sponsored links
TED2014

Andrew Connolly: What's the next window into our universe?

March 19, 2014

Big Data is everywhere — even the skies. In an informative talk, astronomer Andrew Connolly shows how large amounts of data are being collected about our universe, recording it in its ever-changing moods. Just how do scientists capture so many images at scale? It starts with a giant telescope …

Andrew Connolly - Astronomer
Andrew Connolly is helping to build the Large Synoptic Survey Telescope -- as well as tools to handle the massive datasets it will send our way. Full bio

sponsored links
Double-click the English subtitles below to play the video.
So in 1781, an English composer,
00:13
technologist and astronomer called William Herschel
00:16
noticed an object on the sky that
00:19
didn't quite move the way the rest of the stars did.
00:21
And Herschel's recognition
that something was different,
00:24
that something wasn't quite right,
00:27
was the discovery of a planet,
00:29
the planet Uranus,
00:31
a name that has entertained
00:33
countless generations of children,
00:34
but a planet that overnight
00:37
doubled the size of our known solar system.
00:40
Just last month, NASA announced the discovery
00:42
of 517 new planets
00:44
in orbit around nearby stars,
00:46
almost doubling overnight the number of planets
00:48
we know about within our galaxy.
00:51
So astronomy is constantly being transformed by this
00:53
capacity to collect data,
00:56
and with data almost doubling every year,
00:58
within the next two decades, me may even
01:01
reach the point for the first time in history
01:02
where we've discovered the majority of the galaxies
01:05
within the universe.
01:08
But as we enter this era of big data,
01:09
what we're beginning to find is there's a difference
01:12
between more data being just better
01:14
and more data being different,
01:17
capable of changing the questions we want to ask,
01:19
and this difference is not about
how much data we collect,
01:22
it's whether those data open new windows
01:25
into our universe,
01:27
whether they change the way we view the sky.
01:28
So what is the next window into our universe?
01:31
What is the next chapter for astronomy?
01:34
Well, I'm going to show you some
of the tools and the technologies
01:37
that we're going to develop over the next decade,
01:40
and how these technologies,
01:42
together with the smart use of data,
01:44
may once again transform astronomy
01:46
by opening up a window into our universe,
01:49
the window of time.
01:51
Why time? Well, time is about origins,
01:53
and it's about evolution.
01:55
The origins of our solar system,
01:57
how our solar system came into being,
01:58
is it unusual or special in any way?
02:01
About the evolution of our universe.
02:04
Why our universe is continuing to expand,
02:06
and what is this mysterious dark energy
02:09
that drives that expansion?
02:11
But first, I want to show you how technology
02:14
is going to change the way we view the sky.
02:16
So imagine if you were sitting
02:19
in the mountains of northern Chile
02:21
looking out to the west
02:23
towards the Pacific Ocean
02:24
a few hours before sunrise.
02:26
This is the view of the night sky that you would see,
02:29
and it's a beautiful view,
02:32
with the Milky Way just peeking out over the horizon.
02:34
but it's also a static view,
02:37
and in many ways, this is the
way we think of our universe:
02:39
eternal and unchanging.
02:42
But the universe is anything but static.
02:44
It constantly changes on timescales of seconds
02:46
to billions of years.
02:48
Galaxies merge, they collide
02:50
at hundreds of thousands of miles per hour.
02:52
Stars are born, they die,
02:55
they explode in these extravagant displays.
02:57
In fact, if we could go back
03:00
to our tranquil skies above Chile,
03:01
and we allow time to move forward
03:04
to see how the sky might change over the next year,
03:06
the pulsations that you see
03:11
are supernovae, the final remnants of a dying star
03:13
exploding, brightening and then fading from view,
03:17
each one of these supernovae
03:21
five billion times the brightness of our sun,
03:23
so we can see them to great distances
03:26
but only for a short amount of time.
03:28
Ten supernova per second explode somewhere
03:31
in our universe.
03:33
If we could hear it,
03:35
it would be popping like a bag of popcorn.
03:36
Now, if we fade out the supernovae,
03:40
it's not just brightness that changes.
03:43
Our sky is in constant motion.
03:46
This swarm of objects you
see streaming across the sky
03:49
are asteroids as they orbit our sun,
03:52
and it's these changes and the motion
03:54
and it's the dynamics of the system
03:56
that allow us to build our models for our universe,
03:59
to predict its future and to explain its past.
04:01
But the telescopes we've used over the last decade
04:05
are not designed to capture the data at this scale.
04:08
The Hubble Space Telescope:
04:12
for the last 25 years it's been producing
04:14
some of the most detailed views
04:16
of our distant universe,
04:18
but if you tried to use the Hubble to create an image
04:20
of the sky, it would take 13 million individual images,
04:22
about 120 years to do this just once.
04:27
So this is driving us to new technologies
04:30
and new telescopes,
04:33
telescopes that can go faint
04:35
to look at the distant universe
04:36
but also telescopes that can go wide
04:38
to capture the sky as rapidly as possible,
04:41
telescopes like the Large Synoptic Survey Telescope,
04:43
or the LSST,
04:47
possibly the most boring name ever
04:49
for one of the most fascinating experiments
04:51
in the history of astronomy,
04:53
in fact proof, if you should need it,
04:55
that you should never allow
a scientist or an engineer
04:57
to name anything, not even your children.
(Laughter)
05:00
We're building the LSST.
05:06
We expect it to start taking data
by the end of this decade.
05:07
I'm going to show you how we think
05:11
it's going to transform
our views of the universe,
05:12
because one image from the LSST
