1. The Local Cosmic Scene

This two-faceted nature of cosmology perhaps justifies a few comments on important advances in the last decade, in the ‘near universe’ of our own Galaxy, that have rendered our cosmic habitat far more interesting. We have learnt, in particular, about the ubiquity of extra-solar planets – one of the most exciting growth points in astronomy, with potential links to biology. The planets are not imaged directly, but are mainly inferred by high-precision observations of their parent star – either by detecting the small periodic motions they induce in the star they are orbiting, or by detecting the slight dimming when a planet transits in front if a star. The Kepler spacecraft, which operated successfully for more than three years, has detected thousands of candidate planets via the transit method.

Planetary systems display a variety that has surprised astronomers. Hardly any of the details of orbital periods, mass distributions and so forth were predicted. But the existence of extra-Solar planets was not a surprise because astrophysicists had come to expect retinues of planets as a natural by-product of star formation. Stars form via the gravitationally-induced contraction of clouds of dusty gas. If the cloud has any angular momentum, it will rotate faster as it contracts, and spin off a dusty disc around a protostar. In the outer and cooler parts of such a disc, gas condenses into ‘Jupiters’; closer in, less volatile dust agglomerates into rocks and planets. This should be a generic process in all protostars. So we shouldn’t be surprised that, of the hundred billion stars in our Galaxy, there may be billions orbited by planets resembling the Earth (although the likelihood that these alien worlds harbour life is exceedingly uncertain –biological questions are perhaps even more challenging than those in physics).

In this context, it is interesting to flash back to Newton. He famously explained why planetary orbits were ellipses, but did not understand why these orbits all lay close to an ecliptic plane – unlike those of comets, which come from all directions. He thought this alignment was divine providence. But we now realise it is a natural consequence of the ‘dusty disc’ origin of planetary systems. Indeed the explanatory chain has now been extended right back not merely to the formation of stars, but to the first nanosecond of the big bang.

What about stars and atoms? We see stars forming, in places like the Orion nebula.

We understand their evolution; and we see them dying. The sun will die non-explosively, but heavier stars explode as supernovae. Just before they explode, these stars have an onion-skin structure – with hotter inner shells processed, further up the periodic table, by a more extended series of nuclear fusion reactions. This processed material is then flung out in the supernova explosion. The debris then mixes into the interstellar medium and re-condenses into new stars, orbited by planets.

Our Galaxy is a huge ecological system where gas is being recycled through successive generations of stars, building up the periodic table – lots of carbon and oxygen, not much uranium and gold – from pristine hydrogen. The key features of this scenario were developed in the 1950s, predominantly by five people: Fred Hoyle, Willy Fowler, Geoffrey and Margaret Burbidge, and Alistair Cameron.Reference Burbidge, Burbidge, Hoyle and Fowler¹

2. The Evolving Universe

The next step in a backward extension of the causal chain – to understand where the original protons and neutrons came from, and how some of these condensed into the first stars – takes us into the extragalactic realm, and into the province of the astronomers who study galaxies and their evolution. At first sight it would seem surprising that much can be learnt about galaxies. Whereas physicists who study particles can probe them, and crash them together in accelerators, astronomers can’t experiment on galaxies. And galaxies change so slowly that we only see a snapshot of each. But we can do computer simulations, and even crash galaxies together, in a ‘virtual universe’. And we have a huge number of galaxies to observe – to classify, and compare with our simulations. Moreover, astronomers have an advantage over (for instance) geologists: the current stellar population in galaxies, and the way the concentration of the different elements in stars depends on their age, offers ‘fossil’ evidence of galactic evolution. But by observing distant galaxies we can actually ‘see’ the past, and test models of galactic evolution more directly. Indeed, a quasar has been observed that is so remote that the redshift stretches the wavelength of its light by more than a factor of 8: the Lyman alpha line of hydrogen, normally in the far ultraviolet, is shifted almost into the infrared.

Importantly, we find that all galaxies are held together by the gravity of not just what we see. If that were all they contained, their internal kinetic energy would overwhelm the gravitational binding energy and they would fly apart. Their equilibrium implies that they are embedded in a swarm of particles that are invisible, but collectively contribute about five times as much mass as ordinary atoms – the dark matter. This galaxy-scale evidence for dark matter is corroborated by measurements of clusters of galaxies. Moreover, if the dark matter consists of particles surviving from the early universe, computer simulations succeed in reproducing the observed structures of galaxies, and how they are clustered, on the assumption that the particles are uncharged, slow moving (and therefore heavy) and interact very weakly with ordinary matter. The actual nature of the dark matter particles may be clarified by (a) the detection of new candidate particles by the Large Hadron Collider (LHC); (b) the detection, via sensitive underground experiments, of the rare recoils when a dark matter particle interacts with a normal atom; or (c) detection of decay products (high energy photons or positrons) generated by annihilation of dark matter particles in space.

