How to take a picture of a black hole | Katie Bouman | TEDxBeaconStreet

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

In the movie "Interstellar,"

we get an up-close look
at a supermassive black hole.

Set against a backdrop of bright gas,

the black hole's massive
gravitational pull

bends light into a ring.

However, this isn't a real photograph,

but a computer graphic rendering --

an artistic interpretation
of what a black hole might look like.

A hundred years ago,

Albert Einstein first published
his theory of general relativity.

In the years since then,

scientists have provided
a lot of evidence in support of it.

But one thing predicted
from this theory, black holes,

still have not been directly observed.

Although we have some idea
as to what a black hole might look like,

we've never actually taken
a picture of one before.

However, you might be surprised to know

that we may be seeing our first picture
of a black hole in the next couple years.

Getting this first picture will come down
to an international team of scientists,

an Earth-sized telescope

and an algorithm that puts together
the final picture.

Although I won't be able to show you
a real picture of a black hole today,

I'd like to give you a brief glimpse
into the effort involved

in getting that first picture.

My name is Katie Bouman,

and I'm a PhD student at MIT.

I do research in a computer science lab

that works on making computers
see through images and video.

But although I'm not an astronomer,

today I'd like to show you

how I've been able to contribute
to this exciting project.

If you go out past
the bright city lights tonight,

you may just be lucky enough
to see a stunning view

of the Milky Way Galaxy.

And if you could zoom past
millions of stars,

26,000 light-years toward the heart
of the spiraling Milky Way,

we'd eventually reach
a cluster of stars right at the center.

Peering past all the galactic dust
with infrared telescopes,

astronomers have watched these stars
for over 16 years.

But it's what they don't see
that is the most spectacular.

These stars seem to orbit
an invisible object.

By tracking the paths of these stars,

astronomers have concluded

that the only thing small and heavy
enough to cause this motion

is a supermassive black hole --

an object so dense that it sucks up
anything that ventures too close --

even light.

But what happens if we were
to zoom in even further?

Is it possible to see something
that, by definition, is impossible to see?

Well, it turns out that if we were
to zoom in at radio wavelengths,

we'd expect to see a ring of light

caused by the gravitational
lensing of hot plasma

zipping around the black hole.

In other words,

the black hole casts a shadow
on this backdrop of bright material,

carving out a sphere of darkness.

This bright ring reveals
the black hole's event horizon,

where the gravitational pull
becomes so great

that not even light can escape.

Einstein's equations predict
the size and shape of this ring,

so taking a picture of it
wouldn't only be really cool,

it would also help to verify
that these equations hold

in the extreme conditions
around the black hole.

However, this black hole
is so far away from us,

that from Earth, this ring appears
incredibly small --

the same size to us as an orange
on the surface of the moon.

That makes taking a picture of it
extremely difficult.

Why is that?

Well, it all comes down
to a simple equation.

Due to a phenomenon called diffraction,

there are fundamental limits

to the smallest objects
that we can possibly see.

This governing equation says
that in order to see smaller and smaller,

we need to make our telescope
bigger and bigger.

But even with the most powerful
optical telescopes here on Earth,

we can't even get close
to the resolution necessary

to image on the surface of the moon.

In fact, here I show one of the highest
resolution images ever taken

of the moon from Earth.

It contains roughly 13,000 pixels,

and yet each pixel would contain
over 1.5 million oranges.

So how big of a telescope do we need

in order to see an orange
on the surface of the moon

and, by extension, our black hole?

Well, it turns out
that by crunching the numbers,

you can easily calculate
that we would need a telescope

the size of the entire Earth.

(Laughter)

If we could build
this Earth-sized telescope,

we could just start to make out
that distinctive ring of light

indicative of the black
hole's event horizon.

Although this picture wouldn't contain
all the detail we see

in computer graphic renderings,

it would allow us to safely get
our first glimpse

of the immediate environment
around a black hole.

However, as you can imagine,

building a single-dish telescope
the size of the Earth is impossible.

But in the famous words of Mick Jagger,

"You can't always get what you want,

but if you try sometimes,
you just might find

you get what you need."

And by connecting telescopes
from around the world,

an international collaboration
called the Event Horizon Telescope

is creating a computational telescope
the size of the Earth,

capable of resolving structure

on the scale of a black
hole's event horizon.

This network of telescopes is scheduled
to take its very first picture

of a black hole next year.

Each telescope in the worldwide
network works together.

Linked through the precise timing
of atomic clocks,

teams of researchers at each
of the sights freeze light

by collecting thousands
of terabytes of data.

This data is then processed in a lab
right here in Massachusetts.

So how does this even work?

Remember if we want to see the black hole
in the center of our galaxy,

we need to build this impossibly large
Earth-sized telescope?

For just a second,
let's pretend we could build

a telescope the size of the Earth.

This would be a little bit
like turning the Earth

into a giant spinning disco ball.

Each individual mirror would collect light

that we could then combine
together to make a picture.

However, now let's say
we remove most of those mirrors

so only a few remained.

We could still try to combine
this information together,

but now there are a lot of holes.

These remaining mirrors represent
the locations where we have telescopes.

This is an incredibly small number
of measurements to make a picture from.

But although we only collect light
at a few telescope locations,

as the Earth rotates, we get to see
other new measurements.

In other words, as the disco ball spins,
those mirrors change locations

and we get to observe
different parts of the image.

The imaging algorithms we develop
fill in the missing gaps of the disco ball

in order to reconstruct
the underlying black hole image.

If we had telescopes located
everywhere on the globe --

in other words, the entire disco ball --

this would be trivial.

