0:00:01.436,0:00:03.296
In the movie "Interstellar,"

0:00:03.320,0:00:06.647
we get an up-close look[br]at a supermassive black hole.

0:00:06.671,0:00:08.814
Set against a backdrop of bright gas,

0:00:08.838,0:00:10.956
the black hole's massive[br]gravitational pull

0:00:10.980,0:00:12.415
bends light into a ring.

0:00:12.439,0:00:14.548
However, this isn't a real photograph,

0:00:14.572,0:00:16.358
but a computer graphic rendering --

0:00:16.382,0:00:19.772
an artistic interpretation[br]of what a black hole might look like.

0:00:20.401,0:00:21.567
A hundred years ago,

0:00:21.591,0:00:25.192
Albert Einstein first published[br]his theory of general relativity.

0:00:25.216,0:00:26.655
In the years since then,

0:00:26.679,0:00:29.652
scientists have provided[br]a lot of evidence in support of it.

0:00:29.676,0:00:32.760
But one thing predicted[br]from this theory, black holes,

0:00:32.784,0:00:35.134
still have not been directly observed.

0:00:35.158,0:00:38.364
Although we have some idea[br]as to what a black hole might look like,

0:00:38.388,0:00:41.167
we've never actually taken[br]a picture of one before.

0:00:41.191,0:00:45.470
However, you might be surprised to know[br]that that may soon change.

0:00:45.494,0:00:49.658
We may be seeing our first picture[br]of a black hole in the next couple years.

0:00:49.682,0:00:53.640
Getting this first picture will come down[br]to an international team of scientists,

0:00:53.664,0:00:55.231
an Earth-sized telescope

0:00:55.255,0:00:58.087
and an algorithm that puts together[br]the final picture.

0:00:58.111,0:01:01.639
Although I won't be able to show you[br]a real picture of a black hole today,

0:01:01.663,0:01:04.574
I'd like to give you a brief glimpse[br]into the effort involved

0:01:04.598,0:01:06.211
in getting that first picture.

0:01:07.477,0:01:08.914
My name is Katie Bouman,

0:01:08.938,0:01:11.504
and I'm a PhD student at MIT.

0:01:11.528,0:01:13.555
I do research in a computer science lab

0:01:13.579,0:01:16.877
that works on making computers[br]see through images and video.

0:01:16.901,0:01:19.063
But although I'm not an astronomer,

0:01:19.087,0:01:20.372
today I'd like to show you

0:01:20.396,0:01:23.299
how I've been able to contribute[br]to this exciting project.

0:01:23.323,0:01:26.154
If you go out past[br]the bright city lights tonight,

0:01:26.178,0:01:28.614
you may just be lucky enough[br]to see a stunning view

0:01:28.638,0:01:30.131
of the Milky Way Galaxy.

0:01:30.155,0:01:32.617
And if you could zoom past[br]millions of stars,

0:01:32.641,0:01:36.396
26,000 light-years toward the heart[br]of the spiraling Milky Way,

0:01:36.420,0:01:39.941
we'd eventually reach[br]a cluster of stars right at the center.

0:01:39.965,0:01:43.171
Peering past all the galactic dust[br]with infrared telescopes,

0:01:43.195,0:01:47.062
astronomers have watched these stars[br]for over 16 years.

0:01:47.086,0:01:50.675
But it's what they don't see[br]that is the most spectacular.

0:01:50.699,0:01:53.765
These stars seem to orbit[br]an invisible object.

0:01:53.789,0:01:56.112
By tracking the paths of these stars,

0:01:56.136,0:01:57.430
astronomers have concluded

0:01:57.454,0:02:00.583
that the only thing small and heavy[br]enough to cause this motion

0:02:00.607,0:02:02.575
is a supermassive black hole --

0:02:02.599,0:02:06.777
an object so dense that it sucks up[br]anything that ventures too close --

0:02:06.801,0:02:08.295
even light.

0:02:08.319,0:02:11.380
But what happens if we were[br]to zoom in even further?

0:02:11.404,0:02:16.137
Is it possible to see something[br]that, by definition, is impossible to see?

0:02:16.719,0:02:19.963
Well, it turns out that if we were[br]to zoom in at radio wavelengths,

0:02:19.987,0:02:21.669
we'd expect to see a ring of light

0:02:21.693,0:02:24.104
caused by the gravitational[br]lensing of hot plasma

0:02:24.128,0:02:25.957
zipping around the black hole.

0:02:25.981,0:02:27.141
In other words,

0:02:27.165,0:02:30.336
the black hole casts a shadow[br]on this backdrop of bright material,

0:02:30.360,0:02:32.202
carving out a sphere of darkness.

