0:00

I'm Dr Karl, coming to you from the lands

0:02

of the Gadigal people of the Eora Nation.

0:04

I acknowledge Aboriginal and Torres Strait Islander

0:07

peoples as the first Australians and

0:09

traditional custodians of the lands where we

0:11

live, learn and work.

0:14

G'day,

0:15

Dr Karl here, Shortlandism Sciences, part two

0:18

of The Early Universe Ran Slower,

0:20

with my wonderful colleague Professor Geraint Lewis.

0:22

Welcome back. Good morning, Karl. How are you doing? I

0:25

said Geraint wrong again. No, no, no. Have I lost

0:27

the proper way? Yeah, you're fine there, Karl.

0:29

So you were talking about the damped

0:31

drunkard's walk. Yes. So by the way, just

0:34

to give the audience time to speed up,

0:36

albatrosses find their food by something

0:38

different, a Levi walk, where

0:41

in a drunkard's walk it's always the same

0:43

size of step, but in a random

0:45

direction. In a Levi walk, the step

0:47

can be different. And this turns out to be more

0:50

efficient for them to find food. Okay, now

0:52

you've had a chance, beloved audience, we went

0:54

through supernovas and they're relatively

0:56

easy to deal with, with a clock, because

0:59

they were damp and down from their brightness

1:01

at a certain fixed rate. And in the early universe

1:03

it was happening earlier, and then you were saying, ah, but quasars,

1:07

they're really out there, but we're not seeing

1:09

it in quasars. And then you went into

1:11

random walks to take it away. Yeah, if you

1:13

look at the variations of quasars,

1:15

said it's up and down and up and down, etc. In

1:17

this new sample of data that the American

1:20

team released, they were able to characterize

1:23

the light curves and found something

1:25

that we could use for a tick, this timescale

1:27

for the damping of that random

1:29

walk, the drunkard's walk in the light curve. Out of the

1:32

millions of quasars, of which

1:35

we've characterized or numbered probably tens of thousands,

1:38

they focused on a couple of hundred. This

1:41

was at the end of 2022. And so

1:43

I had this sample of

1:45

quasars. And the question is, can

1:47

we find this expected

1:49

signature that the universe appears to run slower

1:52

when we look through our telescopes? So you need

1:54

a statistician, a Bayesian statistician. So

1:57

look up Bayesian statistics on Wikipedia.

1:59

out of the way. So you got your mate

2:02

who can help you with statistics, right? The key

2:04

problem that we have is I've got these

2:06

quasars. Some of them are big, some

2:09

of them are small in terms of their mass, they're not identical.

2:12

And there's another added complication is

2:14

that when I look through my telescope

2:17

at a quasar, I tend to look at it through

2:19

a filter. So I would look

2:21

at it in blue light, or green

2:24

light, or red light. If

2:26

I see red light today, this

2:28

doesn't mean that when it was emitted it was

2:30

red because the expansion of the universe stretches

2:33

the light. And the bigger the wavelengths,

2:35

the different the colour. Yeah. So I could

2:37

be looking at some object and I'd

2:40

be seeing material which is very close to the black

2:42

hole, so that could be varying rapidly.

2:44

Or I could be seeing material which

2:46

is further out where it's quieter

2:48

and would be varying less rapidly. There's

2:51

complications, there's always complications. But

2:54

with this sample what we were able to do is start to group

2:58

quasars together. You want to get something

3:00

you can deal with. Yeah. So we were able to group

3:03

them and so we said, oh, this

3:05

group, we expect their variability

3:07

to be very similar. The only

3:09

difference between them should be their

3:12

distance and their time. The expansion

3:14

of the universe, right? Some of them are nearby, some

3:16

of them are further away and the only difference that we should see

3:18

in their timescales is due to the expansion

3:21

of the universe, the time dilation effect.

3:23

Did I actually make a correct guess? You did.

3:26

For what's in my life? Yeah. Jeez.

3:28

Okay, go on. So we went through and we did this and

3:31

again, I'm personally a big fan of Bayesian statistics.

3:33

I do love them. I don't understand them. That's

3:36

another thing for another time. But it means that you

3:38

make your prediction

3:40

based on past events,

3:43

but more than that. And Bayes was a minister

3:45

who rode around on a horse in the 17 or 1800s or something. Yes.

3:49

So we had a very interesting notion about how you should treat data

3:52

and how you should test hypotheses,

3:55

which is what science is all about. So what

3:57

we did is we said, oh, What

4:00

if we assume that there is no cosmological

4:02

time dilation? What does the data say

4:04

about that? What if we say that there is? What

4:06

does the data say about that? So we proposed

4:09

our hypotheses and said, does

4:11

the data support that or not support

4:13

it? And what we did is we turned the handle.

