Nobel Lecture: Roger Penrose, Nobel Prize in Physics 2020

Scientists won the Nobel Prize
12 followers
Sep 26, 2025
This year's Nobel Prize in Physics celebrates one of the great triumphs of mankind's thinking
and exploration. The formation of black holes was predicted on theoretical grounds. These are
objects that by definition cannot be observed directly. Astronomers took up the challenge
and thanks to great innovations and by pushing the boundaries of technology were able to show
firm evidence for their existence. These bizarre regions of space-time, where gravity is so strong
that even light is trapped, have captivated the interest of both scientists and laymen alike.
Among physicists, many resisted the idea that such beasts could arise in the universe as they
involve a so-called gravitational singularity at their center, a point where gravity becomes
infinitely strong. Albert Einstein was arguably the most notable skeptic. Simply put, the notion
shared by many of the sharpest minds of the first half of the 20th century was that it would take
unphysically perfect symmetry for collapsing matter to become a black hole. Thus, in a real universe,
natural imperfections would prevent such objects from ever forming. This year's Nobel laureates
have drastically changed that perception. Roger Penrose, receiving one half of this year's Nobel
Prize, used ingenious mathematical tools to demonstrate that the formation of black holes
is in fact a robust prediction of the theory of general relativity, regardless of the geometry
of the mass being pulled gravitationally. The other half of the prize is shared by Reinhard
Gensel and Andrea Ghez for their spectacular detective work examining the orbits of stars
around the Milky Way center with exquisite precision, circumventing formidable observational
challenges to disclose the nature of the invisible object in the galactic center, harboring over
four million times more mass than our Sun. With that, I would like to invite our first speaker,
Roger Penrose, to talk about black holes, cosmology, and space-time singularities.
Roger Penrose earned his PhD from Cambridge in 1957. In the late 50s and in the 60s,
he held positions in prestigious institutions in the US and in the UK. And in 1973, he became
the Roosevelt Professor of Mathematics at the University of Oxford, turning into emeritus in
1998. There are numerous mathematical and physical findings that carry his name, including Penrose
tilings, Penrose-Hawking singularity theorems, and of course, Penrose diagrams. He's also
exceptionally gifted as a popularizer of science and has written several best-selling books.
Please join me in welcoming Sir Roger Penrose.
In 1908, Hermann Minkowski introduced the idea of space-time, which was a four-dimensional space
which encapsulated pretty well all of Einstein's 1905 theory of special relativity. At first,
Einstein didn't like the idea very much. He thought it was mathematical sophistry or something.
But then he picked up on it, and it was central to his generalization, to his general theory
of relativity. Now, in the first picture, I've imagined three axes for a three-dimensional space,
and then we can move on to introduce a time axis to see our four-dimensional space.
Now, the most important thing of this is to represent the speed of light. Here we have a
light ray, and we want to see it so that it doesn't sort of lie on the floor. So we want to have units
so that it can be seen as maybe 45 degrees or some reasonable angle, so that you have your
space and time units so they're comparable. Now, here we have the null cone, which represents the
directions of all the null rays. So the light rays in all directions are represented by this cone.
It's very important. In fact, we don't really need the light ray there because we've got them all in
the cone. We don't need the axes. So the important thing is this null cone. Now, in general relativity,
you see there's a null cone at each point representing the local speed of light,
but the cones can be more or less all over the place. Now, you can imagine a point and the light
rays coming out of that point, and that's the light cone, if you like. The null cones can be
tangent to it wherever it goes, but you can see at the back, at the top right-hand side,
where the light rays start to cross each other. This sort of thing makes light cones complicated,
but it's important for what I'm going to talk about later that you understand these things.
You certainly are going to get crossovers, crossing points, caustics, and things like that,
and they're a central feature of what I'm going to discuss.
Okay, now let's consider the following picture. In this picture, we see basically the Oppenheimer
Snyder collapse of a dust cloud to what we now call a black hole. This was in 1939,
when they studied a collapse of what they called a dust cloud. This has no pressure.
What you call dust is simply a fluid or something with no pressure.
The thing was spherically symmetrical, so the fact that you fell inwards and it focused itself
into this central point, and you see as you move up the picture, you see this singularity in the
middle where the dust cloud gets itself focused to, and you find that since the density goes
infinite, the spacetime curvature becomes infinite, and this is what's called a singularity.
