1 00:00:01,436 --> 00:00:03,296 In the movie "Interstellar," 2 00:00:03,320 --> 00:00:06,647 we get an up-close look at a supermassive black hole. 3 00:00:06,671 --> 00:00:08,814 Set against a backdrop of bright gas, 4 00:00:08,838 --> 00:00:10,956 the black hole's massive gravitational pull 5 00:00:10,980 --> 00:00:12,415 bends light into a ring. 6 00:00:12,439 --> 00:00:14,548 However, this isn't a real photograph, 7 00:00:14,572 --> 00:00:16,358 but a computer graphic rendering -- 8 00:00:16,382 --> 00:00:19,772 an artistic interpretation of what a black hole might look like. 9 00:00:20,401 --> 00:00:21,567 A hundred years ago, 10 00:00:21,591 --> 00:00:25,192 Albert Einstein first published his theory of general relativity. 11 00:00:25,216 --> 00:00:26,655 In the years since then, 12 00:00:26,679 --> 00:00:29,652 scientists have provided a lot of evidence in support of it. 13 00:00:29,676 --> 00:00:32,760 But one thing predicted from this theory, black holes, 14 00:00:32,784 --> 00:00:35,134 still have not been directly observed. 15 00:00:35,158 --> 00:00:38,364 Although we have some idea as to what a black hole might look like, 16 00:00:38,388 --> 00:00:41,167 we've never actually taken a picture of one before. 17 00:00:41,191 --> 00:00:45,470 However, you might be surprised to know that that may soon change. 18 00:00:45,494 --> 00:00:49,658 We may be seeing our first picture of a black hole in the next couple years. 19 00:00:49,682 --> 00:00:53,640 Getting this first picture will come down to an international team of scientists, 20 00:00:53,664 --> 00:00:55,231 an Earth-sized telescope 21 00:00:55,255 --> 00:00:58,087 and an algorithm that puts together the final picture. 22 00:00:58,111 --> 00:01:01,639 Although I won't be able to show you a real picture of a black hole today, 23 00:01:01,663 --> 00:01:04,574 I'd like to give you a brief glimpse into the effort involved 24 00:01:04,598 --> 00:01:06,211 in getting that first picture. 25 00:01:07,477 --> 00:01:08,914 My name is Katie Bouman, 26 00:01:08,938 --> 00:01:11,504 and I'm a PhD student at MIT. 27 00:01:11,528 --> 00:01:13,555 I do research in a computer science lab 28 00:01:13,579 --> 00:01:16,877 that works on making computers see through images and video. 29 00:01:16,901 --> 00:01:19,063 But although I'm not an astronomer, 30 00:01:19,087 --> 00:01:20,372 today I'd like to show you 31 00:01:20,396 --> 00:01:23,299 how I've been able to contribute to this exciting project. 32 00:01:23,323 --> 00:01:26,154 If you go out past the bright city lights tonight, 33 00:01:26,178 --> 00:01:28,614 you may just be lucky enough to see a stunning view 34 00:01:28,638 --> 00:01:30,131 of the Milky Way Galaxy. 35 00:01:30,155 --> 00:01:32,617 And if you could zoom past millions of stars, 36 00:01:32,641 --> 00:01:36,396 26,000 light-years toward the heart of the spiraling Milky Way, 37 00:01:36,420 --> 00:01:39,941 we'd eventually reach a cluster of stars right at the center. 38 00:01:39,965 --> 00:01:43,171 Peering past all the galactic dust with infrared telescopes, 39 00:01:43,195 --> 00:01:47,062 astronomers have watched these stars for over 16 years. 40 00:01:47,086 --> 00:01:50,675 But it's what they don't see that is the most spectacular. 41 00:01:50,699 --> 00:01:53,765 These stars seem to orbit an invisible object. 42 00:01:53,789 --> 00:01:56,112 By tracking the paths of these stars, 43 00:01:56,136 --> 00:01:57,430 astronomers have concluded 44 00:01:57,454 --> 00:02:00,583 that the only thing small and heavy enough to cause this motion 45 00:02:00,607 --> 00:02:02,575 is a supermassive black hole -- 46 00:02:02,599 --> 00:02:06,777 an object so dense that it sucks up anything that ventures too close -- 47 00:02:06,801 --> 00:02:08,295 even light. 