4:16

And again, it's a topic

4:18

I love, but clearly it was not

4:20

for the discussion here today. As we

4:23

did our comparison, and what popped out is

4:25

that the best hypothesis to

4:27

explain the data in front of us is

4:29

that there's a cosmic time dilation, that

4:32

the more distant quasars are varying

4:35

more slowly than the nearby quasars. Is

4:37

it easy to describe what you're looking at? No.

4:41

Unless you've done basic statistics. It's

4:43

like looking at the stock market, right? Because

4:45

you can zoom in and zoom in, and there's always

4:47

variability going on. The time scale

4:49

is hidden in that variability, but

4:51

it's there. We can get a handle on statistics. People should read

4:53

your paper. And by the way, I'd like to point out that

4:56

even your brain wasn't enough, that you had to team

4:58

up with somebody who specialised in that subset

5:01

of knowledge called Bayesian statistics, and

5:03

then subspecialise again related to

5:05

astronomical events, and then subspecialise again related

5:08

to quasars. Very rarely is science

5:10

done by individuals anymore, right? You work with people

5:12

that have expertise, and you bring their expertise

5:15

together to answer the big questions.

5:17

There was somebody about 20 or 30 years ago who

5:20

looked at quasars and said, nah. I found

5:22

their paper. What did they do wrong? Or did

5:24

they have Bayesian stuff happening? This is an astronomer

5:26

at Edinburgh called Mike Hawkins. And Mike

5:29

was collecting data actually in Australia.

5:32

So just aside Cunha Barabran,

5:34

he's signed in Springs Observatory. And there's a

5:36

special telescope there called the Schmidt Telescope.

5:39

And so Schmidt telescopes are famous because in

5:41

the old days, they allowed you to get

5:44

in-focused view of a large chunk

5:46

of sky. So it'd be sort of like a wide-angle-y

5:49

thing. Yeah. So large chunk

5:51

means a couple of times bigger than the full moon, but

5:53

not that much bigger. OK. Moon is half a degree.

5:56

Yeah. So you could collect data

5:58

on photographic films. because that's what

6:00

they had back then. And then those

6:02

films were scanned by machines back

6:04

in the UK, one in Cambridge, one in Edinburgh. And

6:07

they could identify quasars in these scans.

6:10

So Mike had been collecting data for a number of years

6:13

and he had identified quasars and so he had

6:15

like 20 to 30 years worth of light

6:18

curves, the variations that you see in quasars.

6:21

But they were poorly

6:24

sampled. And what I mean by that is

6:26

that he was getting data every once a

6:28

year or once every six months, etc.

6:30

So if you look over 30 year span, you see

6:33

a bunch of points, but you've got gaps,

6:35

which could be fatal if you look at something like the stock market

6:38

to make money quickly. Exactly. If you were trying

6:40

to do the stock market and you only got a stock price

6:42

once every year, and you were trying to work

6:44

out the behavior, then there's lots of stuff going on in

6:46

between. Firstly, the data wasn't

6:49

very well sampled. But also he was

6:51

working at a time before people had gotten good

6:53

enough data to even get a statistical

6:55

handle on how quasars varied.

6:58

Ah, so he was a very early person doing the best

7:00

he could with what was available back then. Basically,

7:02

he did the statistical tests that he could

7:04

do with the data. And it came back and it was

7:06

like, no, it probably was

7:09

just is not the handle in the data to

7:11

get a measure of the time scale. So

7:13

it's just not well sampled enough. Okay,

7:16

before we dive into the other objects, so I want to talk

7:18

about, did you use a lot of computing

7:20

power or was it relatively minor?

7:22

This was relatively minor in today's

7:25

world. So one of Brendan's things

7:27

that he's done in his research, he's written a special

7:29

program called D-NEST4.

7:32

D? D-NEST4, Diffusive

7:34

Nested Sampling Version 4. Oh,

7:36

we have three was. Yeah,

7:39

okay. And so what this does is it

7:41

does your Bayesian analysis, it goes around probability

7:44

space and checks your hypothesis, etc.

7:47

And it turns out that for this size of problem,

7:49

we can actually do this on a relatively

7:52

small computer. So I did most of the calculations on my

7:54

desktop computer, which is a computer

7:57

compared to something 15 years ago. Yes, yes.