Now a more fundamental question. Our present complex cosmos manifests a huge range of temperature and density, from blazingly hot stars, to the dark night sky. But the microwave background, the 3° Kelvin radiation pervading all of space, is an afterglow of a phase when everything we can now see was in a hot dense state. So how did the present intricate complexity emerge from an amorphous fireball? The answer lies in the anomalous thermodynamic properties of gravity. Gravity enhances density contrasts rather than wiping them out. Any patch of the early universe that starts off slightly denser than average would decelerate more, because it feels extra gravity; its expansion lags further and further behind, until it eventually stops expanding and condenses out.

Computer simulations are able to model the evolution of the baryons and the dark matter in a domain large enough to make thousands of galaxies. The results can be presented as movies that clearly reveal incipient structures unfolding and evolving. The movies portray how galaxies emerged, 16 powers of 10 faster than it actually happened! Each galaxy is an arena within which stars, planets and perhaps life can emerge.

And there is one important point. The initial fluctuations fed into computer models are not arbitrary: they are derived from the observed fluctuations in the temperature of the microwave background that comes from very early eras. They are initially ‘metric fluctuations’ with an amplitude of only one part in 100,000; but, computing forward, they’re amplified by gravity into the conspicuous structures that dominate the present universe. The microwave background fluctuations were first beautifully observed by WMAP, a project initiated by the late David Wilkinson at Princeton. But the most recent and precise data have come from the European Space Agency’s Planck Spacecraft.

Comparisons of models with data on galaxies and their spatial distribution vindicate the claim that structure emerges by clustering of the gravitationally-dominant dark matter during cosmic expansion.

3. The Limits of Known Physics

We can trace cosmic history back to 1 second, when the thermal energy of each particle was 1 Mev and helium and deuterium form. Indeed we can probably be confident back to a nanosecond. That is when each particle had about 50 GeV of energy – as much as can be achieved in the LHC – and the entire visible universe was squeezed to the size of our solar system. But questions such as ‘where did the fluctuations come from?’ and ‘why did the early universe contain the actual mix we observe of protons, photons and dark matter?’ take us back to the even briefer instants when our universe was hugely more compressed still – ultra-early eras when experiments offer no direct guide to the relevant physics.

According to a popular theory, the entire volume we can see with our telescopes ‘inflated’, at 10¹⁶ GeV, from a hyper-dense blob no bigger than an apple. The physics at that era is of course still conjectural, but nonetheless the generic idea of inflation has achieved success in predicting two features of the fluctuations: that they are Gaussian, and that their amplitude depends on scale in a distinctive way.

A few years ago, cosmologists had a big surprise. It doesn’t change anything I’ve said about cosmic history, but alters our perspective on the far future (see Section 5). Forty years ago, many people, the great John Wheeler for instance, argued that the universe was finite, closed and would re-collapse. Twenty years ago, most thought it would continue expanding: gravity would slow the expansion down, but the mean density of the universe (baryons plus dark matter) was by then known to be only about 30% of what would be needed to bring the expansion to a halt. Today there has been a further revision. It turns out that the expansion is speeding up. On the cosmic scale gravitational attraction is overwhelmed by a mysterious new force latent in empty space which pushes clusters away from each other.

Support for this conclusion comes from two interlinked bits of evidence, both dating from the late 1990s. First, observations of distant supernovae (of type 1A), which serve as accurate ‘standard candles’, show the Hubble expansion was slower 5 billion years ago, not faster. And independent evidence from the microwave background shows that the universe is ‘flat’ (which is, by the way, another prediction of the inflation theory). There is a characteristic length-scale on which the early universe would be especially irregular; and the angular scale that this length-scale subtends then tells us the geometry. If there were just the baryons and dark matter, which make up only 30% of the critical density energy, the geometry would be hyperbolic. (A long thin triangle in a ‘flat’ universe has angles adding up to 180°; in hyperbolic geometry the sum of the angles is smaller, and so consequently would be the angular size corresponding to a given length-scale.) So the energy latent in empty space makes up the rest. According to Einstein’s theory, the ‘pull’ of gravity is not just proportional to the density, ρ, but to ρ + 3p/c ², and since the vacuum has pressure of (–ρ c ²) this leads to acceleration.

These two lines of evidence, which have both been firmed up in the last decade, together clinch the case for a ‘flat’ accelerating universe.