Now, this was known, and at the time that quasars were discovered, people started to wonder whether
there wasn't something like the Oppenheimer Snyder collapse involved. Now, I wasn't aware
of this Oppenheimer Snyder paper in the, I think it was in 1958, when I went to a lecture in London,
in King's College London. My good friend and mentor, Dennis Schauer, drove me there. He said
it would be interesting to me, and this was a talk describing how you get through what was
then thought of as the Schwarzschild singularity. Now, the top part of this picture is what
Finkelstein described. You see, Schwarzschild, shortly after Einstein introduced his general
theory of relativity, solved the equations for a spherically symmetrical body. Now, he also solved
it sort of for the interior of the body, but that wasn't a very realistic model. It wasn't important
so much. What was important was the solution for the exterior of the body, spherically symmetrical
vacuum. Now, the thing about this is, if you imagine squashing the body down, smaller and
smaller and smaller, you get to a point which is called the Schwarzschild singularity. Often, it was
called that because the equations all go crazy, and things go infinite, and people used to think
this was a singularity, which means you simply have some physical nastiness which you can't
extend beyond. But the models I'm showing you here, well, at least the top part of it, was described
to me by David Finkelstein at King's College, where he gave a talk in, I think it was 1958.
And I came away thinking, gosh, you've got this singularity still in the middle, whereas you got
right rid of the one on the outside, but you still have that one in the middle. So, I wondered whether
there was a theorem or something which showed that whatever you did, if you had it complicated,
irregular in some way, you would still get a singularity. I had no idea of how you might prove
such a thing, so I started to think to myself, what do I know about general relativity that maybe
other people don't know, and possibly this would be helpful to me to do something that people aren't
familiar with. What I settled on was two-component spinners. Now, here we have, in the next picture,
a picture on the right-hand side of a two-spinner. I learned really about these things from the great
physicist Paul Dirac, I think it was earlier in the same year, when he sort of deviated from his
normal course and talked about two-spinners. He was famous for discovering the equation for the
electron, but these use four-spinners, four-component spinners, and you can break them down into these
two-spinners. And Dirac was well aware of that, and I was familiar with the idea, but I simply
didn't understand them. And Dirac's lecture made it absolutely clear to me. And so, I began thinking,
okay, this is something that can comply to general relativity, and maybe it will give me an insight
that perhaps isn't familiar to people. In the picture you see now, on the left-hand side,
you see the celestial sphere, and that's the different directions in which the two-spinner
can point in. The two-spinner points along the light cone, but it's also got a little flag
attached to it. I won't go into that at the moment, but it was important to have an understanding of
the geometry of two-spinners. Now, the important thing, mainly for me, was that you can understand
the curvature much better in two-spinners, in particular, the part of the curvature which is
called the Weyl curvature. Now, in this picture, you see the curvature splits into two parts. One
is called the Ricci curvature, and that is what matter directly influences. Einstein's theory
tells you that the matter density, the energy density, the pressure, and all that stuff,
tells you what the Ricci tensor is directly. But what's left, that's 10 components. What's left
is another 10 components which describe the way the gravitational field behaves. So, the free
gravitational field in a gravitational wave, say, is described by the Weyl curvature, and you see it
gives you a distortion in the field of vision. So, somebody is looking back, and you see
the focusing effect due to the Ricci curvature, and then you see the effect due to the distortion
due to the Weyl curvature. Now, the Weyl curvature satisfied a very nice equation
when you write it in two-spinners, and it really attracted me very much. But getting used to how
light rays behaved, it was something that I felt at home with, and how they focused,
and how caustics behaved, and crossing surfaces behaved, and all that sort of thing. So,
I got familiar with all that. But at the time, this was when the quasars were being seen. This
was in 1962, 63, 64, that sort of time, when these bodies which were producing enormous amounts of
energy, and they seemed to be very small. So, they had an energy which was, I forget, 100 times
an entire galaxy, but yet they seemed to vary in a few hours, or days, or something like that,
which meant they had to be very small, and they had to be very big to emit such energy,
very massive to emit such energy. So, how could all that energy be squashed into that small volume?