8:00

The bigger problems we do need to do

8:02

the calculations on actual supercomputers

8:04

that exist today because you

8:06

essentially have to interrogate the

8:08

data many millions of times

8:10

with your hypotheses and then from

8:13

that you draw out the answers. Okay getting back to

8:15

your paper the detection of the cosmological

8:17

time dilation of high redshift quasars,

8:20

you and your mate Dr Brendan J. Brewer

8:23

managed to do something with the complicated things called

8:26

quasars. That

8:28

leads us to other objects gamma

8:30

rays which are very energetic

8:33

and fast radio bursters which

8:35

I don't understand either. So normally astronomers

8:37

look at the sky right and you get your images and you do

8:40

your work and all that kind of stuff. One of the

8:42

things that we haven't really looked at in detail

8:44

is how much the sky changes on a

8:47

relatively short time scale. Things

8:49

that vary quickly are still relatively

8:52

new. So gamma ray bursts have been known for

8:54

a little while. The Vila satellites

8:56

by the Americans discovered something that the

8:58

South Africans were setting off nuclear weapons. Then

9:01

they finally discovered that there were gamma rays coming from space

9:03

not just from nuclear weapons. Exactly. And

9:06

so people found that there are gamma ray bursts. Things

9:08

that explode release these high energy

9:10

gamma rays. Huge amount of energy. Huge amount of energy

9:12

and we found that they are at huge distances,

9:14

cosmological distances. Yeah this was bothering

9:17

the American spy satellite. Billions of layers

9:19

and more recently we discovered that when we look

9:21

at the radio sky we find something similar,

9:24

fast radio bursts. Something goes

9:26

bang and releases radio waves and

9:28

then those radio waves spill out into the universe

9:30

again. They appear to be coming from cosmological

9:33

distances. So these must be powered

9:35

by very energetic, probably

9:37

quite violent events. So like a gamma

9:39

ray burst might be neutron stars colliding.

9:42

Oh yeah. And fast radio bursts must

9:44

be something similar as well. There is

9:46

one hypothesis that the gamma ray bursts might

9:48

be neutron stars colliding. Yeah. And

9:50

be able to tie them in with gravitational

9:53

wave detection if the frequency wavelength is

9:55

right? Yes. So I mean this is

9:57

what people are trying to do. When they detect a gravitational wave

9:59

they... They take a look to see if they can pick up light

10:02

from those objects, so optical light or gamma

10:05

rays or radio waves. So by light, you

10:07

as an astronomer mean anything from gamma rays

10:09

all the way up to the other end? Yes,

10:11

electromagnetic radiation. So again, we

10:13

don't quite understand the physics of what's powering

10:16

these, but we know that they go bang and we

10:18

know that they're at large distances. So

10:20

they too should show this cosmic

10:22

time dilation. A place where

10:25

we are is playing this same game

10:27

again. How do I know that this gamma

10:29

ray burst and that gamma ray burst are

10:31

intrinsically similar objects? Because

10:34

Type 1A supernovas are what we call a standard

10:36

candle and they're essentially the same. You're

10:38

saying there could be many different types of gamma ray burst,

10:41

many different types of fast

10:43

radio burst. Yes. Imagine

10:45

that you've got one fast radio burst that's created

10:47

by an object of this mass and an object

10:49

of that mass interacting. But you change the

10:51

masses and you still get a fast radio burst, but maybe

10:54

it puts out twice as much power or

10:56

half as much power. It's a different time scale. By

10:58

the way, how much slower was the universe

11:00

running back then? With the measurements that we

11:03

made, instead of looking back on the half

11:05

the age of the universe, with these quasals,

11:07

because they're so bright, we looked back over 90% the

11:10

age of the universe. So how

11:12

many billion years after the Big Bang or how many billion

11:14

years back is that? A billion or so. After the

11:16

Big Bang. A billion. So

11:18

at three minutes you get the nuclei

11:20

cooling down. At three, 80,000 years

11:23

you get the electrons joining up and the

11:25

first stars are around 400 million. Is

11:27

that what we still think? Yes,

11:30

something like that. They keep pushing these numbers back. Stars

11:32

seem to form quite early in the universe. Thank

11:34

you. Just wonderful space telescope, I guess. Yes. Yes.

11:38

Okay. So at point two or three or four

11:40

of a billion, you get the first stars somewhere there and then you're

11:42

picking up stuff at a billion years. Between

11:45

a billion and two billion, roughly, how did

11:47

it get so big, so powerful, so fast? Well,

11:49

the universe was very different to the universe

11:51

today, right? It was a lot denser, so

11:53

material gathered together very, very quickly.

11:56

These massive stars, they lived their lives

11:58

very fast. They're born very massive

12:01

and they live for a few million years.

12:04

So like twice the mass of our sun or ten or

12:06

a hundred? A hundred times. A hundred.

12:08

That is big. A hundred times and pure

12:11

hydrogen and helium. Oh because there's no debris

12:13

from nuclear reactions in

12:15

stars making energy because they're the first

12:17

generation of stars. The universe

12:19

was a different place. Massive stars

12:22

could have been the seeds for the new black holes

12:24

because when they die they create black holes. Nuclear

12:27

could flow in very very quickly so you

12:29

can form these quasars in the very

12:31

very early universe. When we saw them

12:34

basically the time scales back

12:36

then from our viewpoint here the universe

12:38

was running at about a fifth the

12:40

speed that we see it today. If

12:43

we were magically able to go in a time machine

12:45

and a distance machine and go back then would we

12:47

notice? No because we'd be

12:50

obeying the laws of the universe. We'd be

12:52

looking at our watches just going one second per second

12:54

everything is fine but if I spied you through

12:56

my telescope and I could see your watch

12:58

I would see it tick five of my seconds

13:01

for your watch to tick one second.