48 00:02:08,319 --> 00:02:11,380 But what happens if we were to zoom in even further? 49 00:02:11,404 --> 00:02:16,137 Is it possible to see something that, by definition, is impossible to see? 50 00:02:16,719 --> 00:02:19,963 Well, it turns out that if we were to zoom in at radio wavelengths, 51 00:02:19,987 --> 00:02:21,669 we'd expect to see a ring of light 52 00:02:21,693 --> 00:02:24,104 caused by the gravitational lensing of hot plasma 53 00:02:24,128 --> 00:02:25,957 zipping around the black hole. 54 00:02:25,981 --> 00:02:27,141 In other words, 55 00:02:27,165 --> 00:02:30,336 the black hole casts a shadow on this backdrop of bright material, 56 00:02:30,360 --> 00:02:32,202 carving out a sphere of darkness. 57 00:02:32,226 --> 00:02:35,565 This bright ring reveals the black hole's event horizon, 58 00:02:35,589 --> 00:02:37,989 where the gravitational pull becomes so great 59 00:02:38,013 --> 00:02:39,639 that not even light can escape. 60 00:02:39,663 --> 00:02:42,522 Einstein's equations predict the size and shape of this ring, 61 00:02:42,546 --> 00:02:45,754 so taking a picture of it wouldn't only be really cool, 62 00:02:45,778 --> 00:02:48,396 it would also help to verify that these equations hold 63 00:02:48,420 --> 00:02:50,886 in the extreme conditions around the black hole. 64 00:02:50,910 --> 00:02:53,468 However, this black hole is so far away from us, 65 00:02:53,492 --> 00:02:56,590 that from Earth, this ring appears incredibly small -- 66 00:02:56,614 --> 00:03:00,204 the same size to us as an orange on the surface of the moon. 67 00:03:00,758 --> 00:03:03,582 That makes taking a picture of it extremely difficult. 68 00:03:04,645 --> 00:03:05,947 Why is that? 69 00:03:06,512 --> 00:03:09,700 Well, it all comes down to a simple equation. 70 00:03:09,724 --> 00:03:12,140 Due to a phenomenon called diffraction, 71 00:03:12,164 --> 00:03:13,519 there are fundamental limits 72 00:03:13,543 --> 00:03:16,213 to the smallest objects that we can possibly see. 73 00:03:16,789 --> 00:03:20,461 This governing equation says that in order to see smaller and smaller, 74 00:03:20,485 --> 00:03:23,072 we need to make our telescope bigger and bigger. 75 00:03:23,096 --> 00:03:26,165 But even with the most powerful optical telescopes here on Earth, 76 00:03:26,189 --> 00:03:28,608 we can't even get close to the resolution necessary 77 00:03:28,632 --> 00:03:30,830 to image on the surface of the moon. 78 00:03:30,854 --> 00:03:34,471 In fact, here I show one of the highest resolution images ever taken 79 00:03:34,495 --> 00:03:35,892 of the moon from Earth. 80 00:03:35,916 --> 00:03:38,473 It contains roughly 13,000 pixels, 81 00:03:38,497 --> 00:03:42,547 and yet each pixel would contain over 1.5 million oranges. 82 00:03:43,396 --> 00:03:45,368 So how big of a telescope do we need 83 00:03:45,392 --> 00:03:48,157 in order to see an orange on the surface of the moon 84 00:03:48,181 --> 00:03:50,395 and, by extension, our black hole? 85 00:03:50,419 --> 00:03:52,759 Well, it turns out that by crunching the numbers, 86 00:03:52,783 --> 00:03:55,393 you can easily calculate that we would need a telescope 87 00:03:55,417 --> 00:03:56,810 the size of the entire Earth. 88 00:03:56,834 --> 00:03:57,858 (Laughter) 89 00:03:57,882 --> 00:04:00,001 If we could build this Earth-sized telescope, 90 00:04:00,025 --> 00:04:02,950 we could just start to make out that distinctive ring of light 91 00:04:02,974 --> 00:04:05,157 indicative of the black hole's event horizon. 92 00:04:05,181 --> 00:04:08,099 Although this picture wouldn't contain all the detail we see 93 00:04:08,123 --> 00:04:09,629 in computer graphic renderings, 94 00:04:09,653 --> 00:04:11,952 it would allow us to safely get our first glimpse 95 00:04:11,976 --> 00:04:14,463 of the immediate environment around a black hole. 96 00:04:14,487 --> 00:04:16,100 However, as you can imagine, 97 00:04:16,124 --> 00:04:19,748 building a single-dish telescope the size of the Earth is impossible. 