How to take a picture of a black hole | Katie Bouman | TEDxBeaconStreet
-
0:19 - 0:21In the movie "Interstellar,"
-
0:21 - 0:25we get an up-close look
at a supermassive black hole. -
0:25 - 0:27Set against a backdrop of bright gas,
-
0:27 - 0:29the black hole's massive
gravitational pull -
0:29 - 0:30bends light into a ring.
-
0:30 - 0:32However, this isn't a real photograph,
-
0:33 - 0:34but a computer graphic rendering --
-
0:34 - 0:38an artistic interpretation
of what a black hole might look like. -
0:38 - 0:40A hundred years ago,
-
0:40 - 0:43Albert Einstein first published
his theory of general relativity. -
0:43 - 0:45In the years since then,
-
0:45 - 0:48scientists have provided
a lot of evidence in support of it. -
0:48 - 0:51But one thing predicted
from this theory, black holes, -
0:51 - 0:53still have not been directly observed.
-
0:53 - 0:56Although we have some idea
as to what a black hole might look like, -
0:56 - 0:59we've never actually taken
a picture of one before. -
0:59 - 1:01However, you might be surprised to know
-
1:01 - 1:06that we may be seeing our first picture
of a black hole in the next couple years. -
1:06 - 1:09Getting this first picture will come down
to an international team of scientists, -
1:10 - 1:11an Earth-sized telescope
-
1:11 - 1:14and an algorithm that puts together
the final picture. -
1:14 - 1:17Although I won't be able to show you
a real picture of a black hole today, -
1:18 - 1:20I'd like to give you a brief glimpse
into the effort involved -
1:20 - 1:22in getting that first picture.
-
1:24 - 1:25My name is Katie Bouman,
-
1:25 - 1:28and I'm a PhD student at MIT.
-
1:28 - 1:30I do research in a computer science lab
-
1:30 - 1:33that works on making computers
see through images and video. -
1:34 - 1:36But although I'm not an astronomer,
-
1:36 - 1:37today I'd like to show you
-
1:37 - 1:40how I've been able to contribute
to this exciting project. -
1:42 - 1:45If you go out past
the bright city lights tonight, -
1:45 - 1:48you may just be lucky enough
to see a stunning view -
1:48 - 1:49of the Milky Way Galaxy.
-
1:50 - 1:52And if you could zoom past
millions of stars, -
1:52 - 1:5626,000 light-years toward the heart
of the spiraling Milky Way, -
1:56 - 1:59we'd eventually reach
a cluster of stars right at the center. -
1:59 - 2:03Peering past all the galactic dust
with infrared telescopes, -
2:03 - 2:07astronomers have watched these stars
for over 16 years. -
2:07 - 2:10But it's what they don't see
that is the most spectacular. -
2:10 - 2:13These stars seem to orbit
an invisible object. -
2:16 - 2:18By tracking the paths of these stars,
-
2:18 - 2:19astronomers have concluded
-
2:19 - 2:22that the only thing small and heavy
enough to cause this motion -
2:22 - 2:24is a supermassive black hole --
-
2:24 - 2:29an object so dense that it sucks up
anything that ventures too close -- -
2:29 - 2:30even light.
-
2:30 - 2:33But what happens if we were
to zoom in even further? -
2:33 - 2:38Is it possible to see something
that, by definition, is impossible to see? -
2:40 - 2:43Well, it turns out that if we were
to zoom in at radio wavelengths, -
2:43 - 2:44we'd expect to see a ring of light
-
2:44 - 2:47caused by the gravitational
lensing of hot plasma -
2:47 - 2:49zipping around the black hole.
-
2:49 - 2:50In other words,
-
2:50 - 2:53the black hole casts a shadow
on this backdrop of bright material, -
2:53 - 2:55carving out a sphere of darkness.
-
2:55 - 2:59This bright ring reveals
the black hole's event horizon, -
2:59 - 3:01where the gravitational pull
becomes so great -
3:01 - 3:03that not even light can escape.