And people started to speculate on whether something like the Oppenheimer-Snyder model
might be relevant. But then, this thing just collapses radially inwards, and doesn't give
you anything. So, if you want to have radiation coming out, you need to have at least a
quadrupole structure, or something complicated, not just a radial collapse. So, I started thinking
about it, partly at the instigation of John Wheeler. And at the time, there was a paper by
two Russians, Lifshitz and Kalatnikov, which seemed to have proved that, in general, you would not get
singularities, and only very special situations. So, if you had a generic collapse, it would just
swish around, and come swirling out again. So, I started to worry about this problem.
I had a look at their paper. I didn't notice a mistake. There was a mistake in the paper. I
didn't see that. But what I did see was the methods they were using, I didn't feel altogether
convincing, and that it was worth trying to see whether you could get singularities in a generic
situation. And I remember walking in the woods, and trying to think about this. And I came to
the conclusion that it couldn't be a local thing. It had to be something non-local,
some kind of criterion, which tells you the point of no return, in some sense, has been reached.
And I then devised the idea of a trapped surface. The picture you see here is the picture that
appeared in my paper, a paper that I published in 1965. I gave a talk at King's College London
in 1964 about it. And the argument was that if you have a collapse which is generic,
that you might still have problems, even though it wouldn't be focused right into the central
region. And I developed this idea of a trapped surface. You can see that's this little ring
in the sort of middle of the picture. You have to imagine that it's not a ring, because I'm only
depicting two dimensions, and the whole thing should be four-dimensional, and that ring is
really a two-dimensional surface, like a sphere. But you can imagine it might be distorted,
not like that. Now, what is a trapped surface? You have to imagine that there's a flash of light
emitted all along that surface here. On the top left, you see a picture of a little surface,
and you imagine, suppose there's a flash of light occurring on that surface. And if it's convex,
well, on the concave side, the flash will be converging. On the convex side, it will be
diverging. And that will be the normal thing for a surface. Now, on the top right, you have
something which is sort of, you might imagine that it can converge on both sides. On the bottom,
you see a much more general situation, which doesn't depend on any kind of curvature. You
get this in flat space. The intersection of two paths like cones, you get this property
of a locally trapped surface. Now, the trouble is that when you have a surface like this,
which is global all the way around, such as in the picture that I have on the central ring,
which is really a two surface, and I'm imagining that a flash of light occurring on that,
it's converging on both sides. And that's what a trapped surface is. And what I was able to show,
that if you have energy which is locally positive, and you look at the behavior of the future of that
surface, that you must run into a contradiction. That is to say, you get a singularity. So,
this was a proof, in general circumstances, that you couldn't have irregularity or something
swishing around in a complicated way and come swirling out again. You would always get a
singularity. And this was the central theorem that I proved, which eventually seems to have won me
the Nobel Prize. Okay. Now, let's think about the universe as a whole. Now, you see, Stephen Hawking
was not at that talk, but Dennis Sharma got me to give a repeat talk in Cambridge. And Stephen was
at that talk in Cambridge. And we had quite a long chat afterwards about methods I was using.
George Ellis was there also. And Stephen picked up on it very quickly and developed these ideas
by generalizing my arguments to apply to cosmological model. I won't go into the
details of this, but the main point was that you could also show, using the arguments that he'd
developed, and also we got together eventually and wrote a paper together, which more or less
encompassed the things that we'd done before. And this showed that under very general circumstances,
in the past now, if you have some kind of diverging ray, some kind of expansion in the universe,
then you couldn't avoid having this singular state in the very early universe.