98 00:04:19,772 --> 00:04:21,659 But in the famous words of Mick Jagger, 99 00:04:21,683 --> 00:04:23,474 "You can't always get what you want, 100 00:04:23,498 --> 00:04:25,685 but if you try sometimes, you just might find 101 00:04:25,709 --> 00:04:26,924 you get what you need." 102 00:04:26,948 --> 00:04:29,412 And by connecting telescopes from around the world, 103 00:04:29,436 --> 00:04:32,974 an international collaboration called the Event Horizon Telescope 104 00:04:32,998 --> 00:04:36,107 is creating a computational telescope the size of the Earth, 105 00:04:36,131 --> 00:04:37,668 capable of resolving structure 106 00:04:37,692 --> 00:04:39,891 on the scale of a black hole's event horizon. 107 00:04:39,915 --> 00:04:43,302 This network of telescopes is scheduled to take its very first picture 108 00:04:43,326 --> 00:04:45,141 of a black hole next year. 109 00:04:45,165 --> 00:04:48,503 Each telescope in the worldwide network works together. 110 00:04:48,527 --> 00:04:51,239 Linked through the precise timing of atomic clocks, 111 00:04:51,263 --> 00:04:53,920 teams of researchers at each of the sites freeze light 112 00:04:53,944 --> 00:04:56,906 by collecting thousands of terabytes of data. 113 00:04:56,930 --> 00:05:01,947 This data is then processed in a lab right here in Massachusetts. 114 00:05:01,971 --> 00:05:03,765 So how does this even work? 115 00:05:03,789 --> 00:05:07,192 Remember if we want to see the black hole in the center of our galaxy, 116 00:05:07,216 --> 00:05:10,198 we need to build this impossibly large Earth-sized telescope? 117 00:05:10,222 --> 00:05:12,454 For just a second, let's pretend we could build 118 00:05:12,478 --> 00:05:14,320 a telescope the size of the Earth. 119 00:05:14,344 --> 00:05:16,799 This would be a little bit like turning the Earth 120 00:05:16,823 --> 00:05:18,570 into a giant spinning disco ball. 121 00:05:18,594 --> 00:05:20,794 Each individual mirror would collect light 122 00:05:20,818 --> 00:05:23,415 that we could then combine together to make a picture. 123 00:05:23,439 --> 00:05:26,100 However, now let's say we remove most of those mirrors 124 00:05:26,124 --> 00:05:28,096 so only a few remained. 125 00:05:28,120 --> 00:05:30,997 We could still try to combine this information together, 126 00:05:31,021 --> 00:05:33,014 but now there are a lot of holes. 127 00:05:33,038 --> 00:05:37,411 These remaining mirrors represent the locations where we have telescopes. 128 00:05:37,435 --> 00:05:41,514 This is an incredibly small number of measurements to make a picture from. 129 00:05:41,538 --> 00:05:45,376 But although we only collect light at a few telescope locations, 130 00:05:45,400 --> 00:05:48,823 as the Earth rotates, we get to see other new measurements. 131 00:05:48,847 --> 00:05:52,666 In other words, as the disco ball spins, those mirrors change locations 132 00:05:52,690 --> 00:05:55,589 and we get to observe different parts of the image. 133 00:05:55,613 --> 00:05:59,631 The imaging algorithms we develop fill in the missing gaps of the disco ball 134 00:05:59,655 --> 00:06:02,688 in order to reconstruct the underlying black hole image. 135 00:06:02,712 --> 00:06:05,348 If we had telescopes located everywhere on the globe -- 136 00:06:05,372 --> 00:06:07,313 in other words, the entire disco ball -- 137 00:06:07,337 --> 00:06:08,621 this would be trivial. 138 00:06:08,645 --> 00:06:11,967 However, we only see a few samples, and for that reason, 139 00:06:11,991 --> 00:06:14,379 there are an infinite number of possible images 140 00:06:14,403 --> 00:06:17,367 that are perfectly consistent with our telescope measurements. 