia supernova. And what we're able to do, also for the first time, is rule out, essentially directly, red giants and helium star progenitors as the origin of that type of supernova. So that's incredibly important, right? And the way in which astronomers usually go about their business is they have a whole bunch of ideas and then go about falsifying them one by one, and the only thing that's left standing is the one that people wind up believing.
But something that was even more remarkable is that if we look at the early-time data from this light curve, we noticed that it was a very, very smooth curve, and there was no extra light that occurred within the first few days that might be expected if the supernova itself — the explosion — was running into the envelope of its nearby companion. So if I blew this star up right here, then if this star is very nearby, they would wind up slamming into this nearby companion and produce a lot of extra light. And the absence of that extra light allowed us to conclude something, also for the first time: that basically a very compact object is indeed the only thing that was allowed — that was left after we went through the whole process of looking at all the possibilities. The only thing that was left was a white dwarf.
So this is, again, sort of direct evidence and confirmation of our idea that a white dwarf is exploding — that makes a Type Ia supernova. And that's something that hadn't been done in the history of Type Ia supernovae. So it was a very remarkable triumph that we basically showed Type Ia supernovae come from the explosion of a white dwarf. That's a pretty big deal. The other possibility, that it came from the explosion of a white dwarf with a red giant companion, was also excluded. But one of the wonderful things about nature is that she really, really likes to mess with us. And so, lest we think that we solved this whole problem, we had another supernova called PTF11kx — perhaps that'll be the subject of another Dean Lecture one day; this is published last week in Science — that showed reasonably conclusive evidence that there was a red giant companion to that supernova. And this is different than the one that we found: we found 11kly; this is 11kx. So we had a pretty good year last year.
And what's so remarkable about this is that we have really strong evidence — even though people have been looking for such evidence for years — for essentially two different ways of making Type Ia supernovae. And now the big question is out: if there's two very different ways of making Type Ia supernovae — again, probably where there's always a white dwarf exploding, but in some cases it's a white dwarf smashing into another white dwarf, and other cases it's some sort of evolved star companion — how is it that they look so similar? And are we really on okay footing to use every Type Ia supernova that we see to make measurements of the expansion of the universe? So this is an incredibly hot topic, and you can see it's still very, very much alive.
So where are we going in the future? Well, the James Webb Space Telescope is due for launch several years out — about five, six years, I think, from now. We'll be able to study 2011fe probably for the next 10 years with the Hubble Space Telescope, and then with the James Webb Space Telescope after that. And as many of you know, California — both Caltech and the universities of California — are endeavoring to build a 30-meter telescope from the ground in Hawaii. And that project's moving very quickly. It's a billion-dollar facility. But one of the great things is, if we get it, we'll be able to take spectra and really look deep inside this explosion for the next decade. It's going to be absolutely wonderful.
So let me summarize here: supernova science in California is thriving. We have established now a major discovery center, both nearby and distant supernovae, and some of the distant work was credited very nicely by the Nobel Committee last year. And we're leveraging computational, theoretical, observational strengths to really try to understand not just 2011fe, but where it sits in the context of what makes supernovae, but even in the broader context of how we understand our place in the universe. And I obviously didn't have time to get into all the different things that we're seeing in surveys like this and what we might expect to see in new surveys. But the sky is really teeming. I hope you're not seeing the sky anymore as just this fixed picture that just stays the same year after year, decade after decade, millennia after millennia. The sky is evolving. You can think of astronomers now as celestial cinematographers. And the job of those of us working in the time domain is try to find those interesting events and make great sense of them and understand where they fit in.
So let me end with, again, the Bard. And he said, “All the world's a stage, and all the men and women merely players; they have their exits and entrances.” We think about this in the context of the universe now. We think about the world now as the universe. We think about the way in which the universe is evolving as the stage in which galaxies and stars and you exist. And it's those men and women — those supernovae or other transients or time-domain events that are coming in and out of the picture — they're actually teaching us a great deal about the stage itself. So I'll end with that. Thank you very much.