0:02:32.226,0:02:35.565
This bright ring reveals[br]the black hole's event horizon,

0:02:35.589,0:02:37.989
where the gravitational pull[br]becomes so great

0:02:38.013,0:02:39.639
that not even light can escape.

0:02:39.663,0:02:42.522
Einstein's equations predict[br]the size and shape of this ring,

0:02:42.546,0:02:45.754
so taking a picture of it[br]wouldn't only be really cool,

0:02:45.778,0:02:48.396
it would also help to verify[br]that these equations hold

0:02:48.420,0:02:50.886
in the extreme conditions[br]around the black hole.

0:02:50.910,0:02:53.468
However, this black hole[br]is so far away from us,

0:02:53.492,0:02:56.590
that from Earth, this ring appears[br]incredibly small --

0:02:56.614,0:03:00.204
the same size to us as an orange[br]on the surface of the moon.

0:03:00.758,0:03:03.582
That makes taking a picture of it[br]extremely difficult.

0:03:04.645,0:03:05.947
Why is that?

0:03:06.512,0:03:09.700
Well, it all comes down[br]to a simple equation.

0:03:09.724,0:03:12.140
Due to a phenomenon called diffraction,

0:03:12.164,0:03:13.519
there are fundamental limits

0:03:13.543,0:03:16.213
to the smallest objects[br]that we can possibly see.

0:03:16.789,0:03:20.461
This governing equation says[br]that in order to see smaller and smaller,

0:03:20.485,0:03:23.072
we need to make our telescope[br]bigger and bigger.

0:03:23.096,0:03:26.165
But even with the most powerful[br]optical telescopes here on Earth,

0:03:26.189,0:03:28.608
we can't even get close[br]to the resolution necessary

0:03:28.632,0:03:30.830
to image on the surface of the moon.

0:03:30.854,0:03:34.471
In fact, here I show one of the highest[br]resolution images ever taken

0:03:34.495,0:03:35.892
of the moon from Earth.

0:03:35.916,0:03:38.473
It contains roughly 13,000 pixels,

0:03:38.497,0:03:42.547
and yet each pixel would contain[br]over 1.5 million oranges.

0:03:43.396,0:03:45.368
So how big of a telescope do we need

0:03:45.392,0:03:48.157
in order to see an orange[br]on the surface of the moon

0:03:48.181,0:03:50.395
and, by extension, our black hole?

0:03:50.419,0:03:52.759
Well, it turns out[br]that by crunching the numbers,

0:03:52.783,0:03:55.393
you can easily calculate[br]that we would need a telescope

0:03:55.417,0:03:56.810
the size of the entire Earth.

0:03:56.834,0:03:57.858
(Laughter)

0:03:57.882,0:04:00.001
If we could build[br]this Earth-sized telescope,

0:04:00.025,0:04:02.950
we could just start to make out[br]that distinctive ring of light

0:04:02.974,0:04:05.157
indicative of the black[br]hole's event horizon.

0:04:05.181,0:04:08.099
Although this picture wouldn't contain[br]all the detail we see

0:04:08.123,0:04:09.629
in computer graphic renderings,

0:04:09.653,0:04:11.952
it would allow us to safely get[br]our first glimpse

0:04:11.976,0:04:14.463
of the immediate environment[br]around a black hole.

0:04:14.487,0:04:16.100
However, as you can imagine,

0:04:16.124,0:04:19.748
building a single-dish telescope[br]the size of the Earth is impossible.

0:04:19.772,0:04:21.659
But in the famous words of Mick Jagger,

0:04:21.683,0:04:23.474
"You can't always get what you want,

0:04:23.498,0:04:25.685
but if you try sometimes,[br]you just might find

0:04:25.709,0:04:26.924
you get what you need."

0:04:26.948,0:04:29.412
And by connecting telescopes[br]from around the world,

0:04:29.436,0:04:32.974
an international collaboration[br]called the Event Horizon Telescope

0:04:32.998,0:04:36.107
is creating a computational telescope[br]the size of the Earth,

0:04:36.131,0:04:37.668
capable of resolving structure

0:04:37.692,0:04:39.891
on the scale of a black[br]hole's event horizon.

0:04:39.915,0:04:43.302
This network of telescopes is scheduled[br]to take its very first picture

0:04:43.326,0:04:45.141
of a black hole next year.

0:04:45.165,0:04:48.503
Each telescope in the worldwide[br]network works together.