-
3:05 - 3:08Einstein's equations predict
the size and shape of this ring, -
3:08 - 3:11so taking a picture of it
wouldn't only be really cool, -
3:11 - 3:14it would also help to verify
that these equations hold -
3:14 - 3:16in the extreme conditions
around the black hole. -
3:16 - 3:19However, this black hole
is so far away from us, -
3:19 - 3:22that from Earth, this ring appears
incredibly small -- -
3:22 - 3:26the same size to us as an orange
on the surface of the moon. -
3:26 - 3:29That makes taking a picture of it
extremely difficult. -
3:30 - 3:32Why is that?
-
3:32 - 3:35Well, it all comes down
to a simple equation. -
3:35 - 3:38Due to a phenomenon called diffraction,
-
3:38 - 3:39there are fundamental limits
-
3:39 - 3:42to the smallest objects
that we can possibly see. -
3:42 - 3:46This governing equation says
that in order to see smaller and smaller, -
3:46 - 3:49we need to make our telescope
bigger and bigger. -
3:49 - 3:52But even with the most powerful
optical telescopes here on Earth, -
3:52 - 3:54we can't even get close
to the resolution necessary -
3:54 - 3:56to image on the surface of the moon.
-
3:56 - 4:00In fact, here I show one of the highest
resolution images ever taken -
4:00 - 4:01of the moon from Earth.
-
4:01 - 4:04It contains roughly 13,000 pixels,
-
4:04 - 4:08and yet each pixel would contain
over 1.5 million oranges. -
4:09 - 4:11So how big of a telescope do we need
-
4:11 - 4:14in order to see an orange
on the surface of the moon -
4:14 - 4:16and, by extension, our black hole?
-
4:16 - 4:18Well, it turns out
that by crunching the numbers, -
4:18 - 4:21you can easily calculate
that we would need a telescope -
4:21 - 4:22the size of the entire Earth.
-
4:22 - 4:23(Laughter)
-
4:23 - 4:26If we could build
this Earth-sized telescope, -
4:26 - 4:29we could just start to make out
that distinctive ring of light -
4:29 - 4:31indicative of the black
hole's event horizon. -
4:31 - 4:34Although this picture wouldn't contain
all the detail we see -
4:34 - 4:35in computer graphic renderings,
-
4:35 - 4:38it would allow us to safely get
our first glimpse -
4:38 - 4:40of the immediate environment
around a black hole. -
4:41 - 4:42However, as you can imagine,
-
4:42 - 4:46building a single-dish telescope
the size of the Earth is impossible. -
4:46 - 4:48But in the famous words of Mick Jagger,
-
4:48 - 4:50"You can't always get what you want,
-
4:50 - 4:52but if you try sometimes,
you just might find -
4:52 - 4:53you get what you need."
-
4:53 - 4:56And by connecting telescopes
from around the world, -
4:56 - 4:59an international collaboration
called the Event Horizon Telescope -
4:59 - 5:02is creating a computational telescope
the size of the Earth, -
5:02 - 5:04capable of resolving structure
-
5:04 - 5:06on the scale of a black
hole's event horizon. -
5:07 - 5:10This network of telescopes is scheduled
to take its very first picture -
5:10 - 5:12of a black hole next year.
-
5:14 - 5:17Each telescope in the worldwide
network works together. -
5:17 - 5:20Linked through the precise timing
of atomic clocks, -
5:20 - 5:23teams of researchers at each
of the sights freeze light -
5:23 - 5:26by collecting thousands
of terabytes of data. -
5:26 - 5:31This data is then processed in a lab
right here in Massachusetts. -
5:33 - 5:34So how does this even work?
-
5:34 - 5:38Remember if we want to see the black hole
in the center of our galaxy, -
5:38 - 5:41we need to build this impossibly large
Earth-sized telescope? -
5:41 - 5:43For just a second,
let's pretend we could build -
5:43 - 5:45a telescope the size of the Earth.