WYATT: We are going to take time for a few questions. If you do need to leave, however, please feel free to do so now, and we'll keep the aisles clear so you can exit either at the top of the theater or down here. If you do have a question, though, we can get the microphone to you without disrupting the flow. Bing Quock and I both have microphones, if you want to raise your hand, and we can get the ball rolling. So it looks like first question from sixth row up.
AUDIENCE: If you wouldn't mind, please go back to that calibration slide where you showed how you calibrate several different intensities. And while you're finding that, I'll just ask my question. I'm trying to remember my basic astronomy. How big a difference is there between, it looks like, about a magnitude 19 and a magnitude 20, slightly more than one magnitude? How big a difference is there in absolute brightness? And what techniques did you use to do the calibration, just at a high conceptual level?
BLOOM: Right. About a factor of two, maybe a factor of three if you've put on some other events, which aren't actually shown here, because those aren't thought to be the normal Ia's. It gets a little bit into the details of what a normal Ia is when you take a spectrum. But the way in which we know, in this case, their absolute brightness — so that is, how bright would they be if I was sitting in the galaxy in which they occurred — is that these are all nearby supernovae whose distances we know, because we have another measurement of distance to the galaxy in which they occurred.
And so this is the famous so-called distance ladder, where — because we don't have any sort of a priori knowledge of how bright these things should be, right? We really haven't been able to produce in supercomputer land the statement that's robust that says they have to be exactly this brightness. If it turned out there was something else, people would dial around the physics to get that answer. But we have observational evidence, because we can make measurements of pulsational variables in our own galaxy, like Cepheids and RR Lyrae; we can make measurements of where those are in the Large Magellanic Cloud, make measurements of other types of variables that give us distances, and then make measurements of those types of objects within the same galaxies that Type Ia supernova go off in.
So if you think about it, climbing a ladder, you better be pretty sure-footed in the lowest rungs. And by the way, for those that know about this, the lowest rungs of the distance ladder — basically it's parallax of the nearest objects, nearest stars to us. And in fact, if we have the distance between the Earth and the Sun wrong, all of this stuff winds up changing. Now, it can't be that wrong, so those rungs are pretty secure. But as we go farther and farther out, there's larger and larger uncertainties, and the systematics of our uncertainties start growing and growing. That's why we're really only able to get distance to about 5% with supernovae. A real holy grail is to try to get that to 1%. But it's the knowledge — which I didn't tell you about until just now — it's the knowledge of the true distances to those supernovae that allow us to calibrate the absolute brightness of those.
WYATT: Next question is from all the way over here on the right, in the fourth row.
AUDIENCE: This actually may be the same question in disguise. The two supernova that you illustrated, of recent vintage — one where you ruled out the red giant, one where you ruled it in — if you take the associated redshifts of them, and the two different mechanisms, do they end up being a standard candle?
BLOOM: Yeah, very good question. Yeah, they fall right smack down on this relationship — this is, by the way, it's called the Phillips relationship — and they are on it. So they are normal Ia's in that sense.
WYATT: So next question, all the way up there, over on the left-hand side.
AUDIENCE: So you mentioned in the closing slides that you can study supernovae for about a year at the moment. Presumably that's a sensor limitation. Is it the case that with better sensors you could learn more after a year, or does a year give you long enough to learn everything that's interesting?
BLOOM: That's a good question. So there are people who make their livelihoods studying supernovae at late times, because as you look at these supernovae and they wind up expanding — so here's a nice expanding supernova — what you wind up seeing is they become transparent over time, and we can basically look directly right through them. So we can see all the elements and things that are created right near the center of this core. So the farther out in time that you go, the better chance you have for seeing the innards of this explosion. As you go farther out in time, there's fewer and fewer people that work on it — not because they wind up getting through their whole career and passing away and stuff. But it's a bit of a niche to work on these things at late times.