Now, this was fine, that you seemed to prove these singularities. But I remember being
very puzzled by why cosmologists didn't study all the different kinds of singularity you could have,
because they're very common. There are many, many solutions known to general relativists,
where you could get very complicated singularities. And I remember this was an occasion when I think
I was at Princeton, and we were about to go to a conference at Stephen's Institute in Hoboken,
and we used to drive up in several cars, and I noticed in the car, in the back of one of these
cars, was Jim Peebles. This was last year's Nobel Prize winner in physics. This was long before that,
of course. But I noticed he was there, and I asked him, I said, why don't you cosmologists ever
consider all these complicated possible singularities that you could have, and you just look at this
simple case? Why don't you look at these other cases? And he looked at me and says, because the
universe is not like that. And I thought, my gosh, it's not, is it? I presume that he was looking
at the microwave background radiation, which is very uniform over the whole sky, and it tells
you that the universe really is very uniform. So it told me that there's something very strange
about these singularities, that the Big Bang singularity is utterly different from the kinds
that you see in the future, in the collapsing and black holes. And I was very puzzled by this,
and since everybody seemed to think that the solution of the singularity problem was that
you had to combine general relativity with quantum mechanics, you had to find a quantum gravity theory,
it must be a very peculiar quantum gravity theory, which is grossly asymmetrical in time,
which gives you a theory which makes the singularities quite different in the past
than the future. Well, I had this view for quite a long time, until I think I started to worry more
about, well, I thought it worried about the singularities in relation to the entropy in
the universe. I'll come to that in a minute, but first of all, let me talk about what the universe
is actually like, according to what people think. You see, what they think is that the Big Bang was
not just like this, and if you have a magnifying glass, you can have a good look at it, there's
something called inflation. Now, inflation, according to the theory, is another exponential
expansion, which took place within a tiny, tiny, tiny fraction, 10 to the minus 30 something,
of a second, and that is supposed to have taken place, this exponential expansion,
and it was supposed to have smoothed out the universe, and do various other things,
do the microwave background, make it look as though it was scale-invariant, and things like
that. So, it had a role to play, but I didn't see that it would smooth out the universe.
Well, the reasoning I was using is, suppose that you imagine a collapsing universe,
so here we have a collapsing universe, and this collapsing universe, suppose we put the
inflaton field, which is supposed to give you inflation, well, it just doesn't do anything.
If you have irregularities, black holes all congealing together, the picture would look
more like this, a great mess, with your vial curvature diverging like mad, and then you
imagine, well, that is a much more likely thing, and you can work out how much more likely it is,
something like, the probability of finding this is 10 to the 10 to the hundred and something,
10 to the power of 120 or something, that you get that more likely than the isometrical model that
we seem to see, at least in the Big Bang, but in the Big Crunch, if you had the universe collapsing
into a mess like this, this is what you'd get. But the question is, why wasn't it like this in
the past? Is it a very strange kind of quantum gravity theory, which is what I thought for a while.
Now, this is connected with another problem, which is the problem of the second law of
thermodynamics. Now, let me describe the picture of the second law of thermodynamics.
People often talk about a gas in the box, so the top three pictures, on the left-hand side,
we see a gas tucked up into the corner in a little box, you open the box, and it spreads out
through the big box, if you like. So, as the gas spreads out, it gets more and more uniform,
the entropy or the randomness increases in accordance with the second law.
Now, that's what you see, the right-hand side of this picture, you see matter, which is very,
very uniform in the beginning, which would be a maximum entropy state. In fact, the microwave
background, where you see it, it looks like a maximum entropy state. So, there's something very
funny going on. How can it start off at a maximum? But then when you think about gravity,
it works the opposite way, because gravity is uniformly attractive. So, here I have a picture
on the left of a lot of stars running around, and then they tend to clump, and then they finally get
black holes, and the entropy goes shooting up enormously by the Bekenstein-Hawking formula,
which tells you what the entropy in a black hole is, and it completely and utterly dominates the
entropy in the universe. Right now, almost the entire entropy in our current universe is in
black holes by an enormous factor. So, we see that what's special about the universe is that
the gravitational degrees of freedom were not activated. Somehow, and I tend to postulate,
just that maybe quantum gravity tells us in some mysterious theory, that somehow the
Valka, which had to be zero in a past singularity, and it could be infinite in a future singularity,
just sort of waving hands around and no theory which tells you that.
And I thought that for a long time. But then, after I became persuaded, and I think I want to
go back now and look at this picture, that the universe, this was another Nobel Prize,
Saul, Paul, Mutter, and Schmidt, and Ries, discovered the distant supernovae stars seem
to be accelerating away from us. And I had to be persuaded of that. And when I came around to
believe it, I had some wrong reason for disbelieving it. But when I came around to believing it, I
thought, gosh, this means, this is interesting. It tells you that the future infinity is space-like.