13:03

Now I'm thinking of a question from

13:07

a nine year old intelligence student

13:09

at primary school can we use this to make a

13:11

time machine? We can use time

13:13

dilation to make a time machine but you

13:15

can't use the cosmological time dilation because it

13:17

would mean that it would have to go into the past anyway.

13:20

Right so they would need a time machine but

13:22

we can make a time machine using gravity. What?

13:25

Because another consequence of Einstein's equations

13:28

is that the rate at which clocks

13:30

tick is relative based upon where you are in a

13:32

gravitational field and we know this works because

13:34

GPS needs to correct for this. So

13:37

if you go down near a black hole

13:39

and hang out near the event horizon

13:41

the point of no return your

13:44

clock down there ticks really slowly

13:46

compared to a clock somewhere else hanging

13:48

out in the universe. So you can go

13:50

down into a black hole

13:52

hang out there and come back out. Hang around

13:55

to be precise hang out just outside

13:57

the event horizon. Definitely not inside. And

14:00

definitely not into the black hole itself which has zero size

14:02

which is inside the event horizon but you go very

14:05

close and you might hang out there for

14:07

months and the universe has aged

14:10

by ... We've all seen Interstellar, right? Come out

14:12

and his daughter's now an old woman. Yeah,

14:15

that kind of thing can happen. So you can time travel

14:17

into the future. Time travel into the future is easy.

14:20

Yep. Right? We haven't cracked time

14:22

travel into the past. Wasn't there

14:24

the one from John Wheeler where you get a neutron star

14:26

and you fashion it into a cylinder one kilometer in diameter

14:28

and 20 kilometers long and spin it at a thousand

14:31

rivers per second. The outside is moving at half the speed of light

14:33

and you go back in time to when it was created you go into a tidal

14:35

over around it.

14:36

Yes,

14:37

but that's an engineering problem, right?

14:39

A major engineering problem. So look, theoretically

14:42

we can work out time travel. That's definitely

14:44

true. The question about whether or not it's

14:47

ever physically realizable

14:49

is it something that we could ever build? We

14:52

don't know. We don't know. And of course

14:54

people have conjectures. They say time

14:56

travel is impossible because essentially I don't like

14:58

it. There was this famous story that Stephen

15:00

Hawking decided that there was no time travel because

15:02

he had a party and invited time

15:05

travelers from the future to come to his party and

15:07

nobody turned up. That's where he sent out invitations

15:09

saying, come to my party four weeks ago. Yeah. Maybe

15:12

time travel has got something better to do than hang out with

15:14

Stephen Hawking at a party, right? He did

15:16

like pizza apparently. This is one of the things that

15:19

I love about Einstein's general relativity.

15:22

Time is malleable. It is definitely

15:24

malleable. And we have the equations that tell us how

15:26

to shape and bend and warp it. What

15:29

we've got to do of course is ask, can

15:31

we ever actually manipulate gravity

15:34

in that kind of way? Because if we can

15:36

and the equations are right

15:38

then time travels possible as

15:41

our warp drives to travel at any speed through

15:43

the universe. At this stage we have to bring it

15:45

to a halt. So you

15:47

and your colleague, Dr. Brendan J.

15:50

Brew, who's a student who's an astro

15:52

statistician, which I didn't even know

15:54

existed until today, you have shown

15:57

that the early universe ran

16:00

one-fifth slower, somewhere between one and

16:02

two billion years after the universe began. Forget

16:04

the universe is 5,000 years old thing. How

16:07

can people follow you and your fine work? So I have a website

16:09

which I update very regularly at garintflewis.com.

16:23

And you like me are still stuck on X or Twitter

16:25

or whatever Elon calls it today depending on

16:27

what he had for breakfast. Yes, but as soon as he introduces

16:29

a charge, I'm gone. Me too. Yeah.

16:33

Thank you so much. We had so much fun.

16:35

Professor Garin? Yeah, that'll do. See

16:37

you next time. Bye. Thanks,

16:39

Carl. It's getting hotter. Our population's

16:42

aging. We're glued to our screens

16:45

and AI, well, it's

16:47

changing the world as we speak. We're

16:50

facing big challenges. We

16:52

need big solutions. I'm

16:56

Mark Scott, the Vice Chancellor at the University

16:58

of Sydney, and I've got a backstage

17:01

past to the people making

17:03

change happen. The solutionists.

17:05

Look for it in your favorite podcast app.