-
5:45 - 5:47This would be a little bit
like turning the Earth -
5:47 - 5:49into a giant spinning disco ball.
-
5:49 - 5:51Each individual mirror would collect light
-
5:51 - 5:54that we could then combine
together to make a picture. -
5:54 - 5:57However, now let's say
we remove most of those mirrors -
5:57 - 5:59so only a few remained.
-
5:59 - 6:02We could still try to combine
this information together, -
6:02 - 6:04but now there are a lot of holes.
-
6:04 - 6:08These remaining mirrors represent
the locations where we have telescopes. -
6:08 - 6:12This is an incredibly small number
of measurements to make a picture from. -
6:12 - 6:16But although we only collect light
at a few telescope locations, -
6:16 - 6:19as the Earth rotates, we get to see
other new measurements. -
6:20 - 6:23In other words, as the disco ball spins,
those mirrors change locations -
6:23 - 6:26and we get to observe
different parts of the image. -
6:26 - 6:30The imaging algorithms we develop
fill in the missing gaps of the disco ball -
6:30 - 6:33in order to reconstruct
the underlying black hole image. -
6:33 - 6:36If we had telescopes located
everywhere on the globe -- -
6:36 - 6:38in other words, the entire disco ball --
-
6:38 - 6:39this would be trivial.
-
6:39 - 6:43However, we only see a few samples,
and for that reason, -
6:43 - 6:45there are an infinite number
of possible images -
6:45 - 6:48that are perfectly consistent
with our telescope measurements. -
6:49 - 6:52However, not all images are created equal.
-
6:52 - 6:57Some of those images look more like
what we think of as images than others. -
6:57 - 7:00And so, my role in helping to take
the first image of a black hole -
7:00 - 7:03is to design algorithms that find
the most reasonable image -
7:03 - 7:05that also fits the telescope measurements.
-
7:06 - 7:10Just as a forensic sketch artist
uses limited descriptions -
7:10 - 7:14to piece together a picture using
their knowledge of face structure, -
7:14 - 7:17the imaging algorithms I develop
use our limited telescope data -
7:17 - 7:22to guide us to a picture that also
looks like stuff in our universe. -
7:22 - 7:26Using these algorithms,
we're able to piece together pictures -
7:26 - 7:28from this sparse, noisy data.
-
7:28 - 7:33So here I show a sample reconstruction
done using simulated data, -
7:33 - 7:35when we pretend to point our telescopes
-
7:35 - 7:37to the black hole
in the center of our galaxy. -
7:37 - 7:42Although this is just a simulation,
reconstruction such as this give us hope -
7:42 - 7:45that we'll soon be able to reliably take
the first image of a black hole -
7:45 - 7:48and from it, determine
the size of its ring. -
7:50 - 7:53Although I'd love to go on
about all the details of this algorithm, -
7:53 - 7:56luckily for you, I don't have the time.
-
7:56 - 7:58But I'd still like
to give you a brief idea -
7:58 - 8:00of how we define
what our universe looks like, -
8:00 - 8:04and how we use this to reconstruct
and verify our results. -
8:05 - 8:08Since there are an infinite number
of possible images -
8:08 - 8:10that perfectly explain
our telescope measurements, -
8:10 - 8:13we have to choose
between them in some way. -
8:13 - 8:15We do this by ranking the images
-
8:15 - 8:17based upon how likely they are
to be the black hole image, -
8:17 - 8:20and then choosing the one
that's most likely. -
8:20 - 8:22So what do I mean by this exactly?
-
8:22 - 8:24Let's say we were trying to make a model
-
8:24 - 8:28that told us how likely an image
were to appear on Facebook. -
8:28 - 8:29We'd probably want the model to say
-
8:29 - 8:33it's pretty unlikely that someone
would post this noise image on the left, -
8:33 - 8:35and pretty likely that someone
would post a selfie -
8:35 - 8:37like this one on the right.