0:04:48.527,0:04:51.239
Linked through the precise timing[br]of atomic clocks,

0:04:51.263,0:04:53.920
teams of researchers at each[br]of the sites freeze light

0:04:53.944,0:04:56.906
by collecting thousands[br]of terabytes of data.

0:04:56.930,0:05:01.947
This data is then processed in a lab[br]right here in Massachusetts.

0:05:01.971,0:05:03.765
So how does this even work?

0:05:03.789,0:05:07.192
Remember if we want to see the black hole[br]in the center of our galaxy,

0:05:07.216,0:05:10.198
we need to build this impossibly large[br]Earth-sized telescope?

0:05:10.222,0:05:12.454
For just a second,[br]let's pretend we could build

0:05:12.478,0:05:14.320
a telescope the size of the Earth.

0:05:14.344,0:05:16.799
This would be a little bit[br]like turning the Earth

0:05:16.823,0:05:18.570
into a giant spinning disco ball.

0:05:18.594,0:05:20.794
Each individual mirror would collect light

0:05:20.818,0:05:23.415
that we could then combine[br]together to make a picture.

0:05:23.439,0:05:26.100
However, now let's say[br]we remove most of those mirrors

0:05:26.124,0:05:28.096
so only a few remained.

0:05:28.120,0:05:30.997
We could still try to combine[br]this information together,

0:05:31.021,0:05:33.014
but now there are a lot of holes.

0:05:33.038,0:05:37.411
These remaining mirrors represent[br]the locations where we have telescopes.

0:05:37.435,0:05:41.514
This is an incredibly small number[br]of measurements to make a picture from.

0:05:41.538,0:05:45.376
But although we only collect light[br]at a few telescope locations,

0:05:45.400,0:05:48.823
as the Earth rotates, we get to see[br]other new measurements.

0:05:48.847,0:05:52.666
In other words, as the disco ball spins,[br]those mirrors change locations

0:05:52.690,0:05:55.589
and we get to observe[br]different parts of the image.

0:05:55.613,0:05:59.631
The imaging algorithms we develop[br]fill in the missing gaps of the disco ball

0:05:59.655,0:06:02.688
in order to reconstruct[br]the underlying black hole image.

0:06:02.712,0:06:05.348
If we had telescopes located[br]everywhere on the globe --

0:06:05.372,0:06:07.313
in other words, the entire disco ball --

0:06:07.337,0:06:08.621
this would be trivial.

0:06:08.645,0:06:11.967
However, we only see a few samples,[br]and for that reason,

0:06:11.991,0:06:14.379
there are an infinite number[br]of possible images

0:06:14.403,0:06:17.367
that are perfectly consistent[br]with our telescope measurements.

0:06:17.391,0:06:20.407
However, not all images are created equal.

0:06:20.849,0:06:25.307
Some of those images look more like[br]what we think of as images than others.

0:06:25.331,0:06:28.553
And so, my role in helping to take[br]the first image of a black hole

0:06:28.577,0:06:31.509
is to design algorithms that find[br]the most reasonable image

0:06:31.533,0:06:33.755
that also fits the telescope measurements.

0:06:34.727,0:06:38.669
Just as a forensic sketch artist[br]uses limited descriptions

0:06:38.693,0:06:42.207
to piece together a picture using[br]their knowledge of face structure,

0:06:42.231,0:06:45.546
the imaging algorithms I develop[br]use our limited telescope data

0:06:45.570,0:06:49.892
to guide us to a picture that also[br]looks like stuff in our universe.

0:06:49.916,0:06:53.567
Using these algorithms,[br]we're able to piece together pictures

0:06:53.591,0:06:55.771
from this sparse, noisy data.

0:06:55.795,0:07:00.324
So here I show a sample reconstruction[br]done using simulated data,

0:07:00.348,0:07:02.281
when we pretend to point our telescopes

0:07:02.305,0:07:04.890
to the black hole[br]in the center of our galaxy.

0:07:04.914,0:07:09.369
Although this is just a simulation,[br]reconstruction such as this give us hope

0:07:09.393,0:07:12.846
that we'll soon be able to reliably take[br]the first image of a black hole

0:07:12.870,0:07:15.465
and from it, determine[br]the size of its ring.

0:07:16.118,0:07:19.317
Although I'd love to go on[br]about all the details of this algorithm,

0:07:19.341,0:07:21.515
luckily for you, I don't have the time.

0:07:21.539,0:07:23.540
But I'd still like[br]to give you a brief idea

0:07:23.564,0:07:25.866
of how we define[br]what our universe looks like,

0:07:25.890,0:07:30.356
and how we use this to reconstruct[br]and verify our results.