That said, there's a very active group of people working around the world on supernova remnants. So if you want to think about it, if we were standing on the Pinwheel Galaxy and we were looking at this remnant right here, this would just look like a faint source, and we wouldn't be able to resolve it, right? So we would see all the same things as if we added up all the light that we see from Earth of this one object. When we study distant supernovae — those outside of basically the Large Magellanic Cloud distances, so even by the time we get to Andromeda — we can't actually resolve those very well. And so when we get a spectrum of them, say a year later or five years later, we're really just getting the integrated light from the whole thing. So there's a lot of work that has to be done to invert that spectrum, turn it back into the physics of what's happening. It's a little bit easier here, because I can take a spectrum here, I can take a spectrum here, I can take a spectrum here. I can take a spectrum across this object right here, and I can really do the 3D tomography through the whole thing. So people are indeed interested in that.
And one of the really remarkable new things that's happening in supernova land within our own galaxy is that people are starting to find echoes of old supernovae. So a supernova goes off, and you think about this expanding sphere of light. Well, that expanding sphere of light is eventually going to wind up hitting a dust cloud somewhere on the other side of the galaxy. And then if that expanding sphere of light actually hit us and we observe the supernova, the light took a long time to hit this other distant dust cloud, and then it's bouncing back to us. So because it's a longer path length, we can actually now very credibly take spectra of old supernovae by taking spectra of their reflections from other places all over the sky. And people have started doing this in the Large Magellanic Cloud and even within our own galaxy, and we can now know for sure what type of supernova Kepler's supernova was, or Brahe's supernova, because we have the echoes of that light. So it's as if we're really reaching back in time and able to actually see those, as if the ancients were taking spectra.
WYATT: That is actually one of the freakiest things — just a comment. Next question, and also from the fourth row.
AUDIENCE: So I'm wondering if you could summarize exactly how your image processing algorithm works — how exactly can you filter all of these artifacts and numbers that you're seeing? And then after you do filter it out, how are you even sure that what you're seeing is what you think it is?
BLOOM: Well, you've just summarized the last four years for me. It's certainly not easy. I may have a couple of backup slides; I'll just start talking to you while I see if I can bring those up. But essentially, as I was saying, the machines are trying to approximate what people would say about those same images. And so what we wind up doing is we wind up putting a bunch of images in front of people, and we say, “What is this? Do you think this is real or not?” And that was one way in which we went about this process, where we got a bunch of data that essentially gave us a bunch of labels, the correct answers.
And then what you can do is produce a bunch of different metrics on, say, this thing right here — this looks basically like crap, and a trained astronomer would say that. Actually, you guys probably say that's crap too, even if you're not a trained astronomer. That's crap, and that's real. So you can show a bunch of experts this and try to get their answers for what these things are. You can then develop metrics on this image, right, where I can try to figure out what the variance is in different directions here. I can try to fit a nice smooth Gaussian to it and get some numbers out of that. We wound up getting about 50 different numbers out of every single image here. And then what you do is you can train machines to take in those numbers and do the best it can to predict what the experts would have said. And we can validate how well we're doing, because we have real-life results where we actually go and, say, we take a spectrum and we know it's a supernova, so it's not an artifact. So it validates that this is actually real.
So what we've been able to do over the last year is go back and look at all of the data that we've gotten from the Palomar Transient Factory and build up essentially a ground truth, and then retrain the machines. So the machines are getting better as we have a better and better understanding of how the survey is actually progressing. And it's an interesting question to know how universal some of these algorithms are, and we believe they probably are when we apply them to new surveys. So that's sort of the state of the art for us, is to take the same algorithms and retrain it on new surveys. Because 50 gigabytes a night is not that much data. I'm involved in the so-called Large Synoptic Survey Telescope, which will probably see its first light at the end of the decade. And it's expected to see something like a million supernova a year, and it will be producing on order a terabyte of data a night of imaging, and we're talking about petabyte databases by the end. So there is a real problem — that's a lot of undergraduates, right? And we really do need to have machines take over that role that actually undergraduates have been part of for a very long time.
WYATT: Next question all the way at the top.