-
8:37 - 8:38The image in the middle is blurry,
-
8:38 - 8:41so even though it's more likely
we'd see it on Facebook -
8:41 - 8:42compared to the noise image,
-
8:42 - 8:45it's probably less likely we'd see it
compared to the selfie. -
8:46 - 8:48But when it comes to images
from the black hole, -
8:48 - 8:52we're posed with a real conundrum:
we've never seen a black hole before. -
8:52 - 8:54In that case, what is a likely
black hole image, -
8:54 - 8:57and what should we assume
about the structure of black holes? -
8:58 - 9:00We could try to use images
from simulations we've done, -
9:00 - 9:03like the image of the black hole
from "Interstellar," -
9:03 - 9:06but if we did this,
it could cause some serious problems. -
9:07 - 9:11What would happen
if Einstein's theories didn't hold? -
9:11 - 9:15We'd still want to reconstruct
an accurate picture of what was going on. -
9:15 - 9:18If we bake Einstein's equations
too much into our algorithms, -
9:18 - 9:21we'll just end up seeing
what we expect to see. -
9:21 - 9:23In other words,
we want to leave the option open -
9:23 - 9:26for there being a giant elephant
at the center of our galaxy. -
9:26 - 9:27(Laughter)
-
9:28 - 9:31Different types of images have
very distinct features. -
9:31 - 9:35We can easily tell the difference
between black hole simulation images -
9:35 - 9:37and images we take
every day here on Earth. -
9:37 - 9:40We need a way to tell our algorithms
what images look like -
9:40 - 9:43without imposing one type
of image's features too much. -
9:44 - 9:46One way we can try to get around this
-
9:46 - 9:49is by imposing the features
of different kinds of images -
9:49 - 9:53and seeing how the type of image we assume
affects our reconstructions. -
9:55 - 9:58If all images' types produce
a very similar-looking image, -
9:58 - 10:00then we can start to become more confident
-
10:00 - 10:04that the image assumptions we're making
are not biasing this picture that much. -
10:04 - 10:07This is a little bit like
giving the same description -
10:07 - 10:10to three different sketch artists
from all around the world. -
10:10 - 10:13If they all produce
a very similar-looking face, -
10:13 - 10:15then we can start to become confident
-
10:15 - 10:19that they're not imposing their own
cultural biases on the drawings. -
10:20 - 10:23One way we can try to impose
different image features -
10:23 - 10:26is by using pieces of existing images.
-
10:26 - 10:29So we take a large collection of images,
-
10:29 - 10:31and we break them down
into their little image patches. -
10:31 - 10:36We then can treat each image patch
a little bit like pieces of a puzzle. -
10:36 - 10:40And we use commonly seen puzzle pieces
to piece together an image -
10:40 - 10:42that also fits our telescope measurements.
-
10:47 - 10:50Different types of images have
very distinctive sets of puzzle pieces. -
10:51 - 10:54So what happens when we take the same data
-
10:54 - 10:58but we use different sets of puzzle pieces
to reconstruct the image? -
10:58 - 11:02Let's first start with black hole
image simulation puzzle pieces. -
11:04 - 11:06OK, this looks reasonable.
-
11:06 - 11:08This looks like what we expect
a black hole to look like. -
11:08 - 11:09But did we just get it
-
11:09 - 11:13because we just fed it little pieces
of black hole simulation images? -
11:13 - 11:15Let's try another set of puzzle pieces
-
11:15 - 11:17from astronomical, non-black hole objects.
-
11:18 - 11:20OK, we get a similar-looking image.
-
11:20 - 11:23And then how about pieces
from everyday images, -
11:23 - 11:25like the images you take
with your own personal camera? -
11:27 - 11:29Great, we see the same image.