0:07:30.380,0:07:32.876
Since there are an infinite number[br]of possible images

0:07:32.900,0:07:35.265
that perfectly explain[br]our telescope measurements,

0:07:35.289,0:07:37.894
we have to choose[br]between them in some way.

0:07:37.918,0:07:39.756
We do this by ranking the images

0:07:39.780,0:07:42.614
based upon how likely they are[br]to be the black hole image,

0:07:42.638,0:07:45.120
and then choosing the one[br]that's most likely.

0:07:45.144,0:07:47.339
So what do I mean by this exactly?

0:07:47.862,0:07:49.840
Let's say we were trying to make a model

0:07:49.864,0:07:53.047
that told us how likely an image[br]were to appear on Facebook.

0:07:53.071,0:07:54.772
We'd probably want the model to say

0:07:54.796,0:07:58.353
it's pretty unlikely that someone[br]would post this noise image on the left,

0:07:58.377,0:08:00.796
and pretty likely that someone[br]would post a selfie

0:08:00.820,0:08:02.154
like this one on the right.

0:08:02.178,0:08:03.817
The image in the middle is blurry,

0:08:03.841,0:08:06.480
so even though it's more likely[br]we'd see it on Facebook

0:08:06.504,0:08:07.864
compared to the noise image,

0:08:07.888,0:08:10.848
it's probably less likely we'd see it[br]compared to the selfie.

0:08:10.872,0:08:13.162
But when it comes to images[br]from the black hole,

0:08:13.186,0:08:16.688
we're posed with a real conundrum:[br]we've never seen a black hole before.

0:08:16.712,0:08:19.003
In that case, what is a likely[br]black hole image,

0:08:19.027,0:08:21.965
and what should we assume[br]about the structure of black holes?

0:08:21.989,0:08:24.621
We could try to use images[br]from simulations we've done,

0:08:24.645,0:08:27.175
like the image of the black hole[br]from "Interstellar,"

0:08:27.199,0:08:30.137
but if we did this,[br]it could cause some serious problems.

0:08:30.161,0:08:33.541
What would happen[br]if Einstein's theories didn't hold?

0:08:33.565,0:08:37.526
We'd still want to reconstruct[br]an accurate picture of what was going on.

0:08:37.550,0:08:40.921
If we bake Einstein's equations[br]too much into our algorithms,

0:08:40.945,0:08:43.700
we'll just end up seeing[br]what we expect to see.

0:08:43.724,0:08:46.000
In other words,[br]we want to leave the option open

0:08:46.024,0:08:48.947
for there being a giant elephant[br]at the center of our galaxy.

0:08:48.971,0:08:50.028
(Laughter)

0:08:50.052,0:08:53.041
Different types of images have[br]very distinct features.

0:08:53.065,0:08:56.613
We can easily tell the difference[br]between black hole simulation images

0:08:56.637,0:08:58.913
and images we take[br]every day here on Earth.

0:08:58.937,0:09:02.041
We need a way to tell our algorithms[br]what images look like

0:09:02.065,0:09:05.314
without imposing one type[br]of image's features too much.

0:09:05.865,0:09:07.758
One way we can try to get around this

0:09:07.782,0:09:10.844
is by imposing the features[br]of different kinds of images

0:09:10.868,0:09:14.998
and seeing how the type of image we assume[br]affects our reconstructions.

0:09:15.712,0:09:19.203
If all images' types produce[br]a very similar-looking image,

0:09:19.227,0:09:21.284
then we can start to become more confident

0:09:21.308,0:09:25.481
that the image assumptions we're making[br]are not biasing this picture that much.

0:09:25.505,0:09:28.495
This is a little bit like[br]giving the same description

0:09:28.519,0:09:31.515
to three different sketch artists[br]from all around the world.

0:09:31.539,0:09:34.399
If they all produce[br]a very similar-looking face,

0:09:34.423,0:09:36.216
then we can start to become confident

0:09:36.240,0:09:39.856
that they're not imposing their own[br]cultural biases on the drawings.

0:09:39.880,0:09:43.195
One way we can try to impose[br]different image features

0:09:43.219,0:09:45.660
is by using pieces of existing images.

0:09:46.214,0:09:48.374
So we take a large collection of images,

0:09:48.398,0:09:51.116
and we break them down[br]into their little image patches.

0:09:51.140,0:09:55.425
We then can treat each image patch[br]a little bit like pieces of a puzzle.

0:09:55.449,0:09:59.727
And we use commonly seen puzzle pieces[br]to piece together an image

0:09:59.751,0:10:02.203
that also fits our telescope measurements.

0:10:03.040,0:10:06.783
Different types of images have[br]very distinctive sets of puzzle pieces.