However, we only see a few samples,
and for that reason,

there are an infinite number
of possible images

that are perfectly consistent
with our telescope measurements.

However, not all images are created equal.

Some of those images look more like
what we think of as images than others.

And so, my role in helping to take
the first image of a black hole

is to design algorithms that find
the most reasonable image

that also fits the telescope measurements.

Just as a forensic sketch artist
uses limited descriptions

to piece together a picture using
their knowledge of face structure,

the imaging algorithms I develop
use our limited telescope data

to guide us to a picture that also
looks like stuff in our universe.

Using these algorithms,
we're able to piece together pictures

from this sparse, noisy data.

So here I show a sample reconstruction
done using simulated data,

when we pretend to point our telescopes

to the black hole
in the center of our galaxy.

Although this is just a simulation,
reconstruction such as this give us hope

that we'll soon be able to reliably take
the first image of a black hole

and from it, determine
the size of its ring.

Although I'd love to go on
about all the details of this algorithm,

luckily for you, I don't have the time.

But I'd still like
to give you a brief idea

of how we define
what our universe looks like,

and how we use this to reconstruct
and verify our results.

Since there are an infinite number
of possible images

that perfectly explain
our telescope measurements,

we have to choose
between them in some way.

We do this by ranking the images

based upon how likely they are
to be the black hole image,

and then choosing the one
that's most likely.

So what do I mean by this exactly?

Let's say we were trying to make a model

that told us how likely an image
were to appear on Facebook.

We'd probably want the model to say

it's pretty unlikely that someone
would post this noise image on the left,

and pretty likely that someone
would post a selfie

like this one on the right.

The image in the middle is blurry,

so even though it's more likely
we'd see it on Facebook

compared to the noise image,

it's probably less likely we'd see it
compared to the selfie.

But when it comes to images
from the black hole,

we're posed with a real conundrum:
we've never seen a black hole before.

In that case, what is a likely
black hole image,

and what should we assume
about the structure of black holes?

We could try to use images
from simulations we've done,

like the image of the black hole
from "Interstellar,"

but if we did this,
it could cause some serious problems.

What would happen
if Einstein's theories didn't hold?

We'd still want to reconstruct
an accurate picture of what was going on.

If we bake Einstein's equations
too much into our algorithms,

we'll just end up seeing
what we expect to see.

In other words,
we want to leave the option open

for there being a giant elephant
at the center of our galaxy.

(Laughter)

Different types of images have
very distinct features.

We can easily tell the difference
between black hole simulation images

and images we take
every day here on Earth.

We need a way to tell our algorithms
what images look like

without imposing one type
of image's features too much.

One way we can try to get around this

is by imposing the features
of different kinds of images

and seeing how the type of image we assume
affects our reconstructions.

If all images' types produce
a very similar-looking image,

then we can start to become more confident

that the image assumptions we're making
are not biasing this picture that much.

This is a little bit like
giving the same description

to three different sketch artists
from all around the world.

If they all produce
a very similar-looking face,

then we can start to become confident

that they're not imposing their own
cultural biases on the drawings.

One way we can try to impose
different image features

is by using pieces of existing images.

So we take a large collection of images,

and we break them down
into their little image patches.

We then can treat each image patch
a little bit like pieces of a puzzle.

And we use commonly seen puzzle pieces
to piece together an image

that also fits our telescope measurements.

Different types of images have
very distinctive sets of puzzle pieces.

So what happens when we take the same data

but we use different sets of puzzle pieces
to reconstruct the image?

Let's first start with black hole
image simulation puzzle pieces.

OK, this looks reasonable.

This looks like what we expect
a black hole to look like.

But did we just get it

because we just fed it little pieces
of black hole simulation images?

Let's try another set of puzzle pieces

from astronomical, non-black hole objects.

OK, we get a similar-looking image.

And then how about pieces
from everyday images,

like the images you take
with your own personal camera?

Great, we see the same image.

When we get the same image
from all different sets of puzzle pieces,

then we can start to become more confident

that the image assumptions we're making

aren't biasing the final
image we get too much.

Another thing we can do is take
the same set of puzzle pieces,

such as the ones derived
from everyday images,

and use them to reconstruct
many different kinds of source images.

So in our simulations,

we pretend a black hole looks like
astronomical non-black hole objects,

as well as everyday images like
the elephant in the center of our galaxy.

When the results of our algorithms
on the bottom look very similar

to the simulation's truth image on top,

then we can start to become
more confident in our algorithms.

And I really want to emphasize here

that all of these pictures were created

by piecing together little pieces
of everyday photographs,

like you'd take with your own
personal camera.

So an image of a black hole
we've never seen before

may eventually be created by piecing
together pictures we see all the time

Imaging ideas like this
will make it possible for us

to take our very first pictures
of a black hole,

and hopefully, verify
those famous theories

on which scientists rely on a daily basis.

But of course, getting
imaging ideas like this working

would never have been possible
without the amazing team of researchers

that I have the privilege to work with.

It still amazes me

that although I began this project
with no background in astrophysics,

what we have achieved
through this unique collaboration

could result in the very first
images of a black hole.

But big projects like
the Event Horizon Telescope

are successful due to all
the interdisciplinary expertise

different people bring to the table.

We're a melting pot of astronomers,

physicists, mathematicians and engineers.

This is what will make it soon possible

to achieve something
once thought impossible.

I'd like to encourage all of you to go out

and help push the boundaries of science,

even if it may at first seem
as mysterious to you as a black hole.

Thank you.

(Applause)