141 00:06:17,391 --> 00:06:20,407 However, not all images are created equal. 142 00:06:20,849 --> 00:06:25,307 Some of those images look more like what we think of as images than others. 143 00:06:25,331 --> 00:06:28,553 And so, my role in helping to take the first image of a black hole 144 00:06:28,577 --> 00:06:31,509 is to design algorithms that find the most reasonable image 145 00:06:31,533 --> 00:06:33,755 that also fits the telescope measurements. 146 00:06:34,727 --> 00:06:38,669 Just as a forensic sketch artist uses limited descriptions 147 00:06:38,693 --> 00:06:42,207 to piece together a picture using their knowledge of face structure, 148 00:06:42,231 --> 00:06:45,546 the imaging algorithms I develop use our limited telescope data 149 00:06:45,570 --> 00:06:49,892 to guide us to a picture that also looks like stuff in our universe. 150 00:06:49,916 --> 00:06:53,567 Using these algorithms, we're able to piece together pictures 151 00:06:53,591 --> 00:06:55,771 from this sparse, noisy data. 152 00:06:55,795 --> 00:07:00,324 So here I show a sample reconstruction done using simulated data, 153 00:07:00,348 --> 00:07:02,281 when we pretend to point our telescopes 154 00:07:02,305 --> 00:07:04,890 to the black hole in the center of our galaxy. 155 00:07:04,914 --> 00:07:09,369 Although this is just a simulation, reconstruction such as this give us hope 156 00:07:09,393 --> 00:07:12,846 that we'll soon be able to reliably take the first image of a black hole 157 00:07:12,870 --> 00:07:15,465 and from it, determine the size of its ring. 158 00:07:16,118 --> 00:07:19,317 Although I'd love to go on about all the details of this algorithm, 159 00:07:19,341 --> 00:07:21,515 luckily for you, I don't have the time. 160 00:07:21,539 --> 00:07:23,540 But I'd still like to give you a brief idea 161 00:07:23,564 --> 00:07:25,866 of how we define what our universe looks like, 162 00:07:25,890 --> 00:07:30,356 and how we use this to reconstruct and verify our results. 163 00:07:30,380 --> 00:07:32,876 Since there are an infinite number of possible images 164 00:07:32,900 --> 00:07:35,265 that perfectly explain our telescope measurements, 165 00:07:35,289 --> 00:07:37,894 we have to choose between them in some way. 166 00:07:37,918 --> 00:07:39,756 We do this by ranking the images 167 00:07:39,780 --> 00:07:42,614 based upon how likely they are to be the black hole image, 168 00:07:42,638 --> 00:07:45,120 and then choosing the one that's most likely. 169 00:07:45,144 --> 00:07:47,339 So what do I mean by this exactly? 170 00:07:47,862 --> 00:07:49,840 Let's say we were trying to make a model 171 00:07:49,864 --> 00:07:53,047 that told us how likely an image were to appear on Facebook. 172 00:07:53,071 --> 00:07:54,772 We'd probably want the model to say 173 00:07:54,796 --> 00:07:58,353 it's pretty unlikely that someone would post this noise image on the left, 174 00:07:58,377 --> 00:08:00,796 and pretty likely that someone would post a selfie 175 00:08:00,820 --> 00:08:02,154 like this one on the right. 176 00:08:02,178 --> 00:08:03,817 The image in the middle is blurry, 177 00:08:03,841 --> 00:08:06,480 so even though it's more likely we'd see it on Facebook 178 00:08:06,504 --> 00:08:07,864 compared to the noise image, 179 00:08:07,888 --> 00:08:10,848 it's probably less likely we'd see it compared to the selfie. 180 00:08:10,872 --> 00:08:13,162 But when it comes to images from the black hole, 181 00:08:13,186 --> 00:08:16,688 we're posed with a real conundrum: we've never seen a black hole before. 182 00:08:16,712 --> 00:08:19,003 In that case, what is a likely black hole image, 183 00:08:19,027 --> 00:08:21,965 and what should we assume about the structure of black holes? 