Thank you to Brilliant for supporting PBS. Here’s the story we like to tell about the beginning of the universe. Space is expanding evenly everywhere, but if you rewind that expansion you find that all of space was once compacted in an infinitesimal point of infinite density—the singularity at the beginning of time.
The expansion of the universe from this point is called the Big Bang. We like to tell this story because it's the correct conclusion from the description of an expanding universe that followed Einstein's general theory of relativity back in the 1910's. But since then we've learned so much more.
Does our modern understanding of the universe still insist on a point-like Big Bang? Recent work actually gives us a way to avoid the beginning of time. The key to understanding whether the universe had a beginning is to decide whether there exists what we call a past singularity—whether all points in space were at the same point at time t=0.
And from the earliest models of the expanding universe, the answer seemed yes. When people like Alexander Friedmann and Georges Lemaitre solved the equations of Einstein’s general relativity back in 19-teens and twenties, they found that, if you rewind the expansion indeed all points converged on one point at the start of time. But these guys made some big assumptions—for example that the universe is perfectly smooth everywhere.
It’s not—it’s at least a little lumpy or we wouldn’t have galaxies. Could this non-perfect-smoothness prevent our rewound universe converging on a single point? We’ll come back to that—it’ll be important later.
The fact that the universe is mostly perfectly smooth hints at something else that may change our prediction of a beginning of time. To explain the general sameness of our modern vast universe, we needed to add something called cosmic inflation to stretch out tiny smooth patches in the early universe into a smooth giant universe. At first inflation was described as a period of extreme exponential expansion that lasted for a tiny fraction of a second after the Big Bang.
But then we realized that this inflationary period didn’t have to stop everywhere—perhaps it stops in spots, creating bubble universes like ours, in which case there’d be a greater inflating spacetime that continues growing everywhere. But if this “eternal inflation” lasts forever into the future, could it last forever into the past? If so, perhaps the universe had no beginning.
So, our description of the universe has gotten a little complicated over the past 100 years. We’re going to need to carefully track our path backwards in time to see if we find a beginning. So we choose a coordinate system for space and time, and we trace those coordinates into the past they and see if they reach a point where we cannot trace them any further.
This can happen if some aspect of the coordinate system “blows up”--becomes infinite. We call such points singularities. In standard big bang cosmology there’s a singulartiy in the past where all paths in space overlap and can’t be traced further.
This is interpreted as the time before which there was no space. But there’s at least one example that you’ve heard of where a singularity doesn’t mean the end of spacetime. Consider the black hole.
The simplest type of black hole is described by the Schwarzschild metric. If we trace the Schwarzschild coordinates from outside the black hole in, we find they blow up in two places. One is expected—at the center of the black hole we find a proper curvature singularity—infinite spacetime warping.
But there’s also a singularity at the event horizon. There, it appears that the coordinate of time blows up. It’s as though we can’t trace time past that horizon.
We might be tempted to interpret this as an uncrossable boundary, but we’d be wrong. Just because Schwarzschild coordinates can’t cross this boundary doesn’t mean space and time end there. Sorry, you don’t bounce off the event horizon, you fall straight through it.
We only need a simple-ish coordinate shift to create a map across the event horizon that is free from infinities—from singularities. For example, Eddington-Finklestein coordinates combine space and time in a way that allows this continuous mapping all the way down to the central singularity—where we do find an infinity that does not go away. The event horizon is what we call a coordinate singularity—it's similar to how spherical coordinates have a singularity at the poles where lines of latitude converge.
We say that spacetime is extendable beyond the event horizon because a coordinate shift reveals the space beyond. In fact, if we switch to a Penrose diagram we reveal new spaces that look like a mirror universe and a past white hole—these are the fully extended coordinates… because we can, in principle, trace light rays to these places. But the central singularity of the black hole remains a physical singularity.
No light ray can be traced through this to the other side. It’s reasonable to state that spacetime ends at the center of a black hole. There’s no coordinate shift that extends our map beyond the central singularity.
So what about the beginning of the universe? Is there a singularity there, and if so is it a real physical end-of-spacetime singularity AKA the beginning of the universe? Or is it a coordinate singularity like the event horizon, with something on the other side.