AUDIENCE: I want to get back to the expansion of the universe, and I've been wondering for a couple of years now why Alex Filippenko wasn't on the list of astronomers who received the Nobel Prize.
BLOOM: So, many of you know that Alex Filippenko is another professor at Cal, and he's a world-famous astronomer for his work in supernovae of all different types. Alex was heavily involved, actually, in both teams. I believe he's the only author on both papers because of his work that he did for both teams on the Keck Telescope. And one of the triumphs — of course people are important, but technology is important too, and facilities. One of the things that I think is unheralded is the important role that California telescopes had in that Nobel Prize. Obviously the people got recognized, but the people that built the large telescopes that made that happen — something that also needs to get recognized, and too often the telescope builders aren't also recognized.
Alex was not the lead author on either of those. Adam Riess was his postdoc at the time that Adam was doing the work, but Adam was the co-lead, essentially, of one of the two major groups. And so there's only three Nobel Prizes given, and Saul Perlmutter was the head of one group, and there were essentially two heads of the other group. It was, I think, a very reasonable statement by the Nobel Committee. You'll have to ask Alex what he thinks about that. I can't speak for him.
And there were a number of other people, like Mark Phillips, who discovered the Phillips relation, surprisingly — basically that work and that discovery is what led to precision cosmology with Type Ia supernovae. He also did not win the prize. There are a lot of people who didn't win the prize. Bob Kirshner was another major figure in all those teams, and he's a professor at Harvard. And if you go down the list, if you gave it out to 10 people, you could probably spread the credit where it's all due.
But for better or for worse — and this is in some sense a statement about how antiquated the Nobel rules are — is that a lot of great science now happens not because of one person who is the person at the top of a paper or the top of a proposal, but it happens because of the vast numbers of people that make each individual piece happen. And it is a shame that a Nobel Prize can't be awarded to more than three people. But we're going to have this problem when the Higgs Nobel Prize is awarded, which probably will be soon. We're going to have the problem when somebody finds dark matter — unless it's one person who does it in their basement, it's very likely going to wind up being a team of hundreds of people that do it.
And so there is a problem with recognition, let alone the fact that if you make a great result and then you pass away, you can't get it posthumously. And there have been a lot of examples throughout history where people have essentially been robbed because they passed away and they weren't recognized in the way that they probably should have. So it's an old prize, and they have the old traditions with it. It's still obviously very, very important.
WYATT: I think our next question is also from the very top.
AUDIENCE: Can you speak to the concept of the universe expansion accelerating in the context of superstring theory?
BLOOM: I could try, but I probably won't. No, so we don't know what causes the expansion of the universe, as I said. And there is an idea that there is a force that acts very differently than the one-over-r-squared law. So gravity acts between two bodies of mass like one over r squared — if I separate their distance by a factor of two, the force that they feel between them decreases by a factor of four. The property of this dark energy, if that is indeed the reason why there is this accelerated expansion, is such that if I increase the volume of space, I now basically increase, proportionally to the increasing volume, the exact amount of this repulsive stuff.
So as if I basically open up this space and all of a sudden there's more stuff in it, rather than — you think about the density of the stuff in it going down because I've taken X number of particles and I've increased the size of this box. There's still X particles in it, so the density has gone down. Here, I increase this thing and now X gets larger. So it's something about the property of spacetime itself, and when you start talking about the property of space and time itself, it's natural to start thinking about forces beyond those envisioned in Newtonian or even Einsteinian gravity. And that's why people are looking for explanation. String theory is a possibility — that there perhaps is, in another dimension, some other forces that we're basically feeling in a lower dimension. But I've just expended how much I know here.
WYATT: We have time for a few more questions. Bing's a little closer to — I have someone over in the sixth row. Rows four, six, and the last row seemed to be our leading question askers for the evening.
AUDIENCE: So please bring up the slide showing the 1998 curve. And my question — from other reading I've done, I've gathered that there's been a lot of uncertainty in the actual value of the Hubble constant, and even now we don't think we know it with very great certainty.