Now, let's talk about infinity. See, people think, how do you talk about infinity? Well,
now, Escher has a very nice way of describing infinity. This is a conformal representation
of what's called hyperbolic geometry. Don't worry about the geometry, but the boundary of this
represents infinity. Now, it's a conformal picture. That means that angles are preserved,
but sizes can be big or small. And the fish, as they get closer to the edge,
they don't really realize that they're small. According to them, they're the same size as the
ones in the middle. So this is a conformal representation. You can squash or stretch
as long as you don't alter the small shapes. And that's conformal geometry.
Now, the importance of conformal geometry I'll come to in a minute, but let me think about the
light cone again. The space-time structure means not just the light cone. You need to know these
surfaces within it, which are the surfaces of equal time from the origin point. So if you have
some light rays, these aren't right. You have two massive particles, and I'm having two different
ones going at different speeds, going through the central point. And the ticks of a clock,
at the top you see little clocks, and the clock registers the time as the world line of that
particle intersects these various bowl-shaped surfaces. The first tick, the second tick,
the first tick. And by the two formulae at the bottom, we have the two most famous formulae of
20th century physics. Einstein's E equals mc squared, and Max Planck's E equals h nu. You
put the two together, and you see that energy and frequency are equivalent from the Planck,
and energy and mass are equivalent from Einstein. Put the two together, and you get that mass and
frequency are equivalent, which tells you that a massive particle is a very perfect clock. So this
is where you get the metric structure. But what about light rays? Well, you see the light ray
doesn't even hit the surfaces. So light ray doesn't register the passage of time at all.
And here we have, we forget about the light ray, the light cone or the null cone is itself,
gives you the structure of space-time. So if you only got null cones and not the scale,
then you have a good picture of infinity. And in fact, you can do this, and not only you can
have a picture of infinity, but you can stretch out the Big Bang as well. And this is, you see,
I was trying to say as a sort of criterion that the Big Bang or initial singularities ought to
have vanishing vial curvature. My student, Paul Todd, had a better way of doing it to say, okay,
let's say that it's extendable to something. You see, you could imagine the Escher picture,
you could extend the little fish to beyond the bounding circle to something outside.
They don't experience that, but you could imagine there was a world continuing beyond infinity.
And here, I imagine there might be a world continuing beyond infinity, and there might
be a world continuing before the Big Bang. Now, does this make physical sense? It makes
geometrical sense, but what about physical sense? Well, you see, the remote future is very rarefied,
very cold, and the density is very, very low. When you scale things to squash it, to make it a finite
boundary, this makes the energies go up and the temperature go up. And if you stretch out the Big
Bang, that makes the energy go down and the temperature go down. And you might imagine that
physically, they are similar. So, would they match? That's a possibility. You need the space-like
nature of infinity, and that comes from the cosmological constant. The space-like nature
of the Big Bang comes about automatically. You also get something more if you do what I'm now
going to suggest, which is to imagine not just that you have a finite boundary at the two ends,
but you continue. Our Big Bang was the continuation of something before, and our remote future
will continue to something beyond it. Now, does this make physical sense?
It only really makes physical sense if you've got massless things around. What about the remote
future? Well, the main things running around will be photons, and they're massless. So, let's say
there's something more. You've got mass as well. I have to sort of suggest that the mass does fade
out eventually, but let's say it's dominated by the photons. What about the Big Bang? Well,
it's the opposite reason. Here, by Einstein's E equals mc squared, you find that the energy in the
mass is pretty well completely dominated by the energy in the motion. So, the kinetic energy,
when the motions get so big, the energy gets so great, then it's almost entirely in the kinetic
energy and not in the mass. So, you might as well consider that the mass is zero, and the physics
relevant to the two ends is the physics of zero massless particles. In fact, in the remote future,
the photons, that's Maxwell's equations, Maxwell's equations don't even notice the metric. They are
conformally invariant. In fact, you can consider that you could actually, if you have electromagnetic
things or massless things or gravitational wave things, they could get across from one side to
the other. So, I can imagine that the crossing over from one side to the other makes sense if
your physics is sufficiently conformal and you're not worrying too much about the mass.