To figure this out, we need another tool. It’s something we’ve discussed before: geodesic incompleteness. So, a geodesic is the shortest path through spacetime in general relativity.
It describes the curved path travelled in a gravitational field, or the straight path traveled out of one. Light travels something called a null geodesic, which is a path on which time itself does not pass for the traveller. Geodesics are normally thought of as continuing forever into the past and future, in that you can keep tracing these spacetime lines even beyond the segment travelled by a particle.
But sometimes a geodesic will reach a dead end where you can’t trace it any further. These are points of geodesic incompleteness, and they are interpreted as the literal end of spacetime—like the end of the map. The singularity at the center of a black hole is an example of such a point.
Geodesic incompleteness is how Roger Penrose proved that black holes contain physical singularities with his black hole singularity theorem—something we have of course covered. Geodesic incompleteness by itself is pretty convincing, but we’re always left with a nagging worry that we plotted our geodesics in the wrong coordinate system, like with the black hole event horizon. Another sure sign of the end of spacetime in the black hole singularity is that the curvature blows up— the strength of the gravitational field becomes infinite.
If we can find such a curvature singularity in a way that’s independent of the coordinate system, we can pretty much guarantee that the fabric of spacetime meets a bad end at that point. OK, so now that we have some sharper tools to pry open the beginning of the universe, let’s see what we can find. Well, for geodesic incompleteness the answer appears to be that the universe did have a beginning.
All geodesics—from the arc followed by a thrown ball to the Earth’s orbit around the Sun—can be extended back in time and wind up in the same infinitesimal point. And that’s even true if you include cosmic inflation—even eternal inflation. Demonstrating this past geodesic incompleteness has typically required certain assumptions.
For example, many arguments have depended on the weak energy condition, which is a sort of add-on to the main equation of general relativity that says you can’t have negative mass densities. But things probably get pretty weird when you approach the past singularity, so it would be nice if we didn’t even have to depend on this particular condition. More recently in 2003, Borde, Guth and Vilenkin made a stronger argument for geodesic past incompleteness that didn’t require an energy condition.
Instead, they only needed to assume that the average expansion rate was always positive. This led to the Borde–Guth–Vilenkin (BGV) theorem, which states that any universe that has, on average, been expanding throughout its history can’t have been expanding forever—it must have a past boundary. OK, great.
But can we be sure that this past boundary is really an end … or a beginning to spacetime? What if spacetime is really extendible, like it was through the black hole event horizon? Let’s look at the other criteria that we used for black holes—whether curvature becomes infinite.
If so then our universe surely must contain a physical singularity in its past. If it doesn’t contain curvature singularity then this doesn’t guarantee that we can extend our spacetime, but we’d need to do a bit more investigating to be sure. This is what Geshnizjani, Ling, and Quintin do in their paper published last year.
They applied a curvature test to the beginnings of a range of different universes to check the nature of any singularities. The results depend on the expansion history of the universe. Remember that the BGV theorem tells us that universes that have, on average, only expanded over their history must have a past boundary, which might be interpreted as a beginning.
This new study finds that universes with different expansion histories don’t have to begin with a singularity, and so, perhaps, don’t have to begin. For example, a universe that expanded after a prior phase of contraction or from a prior static phase don’t have to have past singularities. At the same time these expansion histories also seem to violate certain energy conditions of general relativity, and so may not be possible at all.
This isn’t really an entirely new finding—more a confirmation that you need to break some aspect of general relativity to avoid a beginning to the universe. Otherwise they mostly agree with the BGV theorem. Universe that have only ever expanded, do have a beginning.
But in the researchers do find one unusual case in which the BGV theorem may be challenged. Remember that an inflating universe has exponential expansion. A pure exponential function with no vertical offset looks like this: it grows rapidly into the future, but also plateaus approaching but never reaching zero size in the past.
The BGV theorem tells us that even this ever-expanding universe should suffer geodesic incompleteness into the past. There should be some sort of past boundary that we’re tempted to identify as the beginning. That boundary can be depicted as the lower diagonal edges on a Penrose diagram.