05:16
is equivalent to 3,000 images
05:18
from the Hubble Space Telescope,
05:21
each image three and a half degrees on the sky,
05:23
seven times the width of the full moon.
05:26
Well, how do you capture an image at this scale?
05:29
Well, you build the largest digital camera in history,
05:31
using the same technology you find
in the cameras in your cell phone
05:35
or in the digital cameras you
can buy in the High Street,
05:38
but now at a scale that is five and a half feet across,
05:42
about the size of a Volkswagen Beetle,
05:45
where one image is three billion pixels.
05:48
So if you wanted to look at an image
05:51
in its full resolution, just a single LSST image,
05:52
it would take about 1,500
high-definition TV screens.
05:55
And this camera will image the sky,
06:00
taking a new picture every 20 seconds,
06:03
constantly scanning the sky
06:06
so every three nights, we'll get a completely new view
06:08
of the skies above Chile.
06:11
Over the mission lifetime of this telescope,
06:13
it will detect 40 billion stars and galaxies,
06:16
and that will be for the first time
06:19
we'll have detected more objects in our universe
06:21
than people on the Earth.
06:24
Now, we can talk about this
06:26
in terms of terabytes and petabytes
06:28
and billions of objects,
06:30
but a way to get a sense of the amount of data
06:31
that will come off this camera
06:33
is that it's like playing every TED Talk ever recorded
06:35
simultaneously, 24 hours a day,
06:40
seven days a week, for 10 years.
06:43
And to process this data means
06:46
searching through all of those talks
06:48
for every new idea and every new concept,
06:50
looking at each part of the video
06:52
to see how one frame may have changed
06:54
from the next.
06:56
And this is changing the way that we do science,
06:58
changing the way that we do astronomy,
07:00
to a place where software and algorithms
07:02
have to mine through this data,
07:05
where the software is as critical to the science
07:07
as the telescopes and the
cameras that we've built.
07:10
Now, thousands of discoveries
07:14
will come from this project,
07:16
but I'm just going to tell you about two
07:18
of the ideas about origins and evolution
07:20
that may be transformed by our access
07:22
to data at this scale.
07:24
In the last five years, NASA has discovered
07:27
over 1,000 planetary systems
07:29
around nearby stars,
07:32
but the systems we're finding
07:34
aren't much like our own solar system,
07:36
and one of the questions we face is
07:38
is it just that we haven't been looking hard enough
07:40
or is there something special or unusual
07:42
about how our solar system formed?
07:44
And if we want to answer that question,
07:46
we have to know and understand
07:48
the history of our solar system in detail,
07:50
and it's the details that are crucial.
07:53
So now, if we look back at the sky,
07:55
at our asteroids that were streaming across the sky,
07:59
these asteroids are like the
debris of our solar system.
08:02
The positions of the asteroids
08:06
are like a fingerprint of an earlier time
08:08
when the orbits of Neptune and Jupiter
08:10
were much closer to the sun,
08:12
and as these giant planets migrated
through our solar system,
08:14
they were scattering the asteroids in their wake.
08:18
So studying the asteroids
08:21
is like performing forensics,
08:22
performing forensics on our solar system,
08:25
but to do this, we need distance,
08:27
and we get the distance from the motion,
08:30
and we get the motion because of our access to time.
08:32
So what does this tell us?
08:36
Well, if you look at the little yellow asteroids
08:38
flitting across the screen,
08:40
these are the asteroids that are moving fastest,
08:43
because they're closest to us, closest to Earth.
08:45
These are the asteroids we may one day
08:48
send spacecraft to, to mine them for minerals,
08:50
but they're also the asteroids that may one day
08:53
impact the Earth,
08:55
like happened 60 million years ago
08:57
with the extinction of the dinosaurs,
08:58
or just at the beginning of the last century,
09:01
when an asteroid wiped out
09:03
almost 1,000 square miles of Siberian forest,
09:04
or even just last year, as one burnt up over Russia,
09:08
releasing the energy of a small nuclear bomb.
09:11
So studying the forensics of our solar system
09:14
doesn't just tell us about the past,
09:18
it can also predict the future,
including our future.
09:20
Now when we get distance,
09:26
we get to see the asteroids
in their natural habitat,
09:28
in orbit around the sun.
09:32
So every point in this visualization that you can see
09:33
is a real asteroid.
09:36
Its orbit has been calculated
from its motion across the sky.
09:39
The colors reflect the composition of these asteroids,
09:43
dry and stony in the center,
09:46
water-rich and primitive towards the edge,
09:48
water-rich asteroids which may have seeded
09:51
the oceans and the seas that we find on our planet
09:53
when they bombarded the
Earth at an earlier time.
09:57
Because the LSST will be able to go faint
10:02
and not just wide,
10:04
we will be able to see these asteroids
10:06
far beyond the inner part of our solar system,
10:08
to asteroids beyond the
orbits of Neptune and Mars,
10:11
to comets and asteroids that may exist
10:15
almost a light year from our sun.
10:17
And as we increase the detail of this picture,
10:20
increasing the detail by factors of 10 to 100,
10:23
we will be able to answer questions such as,
10:26
is there evidence for planets
outside the orbit of Neptune,
10:29
to find Earth-impacting asteroids
10:32
long before they're a danger,
10:35
and to find out whether, maybe,
10:37
our sun formed on its own or in a cluster of stars,
10:39
and maybe it's this sun's stellar siblings
10:42
that influenced the formation of our solar system,
10:45
and maybe that's one of the reasons why
solar systems like ours seem to be so rare.
10:49
Now, distance and changes in our universe —
10:54
distance equates to time,
10:59
as well as changes on the sky.
11:03