184 00:08:21,989 --> 00:08:24,621 We could try to use images from simulations we've done, 185 00:08:24,645 --> 00:08:27,175 like the image of the black hole from "Interstellar," 186 00:08:27,199 --> 00:08:30,137 but if we did this, it could cause some serious problems. 187 00:08:30,161 --> 00:08:33,541 What would happen if Einstein's theories didn't hold? 188 00:08:33,565 --> 00:08:37,526 We'd still want to reconstruct an accurate picture of what was going on. 189 00:08:37,550 --> 00:08:40,921 If we bake Einstein's equations too much into our algorithms, 190 00:08:40,945 --> 00:08:43,700 we'll just end up seeing what we expect to see. 191 00:08:43,724 --> 00:08:46,000 In other words, we want to leave the option open 192 00:08:46,024 --> 00:08:48,947 for there being a giant elephant at the center of our galaxy. 193 00:08:48,971 --> 00:08:50,028 (Laughter) 194 00:08:50,052 --> 00:08:53,041 Different types of images have very distinct features. 195 00:08:53,065 --> 00:08:56,613 We can easily tell the difference between black hole simulation images 196 00:08:56,637 --> 00:08:58,913 and images we take every day here on Earth. 197 00:08:58,937 --> 00:09:02,041 We need a way to tell our algorithms what images look like 198 00:09:02,065 --> 00:09:05,314 without imposing one type of image's features too much. 199 00:09:05,865 --> 00:09:07,758 One way we can try to get around this 200 00:09:07,782 --> 00:09:10,844 is by imposing the features of different kinds of images 201 00:09:10,868 --> 00:09:14,998 and seeing how the type of image we assume affects our reconstructions. 202 00:09:15,712 --> 00:09:19,203 If all images' types produce a very similar-looking image, 203 00:09:19,227 --> 00:09:21,284 then we can start to become more confident 204 00:09:21,308 --> 00:09:25,481 that the image assumptions we're making are not biasing this picture that much. 205 00:09:25,505 --> 00:09:28,495 This is a little bit like giving the same description 206 00:09:28,519 --> 00:09:31,515 to three different sketch artists from all around the world. 207 00:09:31,539 --> 00:09:34,399 If they all produce a very similar-looking face, 208 00:09:34,423 --> 00:09:36,216 then we can start to become confident 209 00:09:36,240 --> 00:09:39,856 that they're not imposing their own cultural biases on the drawings. 210 00:09:39,880 --> 00:09:43,195 One way we can try to impose different image features 211 00:09:43,219 --> 00:09:45,660 is by using pieces of existing images. 212 00:09:46,214 --> 00:09:48,374 So we take a large collection of images, 213 00:09:48,398 --> 00:09:51,116 and we break them down into their little image patches. 214 00:09:51,140 --> 00:09:55,425 We then can treat each image patch a little bit like pieces of a puzzle. 215 00:09:55,449 --> 00:09:59,727 And we use commonly seen puzzle pieces to piece together an image 216 00:09:59,751 --> 00:10:02,203 that also fits our telescope measurements. 217 00:10:03,040 --> 00:10:06,783 Different types of images have very distinctive sets of puzzle pieces. 218 00:10:06,807 --> 00:10:09,613 So what happens when we take the same data 219 00:10:09,637 --> 00:10:13,767 but we use different sets of puzzle pieces to reconstruct the image? 220 00:10:13,791 --> 00:10:18,557 Let's first start with black hole image simulation puzzle pieces. 221 00:10:18,581 --> 00:10:20,172 OK, this looks reasonable. 222 00:10:20,196 --> 00:10:22,890 This looks like what we expect a black hole to look like. 223 00:10:22,914 --> 00:10:24,107 But did we just get it 224 00:10:24,131 --> 00:10:27,445 because we just fed it little pieces of black hole simulation images? 225 00:10:27,469 --> 00:10:29,349 Let's try another set of puzzle pieces 226 00:10:29,373 --> 00:10:31,882 from astronomical, non-black hole objects. 227 00:10:32,914 --> 00:10:35,040 OK, we get a similar-looking image. 