Whereas the nature of dark matter may well be pinned down quite soon, this vacuum energy – Einstein’s cosmological constant lambda in a modern guise – won’t be understood until we have a model for the graininess of space on the Planck scale. And this is of course a goal of string theory. (But I want to add a, perhaps, curmudgeonly remark. Popular presentations of cosmology customarily present a ‘pie diagram’ according to which about 5% of the universe is baryons, 25% dark matter, and 70% ‘dark energy’. This is of course true of the present-day universe, but it distorts the relative astrophysical importance of dark matter and dark energy. The dark matter is crucial both within and outside galaxies – and it was crucial in the past in controlling how galaxies formed. In contrast, the dark energy, which has a density that is uniform in space, and probably in time too, is everywhere swamped by dark matter in the first half of cosmic history, so it had no influence on how cosmic structures were laid down; moreover it is swamped even today within galaxies and clusters. Dark energy is the number one challenge for fundamental physics, but it is almost irrelevant to astrophysicists.)

Now another basic question. How much space is there altogether? How large is physical reality? We can only see a finite volume – a finite number of galaxies. That is essentially because there’s a horizon – a shell around us, delineating the distance light can have travelled since the big bang. But that shell has no more physical significance than the circle that delineates your horizon if you’re in the middle of the ocean. There’s no perceptible gradient across the visible universe – that suggests it stretches thousands of times further. But that’s just a minimum. If it stretched far enough, then all combinatorial possibilities would be repeated. Far beyond the horizon, we could all have avatars.

Even conservative astronomers are confident that the volume of space–time within range of our telescopes – what astronomers have traditionally called ‘the universe’ – is only a tiny fraction of the aftermath of our big bang. We expect far more galaxies to be located unobservably beyond the horizon.

Moreover, if the acceleration persists, these galaxies will never, ever, be observable – even in principle. But that’s not all. Plausible models for 10¹⁶ GeV physics lead to so-called ‘eternal inflation’. ‘Our’ big bang could be just one island of space–time in a vast cosmic archipelago. This is speculative physics – it’s perplexing today, just as the shape of the Solar System was to Newton. But it’s physics, not metaphysics – we can hope to push the casual chain back further still.

Mention of a multiverse often triggers the response that unobservable domains aren’t part of science. I want to contest this by systematic desensitisation – the psychological process whereby you are at first presented with a spider a long way away, but end up at ease even with tarantulas crawling over you.

I mentioned that there were galaxies beyond out horizon: we’re relaxed about that in a decelerating universe, because they would come into view in the far future, so eventually they would be observable. However, we realise now that because they’re accelerating away from us they’ll never even in principle be observable. But does that make them any less ‘real’? They’re the aftermath of ‘our’ big bang. But since they would never – even in principle – be observable, why is their reality status necessarily higher than that of galaxies in the aftermaths of other big bangs? If, of course, there are other big bangs – and, of course, we don’t know if there are. We’ll only take other big bangs seriously if they’re a prediction of a unified theory that gains credibility by being ‘battle tested’ in other ways.

So a challenge for twenty-first century physics is to address two fundamental questions. First, are there many ‘big bangs’ rather than just one? Second, if there are many, are they all governed by the same physics or not? Many string theorists don’t think so. They think there could be a huge number of different vacuum states – different vacua, arenas for different microphysics. If they are right, what we call ‘laws of nature’ may in this grander perspective be local bylaws governing our cosmic patch.

Many patches could be still-born or sterile – the laws prevailing in them might not allow any kind of complexity. We therefore wouldn’t expect to find ourselves in a typical universe – rather, we’d be in a typical member of the subset where an observer could evolve. Such conjectures motivate us to explore what range of parameters would allow complexity to emerge. Those who are allergic to multiverses can regard this just as an exercise in counterfactual history (rather as historians speculate on what might have happened to America if the British had fought more competently in 1776, and biologists conjecture how our biosphere might have evolved if the dinosaurs hadn’t been wiped out).

4. Prerequisites for a Complex Cosmos

We can highlight several prerequisites for the emergence of our complex and structured cosmos from simple amorphous beginnings.

First, gravity – which enhances density contrasts, and binds planets and stars together. It’s a very weak force. On the atomic scale, it’s about 40 powers of ten weaker than the electric force between electrons and protons, so it only ‘wins’ on large scales. Sugar lumps, rocks, and asteroids are unaffected their self-gravity. But this force makes planets round, and any object more massive than Jupiter is squeezed to make a star. Were gravity not so weak, there would be fewer powers of ten between the micro and the cosmic scales – and less space and time for complexity to emerge. (Incidentally, it is sometimes claimed that gravity is ‘fine-tuned’. That’s not the case. It merely has to be weak. If gravity were (say) 10 times weaker still, the universe would be even more propitious for emergent complexity: stars and planets would be 30 times heavier, stars would live ten times as long.)

And there are other requirements. There could be no complexity if the universe stayed in thermal equilibrium, as it was for its first half million years. And there must be an excess of matter over antimatter, and enough baryons relative to radiation and to dark matter. Another requirement for stars, planets and biospheres is that chemistry should be non-trivial. If hydrogen were the only element, chemistry would be dull, and there would be no nuclear fusion to power the stars. A periodic table of stable elements requires a balance between the nuclear binding force (the ‘strong interactions’) and the electric repulsive force that drives them apart.