Okay. Now, I'm going to now finish this talk by talking about two classes of observation. The
first one was an idea I had about maybe you would see in the previous eon, if it's like ours, I'm
calling the different sections, the different eons. So, if I go back to this picture, our eon is
this initial cylinder. It's not really a cylinder. You see at the back, there's some curly stuff. It
may not close up at the back. So, I don't mind whether the universe is closed spatially or open.
Now, here I have the join from one to the next, and you might imagine light rays getting across.
But what else might get across? Well, here, the previous eon, I'm calling all those different,
what we previously called the universe was our eon. There was an eon prior to ours,
an eon after ours, and so on. Now, the big bang of our eon was the remote future of the previous
eon. Now, there might have been in that previous eon, black holes, supermassive black holes running
into each other. As they run into each other, they would emit gravitational wave signals. These would
get through and make a signal that we might observe in our eon. Moreover, if you imagine
what the big, really superduper massive black holes, which ultimately swallow clusters of
galaxies, they will be the result of many, many supermassive black hole collisions,
and there will be maybe one, two, three, four, five, six, several of them. So, you might see
not just a signal of one of these, but several of them. Now, what you would expect to see,
according to this theory, would be rings in the cosmic microwave background, where the temperature
is slightly lower variance as you go around, or higher or lower temperature. My colleague,
Vaheguru Zidane, who looked at this, he looked at concentric sets of at least three rings
of low variance, low temperature variance. He plotted these things out.
In the next picture I'm going to show you, which is rather remarkable to me,
he didn't select these things from the color, which is the temperature. He didn't select it
for that at all. He selected it just for the low variance. These are plotted out in the Planck data
centers of at least three concentric low variance rings. Now, whatever you believe the origin of
the signal is, what is very remarkable is the extreme anisotropy of the picture, or inhomogeneity
of the apparent universe here. On the lower right, you see a very large collection of red points.
Now, in the color coding and in the theory, I won't go into the details here, the red means
distant. So, these would, according to the theory, be a collection of very distant, supermassive
black holes clumped together in some super-duper cluster. On the top right, you see a bluish region
that would be closer to us within our past light cone, so you probably could see that,
of something which, again, not quite so big and super-duper, but pretty super-duper.
On the slightly lower left, you see something intermediate. The red ones would be distant,
according to my theory, the blue ones. But what's remarkable is not just they're clumping,
they're clumped together in where you see them, but they're clumped together in the color, which
means, according to the theory, in the distance. So, if you don't believe my theory, you have to
think of another explanation for this, but it's very much at odds with the current view, where
the inflation is supposed to stretch things out and flatten everything out. Another possible observation
of events in the previous eon would be the evaporation of supermassive black holes,
according to Stephen Hawking's Hawking Evaporation, where the entire mass, ultimately, it may take
something like 10 to the 100 years or more, for a supermassive black hole finally to disappear
in form of radiation. Now, this radiation would get through into the next eon. So, here, I have
a picture taken from a paper written by Christoph Meissner, Pavel Neurofsky, Daniel Ann, and myself
in the Monthly Notices of Royal Astronomical Society, and we see the crossover is the lowest
horizontal line in the picture, and the vertical line just meeting it is the world line of a
supermassive black hole, which finally evaporates right at the crossover, almost, into the next eon,
really squashed up into that little point. It spreads out through 380,000 years until you
finally see it when it reaches the last scattering or decoupling surface, and according to the work
of James Peeble and his colleagues, you have a good knowledge of what happens in that previous
380,000 years. But what we actually see is the spread of that point to about eight times the
diameter of the moon. So, this would be a heated point, which is hotter in the middle by about
something like 15 times, it could be as much as that, times the background variations in
temperature, the normal variations. And we seem to see, in both the WMAP and Planck satellite data,
these points. In the Planck data, we see it with a confidence in something like 99.98% confidence
level. So, this is a very strong signal that these points are there. Of the five strongest points in
the Planck data, these are also seen as exactly the same places in the WMAP data. There's another
one in WMAP which is seeing the same point in the Planck data. So, these six points, I think,
are pretty... I have pretty great confidence that they are genuine Hawking points. If not,
I think somebody else will have to come up with an explanation for this effect and explain
how we can have a confidence of the level we seem to see on the basis of some other theory.
Thank you very much.