These represent past boundaries of the universe if there’s no way to extend a null geodesic beyond them. Except, as it turns out there is a way. This new study found that there’s a type of spacetime that could extend to this region beyond the past boundary.
Our universe is pretty well described with something called the Friedmann-Lemaitre-Robertson-Walker metric. It’s that perfectly smooth, flat, expanding universe that I mentioned earlier this episode. If you add exponential expansion to that—whether as cosmic inflation or as dark energy—then FLRW metric can be considered to be a subset of something called de Sitter space.
In fact, as these researchers showed, it’s possible to think of our universe—our patch of FLRW space—as just a segment of a larger de Sitter space. The details of this are a bit much for now, but long-story-short: the BGV theorem insists that our universe has a past boundary. If there’s no way to map a null geodesic across that boundary then we should interpret the boundary as a beginning of time.
But the new work shows there may be a place for those null geodesics to travel to, or have traveled from—the larger de Sitter space that extends beyond our FLRW universe. If our universe is extendable in this way, that might just turn the past boundary of our universe into a coordinate singularity rather than a real, physical one with infinite curvature etc. Now this extended spacetime may be - probably is - as illusory as the white hole and mirror universe of the Penrose diagram.
But it’s intriguing that we have here a case where it’s not just blatantly obvious that the universe rewinds into a non-traversable curvature singularity at least in this one case. A word of caution though. In order to make this work, our section of the universe—the FLRW section—has to transition smoothly into the de Sitter space.
The only way for it to do that is for the exponentially accelerating component—often described as the cosmological constant—to dominate over any density fluctuations at the beginning of “our” universe. If those density fluctuations are very small then spacetime is extendable past the past boundary of our universe and we may not have a physical past boundary. But any significant density fluctuations would actually close off the boundary and should turn this transition into a hard curvature singularity—eliminating the possibility that there’s anything “before” it.
And our universe definitely had density fluctuations at very, very early times—otherwise we would not have galaxies and planets and YouTubers today. It’s interesting that the same lumpiness that Friedman et al glossed over when they predicted the big bang may make the big bang more likely. OK, that bodes poorly for an infinite past for the universe.
There probably was a beginning of time. But the resolution of this is buried in the unknowns of inflationary cosmology. It also awaits our theory of quantum gravity, because, even if the universe approaches a point of infinite density in the past, our current understanding of physics breaks down before we reach that point.
The most stunning thing about this sort of work isn’t even the answer—whether or not there was a beginning to the universe. And by the way, the current answer is still probably yes. To me the most incredible thing is that we can even hope to answer questions that seem like they should be far beyond our grasp.
That through pure reason we can learn one way or another whether there was a beginning to spacetime. Thank you to Brilliant for supporting PBS. Brilliant is where you learn by doing, with thousands of interactive lessons in physics, math, programming, and AI.
Brilliant helps build your critical thinking skills through problem solving—not memorizing. Each lesson is filled with hands-on problem solving that let you actually play with concepts. And if you’re into vector mathematics, a fundamental concept for describing motion and orientation in space, Brilliant released a new course on Vectors.
With this course you’ll learn the basics of vector operations, including scaling, transformations, polar coordinates, and the dot product. You can enhance your ability to visualize and solve complex problems in multidimensional spaces. And you’ll apply your newly developed skills to program a game, and design a logo to gain insights into how the physics of motion translate to a digital environment.
To try everything Brilliant has to offer for free for a full 30 days, visit brilliant. org/spacetime or click on the link in the description. If you go now, you’ll also get 20% off an annual premium subscription.
Hey Everyone. On September 26th, 1905 Albert Einstein published his theory of special relativity and altered the course of physics history. To celebrate the over 100 years of relativity, we have two new items at the merch store.
First up is our universal speed limited embroidered patch. And next up is our 2D Light Cone enamel pin with holofoil backing. With both pin and the patch, you have everything you need to derive both special relativity all on your own in case you ever get time traveled back to 1904.
This limited edition pre-sale will run through October 7th and items will ship at the end of November. Just go to PBSspacetime. com/shop or click the link the description.