BLOOM: Is that the one?
AUDIENCE: No, sorry, there was the slide that basically shows the original data for the accelerating expansion.
BLOOM: Oh, Hubble's data.
AUDIENCE: Yeah. Okay, or some of the more recent data for the same thing. Yeah, any of those are fine. Looking at that data, I would think we'd have the value of the Hubble constant to within a pretty small percentage, maybe the 5% that you quoted for the uncertainty in the standard candle. But from what I've read, I understand that that's not the case, that we've had many different values for the constant and there's still a lot of uncertainty. Please comment on that. Tell me what the situation is.
BLOOM: To make a measurement of the Hubble constant, which is basically the slope of this line at small redshift — so basically low velocity — you need precision distance measurements, and you need to map distance to this number of redshift expansion. The problem is — and this is a bit of a pernicious coincidence in some ways — that the galaxies out to which we can make precision measurements of distance, using these things like variable stars or by other techniques, are actually reasonably close to us. And when they're reasonably close, they actually are moving not just under the influence of the expansion of the universe, but they're also moving because of the gravity that they're feeling of nearby galaxies. And the typical velocities that we talk about when we think about how galaxies are falling into each other or moving past each other is something on the order of a few hundred kilometers a second.
And it turns out that if you use Hubble's constant here, the current value of that means that we have to go out to distances for a number of velocity which is much larger — let's say 2,000 kilometers a second. You have to go out so far that you can't make precision measurements anymore. And so we have to make precision measurements in essentially a very, very noisy environment. It's not just noisy, it's also correlated. So in different parts of the sky there are different mass distributions of what's called large-scale structure, and galaxies are falling into this, right? The Milky Way itself is falling into Andromeda at, whatever, 350 kilometers a second, and our whole Local Group of galaxies itself is falling into a larger mass distribution at something like 650 kilometers a second.
So all around us, we're basically trying to make the measurement of a car that's speeding by us, but we are ourselves also on a car, and we don't know exactly the direction that we're heading with respect to these other cars, right? And we don't know what their intrinsic velocities are. So we're confounded by the fact that the best measurements are made locally, but to make the actual best, very precise measurement of the Hubble constant, you'd want to make this at very large distances.
WYATT: Next question from the upper left over there.
AUDIENCE: You talked about using machine learning with up to 50 different features to classify novel events. Now, I've used machine learning to classify things in ways that I expect before. I'm curious to know if you're only classifying novel events that you already understand, or if you're able to classify novel events that you've yet to experience and classify.
BLOOM: Yeah, that's a very good question. So we actually have machine learning through the whole pipeline. The one that I described and I answered a question on was related to imaging — so asking a very simple question: is this real or not? And we know how to ask that question. We know what the potpourri of real things are, because in subtraction space, they basically look like a new star, or they look like a divot because the star got fainter. But then you ask a perhaps even more interesting question, which is: now that I know it's real, what is it? Is it a supernova — it's just a Ia, right? Okay, that's great, because I can follow that up, and I know what to do with that. Is it a variable star that belongs to a class that I know about?
And one of the other forefronts that we're pushing on is trying to take the data that we get after we do the discovery and put it into the appropriate buckets of where this belongs. And what we're now working on — and we have a paper coming out soon on this — is anomaly detection, basically finding novelties of objects that stick out because they're real, but they don't look at all like the stuff that we already know about. Many times it's mundane explanations for why that is, and we've yet to find a completely new object because of this technique, but that is coming.
And in some sense, that's where the logical extreme is with needle-in-the-haystack work, is that you're no longer looking for the needles you know about, but you're trying to find all the hay and you throw those out. You find all the needles and you throw that out, and you find a diamond, right? And you didn't even know about those before. So that is the destiny of where things are heading, and our group is definitely working on that front.
WYATT: Well, with that thought — diamonds in a haystack — we can call our questioning to a close, but thank you very much for your talk this evening.
BLOOM: Thank you.
QUOCK: The recording ends here, but you can explore more on iTunes or at www.calacademy.org.