The universe must expand at the ‘right’ rate – not collapse too soon, nor expand so fast that gravity can’t pull together the structures. This constrains lambda, the vacuum energy. If the cosmic repulsion were too strong, galaxies could never form in the expanding universe. And there must be some fluctuations for gravity to feed on – sufficient in amplitude to permit the emergence of structures – otherwise the universe would now be cold ultra-diffuse hydrogen – no stars, no heavy elements, no planets and no people.

The fluctuation amplitude is denoted by a number Q, which measures how ‘rough’ the early universe is. Its observed value is 10^–5 – there is no firm theoretical argument for this number. In inflationary models it depends on the scale of the universe when inflation occurs (inversely related to the amplitude of quantum fluctuations). The value of Q determines the scale of structure in the present universe. The bigger it is, the earlier structure forms and the bigger the structures get. We’re far from knowing what determines lambda or Q; and even further from putting a measure on the relative likelihood of each combination of constants. But if we could we’d know whether we were in a typical habitable universe. These explorations are interesting to theorists; they delineate the anthropically allowed range.

Anthropic arguments are irrelevant if the constants are unique. Otherwise, they’re the best explanation we’ll ever have. This is reminiscent of planetary science 400 years ago – even before Newton. At that time, Kepler thought that the Earth was unique, its orbit related to the other planets by beautiful mathematical ratios. We now realise that even within our own Galaxy there are billions of stars, each with planetary systems. Earth’s orbit is special only insofar as it’s in the range of radii and eccentricities compatible with life. Maybe we are due for an analogous conceptual shift, on a far grander scale. Our big bang may not be unique, any more than planetary systems are. It parameters may be ‘environmental accidents’, like the details of the Earth’s orbit. The hope for neat explanations in cosmology may be as vain as Kepler’s numerological quest.

If there is a multiverse, it will take our Copernican demotion one stage further – ‘our’ big bang may be one among billions. It may disappoint some physicists if some of the key numbers they are trying to explain turn out to be mere environmental contingencies – no more ‘fundamental’ than the parameters of the Earth’s orbit around the Sun. But in compensation, we’d realise space and time were richly textured – but on a scale so vast that astronomers aren’t directly aware of it – any more than a plankton whose ‘universe’ was a spoonful of water, would be aware of the world’s topography and biosphere.

5. The Far Future

Next, a few words about the far future. The biggest uncertainty is about the nature of empty space. Is the vacuum energy indeed constant – is it really like Einstein’s lambda? If this force decayed away or reversed – space undergoing a phase transition, as it were, then the universe could stop expanding, and collapse to a big crunch. On the other hand, if it got stronger, we’d undergo what’s called the ‘big rip’ – where galaxies, stars and our remote descendants get torn apart.

But the best and most ‘conservative’ bet is that this force is unchanging. If that’s so, galaxies accelerate away and disappear over an ‘event horizon’ – rather like an inside-out black hole. All that’s left will be the remnants of our Galaxy, Andromeda, and smaller neighbours. Protons may decay, dark matter particles will annihilate; there will be occasional flashes when black holes evaporate – and then silence.

That is the case if we assume that ‘nature takes its course’. But even some serious scientists, such as Freeman Dyson, have speculated on the limits of ‘posthuman’ technology. In a classic article in Reviews of Modern Physics Reference Dyson² he considered what would happen if there were no technological limitations. Indeed his aim was ‘to establish numerical bounds within which the universe’s destiny must lie’. Could an unbounded amount of information be processed – could an unbounded number of thoughts be thought – if all material was optimally converted into a computer or superintelligence?

In fact, 10 years earlier I’d written a juvenile paper discussing a re-collapsing universe.Reference Rees³ The ‘eschatology’ is then a depressing one – our remotest descendants will eventually fry as the temperature rises during the cosmic collapse. Dyson refers to my work only briefly. A ‘miserable end’ in a ‘closed’ universe gives him claustrophobia. So he discusses the ever-expanding case

The best prospect for ‘infinite subjective time’ would be an Einstein–de Sitter universe that decelerates, so that arbitrarily large amounts of material come within causal contact. But the cosmos we seem to be in – the aftermath of ‘our’ big bang – is one where the amount of matter within the horizon is going down not up. Unless our descendants engage in cosmic scale engineering to ‘lasso’ distant galaxies before they disappear, then eventually only the remnants of our local group of galaxies will remain within the horizon. There’s a definite upper bound to complexity and information processing.

But perhaps it’s some consolation that, if indeed there’s a multiverse; we may have avatars in cosmoses with far greater potential than our own. Time prevents me from pursuing these speculations further – which is perhaps a good thing!