0:10:06.807,0:10:09.613
So what happens when we take the same data

0:10:09.637,0:10:13.767
but we use different sets of puzzle pieces[br]to reconstruct the image?

0:10:13.791,0:10:18.557
Let's first start with black hole[br]image simulation puzzle pieces.

0:10:18.581,0:10:20.172
OK, this looks reasonable.

0:10:20.196,0:10:22.890
This looks like what we expect[br]a black hole to look like.

0:10:22.914,0:10:24.107
But did we just get it

0:10:24.131,0:10:27.445
because we just fed it little pieces[br]of black hole simulation images?

0:10:27.469,0:10:29.349
Let's try another set of puzzle pieces

0:10:29.373,0:10:31.882
from astronomical, non-black hole objects.

0:10:32.914,0:10:35.040
OK, we get a similar-looking image.

0:10:35.064,0:10:37.300
And then how about pieces[br]from everyday images,

0:10:37.324,0:10:40.109
like the images you take[br]with your own personal camera?

0:10:41.312,0:10:43.427
Great, we see the same image.

0:10:43.451,0:10:46.817
When we get the same image[br]from all different sets of puzzle pieces,

0:10:46.841,0:10:48.887
then we can start to become more confident

0:10:48.911,0:10:50.877
that the image assumptions we're making

0:10:50.901,0:10:53.822
aren't biasing the final[br]image we get too much.

0:10:53.846,0:10:57.099
Another thing we can do is take[br]the same set of puzzle pieces,

0:10:57.123,0:10:59.612
such as the ones derived[br]from everyday images,

0:10:59.636,0:11:03.236
and use them to reconstruct[br]many different kinds of source images.

0:11:03.260,0:11:04.531
So in our simulations,

0:11:04.555,0:11:08.330
we pretend a black hole looks like[br]astronomical non-black hole objects,

0:11:08.354,0:11:12.203
as well as everyday images like[br]the elephant in the center of our galaxy.

0:11:12.227,0:11:15.395
When the results of our algorithms[br]on the bottom look very similar

0:11:15.419,0:11:17.515
to the simulation's truth image on top,

0:11:17.539,0:11:20.885
then we can start to become[br]more confident in our algorithms.

0:11:20.909,0:11:22.776
And I really want to emphasize here

0:11:22.800,0:11:24.734
that all of these pictures were created

0:11:24.758,0:11:27.694
by piecing together little pieces[br]of everyday photographs,

0:11:27.718,0:11:29.933
like you'd take with your own[br]personal camera.

0:11:29.957,0:11:33.233
So an image of a black hole[br]we've never seen before

0:11:33.257,0:11:37.200
may eventually be created by piecing[br]together pictures we see all the time

0:11:37.224,0:11:39.969
of people, buildings,[br]trees, cats and dogs.

0:11:39.993,0:11:42.638
Imaging ideas like this[br]will make it possible for us

0:11:42.662,0:11:45.281
to take our very first pictures[br]of a black hole,

0:11:45.305,0:11:47.752
and hopefully, verify[br]those famous theories

0:11:47.776,0:11:50.197
on which scientists rely on a daily basis.

0:11:50.221,0:11:52.829
But of course, getting[br]imaging ideas like this working

0:11:52.853,0:11:56.175
would never have been possible[br]without the amazing team of researchers

0:11:56.199,0:11:58.086
that I have the privilege to work with.

0:11:58.110,0:11:59.273
It still amazes me

0:11:59.297,0:12:02.648
that although I began this project[br]with no background in astrophysics,

0:12:02.672,0:12:05.291
what we have achieved[br]through this unique collaboration

0:12:05.315,0:12:08.074
could result in the very first[br]images of a black hole.

0:12:08.098,0:12:10.796
But big projects like[br]the Event Horizon Telescope

0:12:10.820,0:12:13.634
are successful due to all[br]the interdisciplinary expertise

0:12:13.658,0:12:15.448
different people bring to the table.

0:12:15.472,0:12:17.178
We're a melting pot of astronomers,

0:12:17.202,0:12:19.434
physicists, mathematicians and engineers.

0:12:19.458,0:12:22.012
This is what will make it soon possible

0:12:22.036,0:12:24.889
to achieve something[br]once thought impossible.

0:12:24.913,0:12:27.169
I'd like to encourage all of you to go out

0:12:27.193,0:12:29.289
and help push the boundaries of science,

0:12:29.313,0:12:33.214
even if it may at first seem[br]as mysterious to you as a black hole.

0:12:33.238,0:12:34.412
Thank you.

0:12:34.436,0:12:36.833
(Applause)