Every foot of distance you look away,
11:05
or every foot of distance an object is away,
11:08
you're looking back about a
billionth of a second in time,
11:10
and this idea or this notion of looking back in time
11:14
has revolutionized our ideas about the universe,
11:16
not once but multiple times.
11:19
The first time was in 1929,
11:21
when an astronomer called Edwin Hubble
11:24
showed that the universe was expanding,
11:26
leading to the ideas of the Big Bang.
11:28
And the observations were simple:
11:31
just 24 galaxies
11:34
and a hand-drawn picture.
11:36
But just the idea that the more distant a galaxy,
11:41
the faster it was receding,
11:45
was enough to give rise to modern cosmology.
11:47
A second revolution happened 70 years later,
11:51
when two groups of astronomers showed
11:53
that the universe wasn't just expanding,
11:55
it was accelerating,
11:58
a surprise like throwing up a ball into the sky
11:59
and finding out the higher that it gets,
12:02
the faster it moves away.
12:05
And they showed this
12:07
by measuring the brightness of supernovae,
12:08
and how the brightness of the supernovae
12:11
got fainter with distance.
12:13
And these observations were more complex.
12:15
They required new technologies and new telescopes,
12:17
because the supernovae were in galaxies
12:20
that were 2,000 times more distant
12:24
than the ones used by Hubble.
12:26
And it took three years to find just 42 supernovae,
12:29
because a supernova only explodes
12:34
once every hundred years within a galaxy.
12:36
Three years to find 42 supernovae
12:39
by searching through tens of thousands of galaxies.
12:41
And once they'd collected their data,
12:45
this is what they found.
12:47
Now, this may not look impressive,
12:51
but this is what a revolution in physics looks like:
12:54
a line predicting the brightness of a supernova
12:58
11 billion light years away,
13:00
and a handful of points that don't quite fit that line.
13:02
Small changes give rise to big consequences.
13:06
Small changes allow us to make discoveries,
13:10
like the planet found by Herschel.
13:13
Small changes turn our understanding
13:16
of the universe on its head.
13:18
So 42 supernovae, slightly too faint,
13:21
meaning slightly further away,
13:24
requiring that a universe must not just be expanding,
13:26
but this expansion must be accelerating,
13:29
revealing a component of our universe
13:33
which we now call dark energy,
13:35
a component that drives this expansion
13:37
and makes up 68 percent of the energy budget
13:40
of our universe today.
13:43
So what is the next revolution likely to be?
13:46
Well, what is dark energy and why does it exist?
13:50
Each of these lines shows a different model
13:53
for what dark energy might be,
13:55
showing the properties of dark energy.
13:58
They all are consistent with the 42 points,
14:00
but the ideas behind these lines
14:04
are dramatically different.
14:06
Some people think about a dark energy
14:08
that changes with time,
14:11
or whether the properties of the dark energy
14:12
are different depending on where you look on the sky.
14:15
Others make differences and changes
14:17
to the physics at the sub-atomic level.
14:19
Or, they look at large scales
14:22
and change how gravity and general relativity work,
14:25
or they say our universe is just one of many,
14:29
part of this mysterious multiverse,
14:31
but all of these ideas, all of these theories,
14:34
amazing and admittedly some of them a little crazy,
14:37
but all of them consistent with our 42 points.
14:41
So how can we hope to make sense of this
14:45
over the next decade?
14:47
Well, imagine if I gave you a pair of dice,
14:49
and I said you wanted to see whether those dice
14:52
were loaded or fair.
14:54
One roll of the dice would tell you very little,
14:56
but the more times you rolled them,
14:59
the more data you collected,
15:01
the more confident you would become,
15:03
not just whether they're loaded or fair,
15:05
but by how much, and in what way.
15:08
It took three years to find just 42 supernovae
15:12
because the telescopes that we built
15:16
could only survey a small part of the sky.
15:19
With the LSST, we get a completely new view
15:22
of the skies above Chile every three nights.
15:25
In its first night of operation,
15:29
it will find 10 times the number of supernovae
15:31
used in the discovery of dark energy.
15:34
This will increase by 1,000
15:37
within the first four months:
15:39
1.5 million supernovae by the end of its survey,
15:42
each supernova a roll of the dice,
15:46
each supernova testing which theories of dark energy
15:50
are consistent, and which ones are not.
15:53
And so, by combining these supernova data
15:57
with other measures of cosmology,
16:01
we'll progressively rule out the different ideas
16:03
and theories of dark energy
16:06
until hopefully at the end of this survey around 2030,
16:08
we would expect to hopefully see
16:15
a theory for our universe,
16:18
a fundamental theory for the physics of our universe,
16:20
to gradually emerge.
16:23
Now, in many ways, the questions that I posed
16:26
are in reality the simplest of questions.
16:29
We may not know the answers,
16:33
but we at least know how to ask the questions.
16:35
But if looking through tens of thousands of galaxies
16:39
revealed 42 supernovae that turned
16:42
our understanding of the universe on its head,
16:45
when we're working with billions of galaxies,
16:48
how many more times are we going to find
16:51
42 points that don't quite match what we expect?
16:53
Like the planet found by Herschel
16:59
or dark energy
17:01
or quantum mechanics or general relativity,
17:04
all ideas that came because the data
17:08
didn't quite match what we expected.
17:10
What's so exciting about the next decade of data
17:13
in astronomy is,
17:17
we don't even know how many answers
17:18
are out there waiting,
17:21
answers about our origins and our evolution.
17:22
How many answers are out there
17:26
that we don't even know the questions
17:27
that we want to ask?
17:31
Thank you.
17:33
(Applause)
17:35