228 00:10:35,064 --> 00:10:37,300 And then how about pieces from everyday images, 229 00:10:37,324 --> 00:10:40,109 like the images you take with your own personal camera? 230 00:10:41,312 --> 00:10:43,427 Great, we see the same image. 231 00:10:43,451 --> 00:10:46,817 When we get the same image from all different sets of puzzle pieces, 232 00:10:46,841 --> 00:10:48,887 then we can start to become more confident 233 00:10:48,911 --> 00:10:50,877 that the image assumptions we're making 234 00:10:50,901 --> 00:10:53,822 aren't biasing the final image we get too much. 235 00:10:53,846 --> 00:10:57,099 Another thing we can do is take the same set of puzzle pieces, 236 00:10:57,123 --> 00:10:59,612 such as the ones derived from everyday images, 237 00:10:59,636 --> 00:11:03,236 and use them to reconstruct many different kinds of source images. 238 00:11:03,260 --> 00:11:04,531 So in our simulations, 239 00:11:04,555 --> 00:11:08,330 we pretend a black hole looks like astronomical non-black hole objects, 240 00:11:08,354 --> 00:11:12,203 as well as everyday images like the elephant in the center of our galaxy. 241 00:11:12,227 --> 00:11:15,395 When the results of our algorithms on the bottom look very similar 242 00:11:15,419 --> 00:11:17,515 to the simulation's truth image on top, 243 00:11:17,539 --> 00:11:20,885 then we can start to become more confident in our algorithms. 244 00:11:20,909 --> 00:11:22,776 And I really want to emphasize here 245 00:11:22,800 --> 00:11:24,734 that all of these pictures were created 246 00:11:24,758 --> 00:11:27,694 by piecing together little pieces of everyday photographs, 247 00:11:27,718 --> 00:11:29,933 like you'd take with your own personal camera. 248 00:11:29,957 --> 00:11:33,233 So an image of a black hole we've never seen before 249 00:11:33,257 --> 00:11:37,200 may eventually be created by piecing together pictures we see all the time 250 00:11:37,224 --> 00:11:39,969 of people, buildings, trees, cats and dogs. 251 00:11:39,993 --> 00:11:42,638 Imaging ideas like this will make it possible for us 252 00:11:42,662 --> 00:11:45,281 to take our very first pictures of a black hole, 253 00:11:45,305 --> 00:11:47,752 and hopefully, verify those famous theories 254 00:11:47,776 --> 00:11:50,197 on which scientists rely on a daily basis. 255 00:11:50,221 --> 00:11:52,829 But of course, getting imaging ideas like this working 256 00:11:52,853 --> 00:11:56,175 would never have been possible without the amazing team of researchers 257 00:11:56,199 --> 00:11:58,086 that I have the privilege to work with. 258 00:11:58,110 --> 00:11:59,273 It still amazes me 259 00:11:59,297 --> 00:12:02,648 that although I began this project with no background in astrophysics, 260 00:12:02,672 --> 00:12:05,291 what we have achieved through this unique collaboration 261 00:12:05,315 --> 00:12:08,074 could result in the very first images of a black hole. 262 00:12:08,098 --> 00:12:10,796 But big projects like the Event Horizon Telescope 263 00:12:10,820 --> 00:12:13,634 are successful due to all the interdisciplinary expertise 264 00:12:13,658 --> 00:12:15,448 different people bring to the table. 265 00:12:15,472 --> 00:12:17,178 We're a melting pot of astronomers, 266 00:12:17,202 --> 00:12:19,434 physicists, mathematicians and engineers. 267 00:12:19,458 --> 00:12:22,012 This is what will make it soon possible 268 00:12:22,036 --> 00:12:24,889 to achieve something once thought impossible. 269 00:12:24,913 --> 00:12:27,169 I'd like to encourage all of you to go out 270 00:12:27,193 --> 00:12:29,289 and help push the boundaries of science, 271 00:12:29,313 --> 00:12:33,214 even if it may at first seem as mysterious to you as a black hole. 272 00:12:33,238 --> 00:12:34,412 Thank you. 273 00:12:34,436 --> 00:12:36,833 (Applause)