-
11:29 - 11:32When we get the same image
from all different sets of puzzle pieces, -
11:32 - 11:34then we can start to become more confident
-
11:34 - 11:36that the image assumptions we're making
-
11:36 - 11:39aren't biasing the final
image we get too much. -
11:40 - 11:43Another thing we can do is take
the same set of puzzle pieces, -
11:43 - 11:46such as the ones derived
from everyday images, -
11:46 - 11:49and use them to reconstruct
many different kinds of source images. -
11:49 - 11:51So in our simulations,
-
11:51 - 11:55we pretend a black hole looks like
astronomical non-black hole objects, -
11:55 - 11:58as well as everyday images like
the elephant in the center of our galaxy. -
11:58 - 12:02When the results of our algorithms
on the bottom look very similar -
12:02 - 12:04to the simulation's truth image on top,
-
12:04 - 12:07then we can start to become
more confident in our algorithms. -
12:07 - 12:09And I really want to emphasize here
-
12:09 - 12:11that all of these pictures were created
-
12:11 - 12:14by piecing together little pieces
of everyday photographs, -
12:14 - 12:16like you'd take with your own
personal camera. -
12:16 - 12:20So an image of a black hole
we've never seen before -
12:20 - 12:24may eventually be created by piecing
together pictures we see all the time -
12:25 - 12:27Imaging ideas like this
will make it possible for us -
12:27 - 12:30to take our very first pictures
of a black hole, -
12:30 - 12:32and hopefully, verify
those famous theories -
12:32 - 12:35on which scientists rely on a daily basis.
-
12:36 - 12:38But of course, getting
imaging ideas like this working -
12:38 - 12:42would never have been possible
without the amazing team of researchers -
12:42 - 12:44that I have the privilege to work with.
-
12:44 - 12:45It still amazes me
-
12:45 - 12:48that although I began this project
with no background in astrophysics, -
12:48 - 12:51what we have achieved
through this unique collaboration -
12:51 - 12:54could result in the very first
images of a black hole. -
12:54 - 12:57But big projects like
the Event Horizon Telescope -
12:57 - 13:00are successful due to all
the interdisciplinary expertise -
13:00 - 13:02different people bring to the table.
-
13:02 - 13:04We're a melting pot of astronomers,
-
13:04 - 13:06physicists, mathematicians and engineers.
-
13:06 - 13:08This is what will make it soon possible
-
13:08 - 13:11to achieve something
once thought impossible. -
13:11 - 13:13I'd like to encourage all of you to go out
-
13:13 - 13:15and help push the boundaries of science,
-
13:15 - 13:19even if it may at first seem
as mysterious to you as a black hole. -
13:19 - 13:20Thank you.
-
13:20 - 13:26(Applause)
- Title:
- How to take a picture of a black hole | Katie Bouman | TEDxBeaconStreet
- Description:
-
To take a photo of a black hole, you'd need a telescope the size of a planet. That's not really feasible, but Katie Bouman and her team came up with an alternative solution involving complex algorithms and global cooperation. Check out this talk to learn about how we can see in the ultimate dark.
Katie Bouman is a Ph.D. candidate in the Computer Science and Artificial Intelligence Laboratory (CSAIL) at the Massachusetts Institute of Technology (MIT), under the supervision of William T. Freeman. She previously received a B.S.E. in Electrical Engineering from the University of Michigan, Ann Arbor, MI in 2011 and an S.M. degree in Electrical Engineering and Computer Science from MIT, Cambridge, MA in 2013. The focus of Katie’s research is on using emerging computational methods to push the boundaries of interdisciplinary imaging.
This talk was given at a TEDx event using the TED conference format but independently organized by a local community. Learn more at http://ted.com/tedx
- Video Language:
- English
- Team:
- closed TED
- Project:
- TEDxTalks
- Duration:
- 13:33
Krystian Aparta edited English subtitles for How to take a picture of a black hole | Katie Bouman | TEDxBeaconStreet | ||
Krystian Aparta edited English subtitles for How to take a picture of a black hole | Katie Bouman | TEDxBeaconStreet | ||
Krystian Aparta edited English subtitles for How to take a picture of a black hole | Katie Bouman | TEDxBeaconStreet |