sponsored links

Andrew Connolly - Astronomer
Andrew Connolly is helping to build the Large Synoptic Survey Telescope -- as well as tools to handle the massive datasets it will send our way.

Why you should listen
Andrew Connolly's research focuses on understanding the evolution of our universe, by studying how structure forms and evolves on small and large scales -- from the search for asteroids to the clustering of distant galaxies. He's a ten-year veteran of the Large Synoptic Sky Survey, and is now prepping for the unprecedented data streams we could expect from the under-construction Large Synoptic Survey Telescope.
 
Set on an 8,800-foot peak in northern Chile, the LSST will have an 8.4-meter primary mirror, a 10-square-degree field of view and a 3.2 gigapixel camera. It will survey half the sky every three nights, creating about 100 terabytes of data every week. Astronomers, Connolly suggests, will need wholly new tools to wrangle this amount of data -- so he has been helping bring together computer scientists, statisticians and astronomers to develop scalable algorithms for processing massive data streams.
 
On sabbatical from the University of Washington, Connolly led the development of Google Sky, and he's now working with Microsoft to develop affordable digital planetariums.
sponsored links

If you need translations, you can install "Google Translate" extension into your Chrome Browser.
Furthermore, you can change playback rate by installing "Video Speed Controller" extension.

Data provided by TED.

This website is owned and operated by Tokyo English Network.
The developer's blog is here.