The universe is misbehaving.
Upon its birth in the big bang, nearly 14 billion years ago, the cosmos began expanding. And for most of the 20th century, scientists assumed gravity would gradually slow down that expansion, with all the universe’s matter acting as drag.
But observations in the late 1990s showed that’s far from the case. Careful, high-precision estimates of cosmic distances using a special class of exploding stars called type Ia supernovae revealed that, against all expectations, the universe’s expansion is actually speeding up. This is akin, on cosmic scales, to tossing a ball overhead only to see it fly away at an ever increasing speed rather than fall back down. The matter of “why” this can happen remains one of the most pressing mysteries in physics.
On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
“The theorists are having a field day,” says Tamara Davis, an astrophysicist at the University of Queensland in Australia. “There are hundreds upon hundreds of theories about what’s driving cosmic acceleration.”
Scientists generally call the culprit dark energy, but no one knows what it is. Its behavior, however, may offer a potent clue to its identity: if dark energy’s expansion-accelerating effect holds steady over time, this would fit quite comfortably within what’s known as the standard model of cosmology, the best overarching explanation of the universe’s evolution that scientists have yet devised. The trouble is that no one has been able to say with certainty whether dark energy is actually so fixed—and if dark energy’s strength can change over time, reconciling this with the rest of physics may require rethinking our understanding of gravity.
The problem spurred an international group of more than 400 scientists, including Davis, to begin collaborating about a decade ago on an observational program called the Dark Energy Survey (DES). The researchers built a special camera to use with the Victor M. Blanco Telescope at the Cerro Tololo Inter-American Observatory in Chile, and they reported the latest, most authoritative results from their ambitious efforts earlier this month at an annual meeting of the American Astronomical Society. The group has also posted its findings to the preprint server arXiv.org.
Night after night for five years, the team hunted for the same kind of supernovae that first led to dark energy’s discovery. Type Ia supernovae all release about the same amount of light when they occur, which makes them handy “rulers” for reckoning cosmic distances. Astronomers can tell how far away a type Ia supernova is by measuring its brightness. Coupling that distance with a separate measurement of how fast the supernova is moving away from Earth reveals the speed at which the cosmos was expanding when that stellar cataclysm took place. This second measurement involves examining a supernova’s spectrum—splitting its light into constituent colors and seeing how reddened they are from the light’s wavelengths being stretched by cosmic expansion. The higher an object’s “redshift” is, the faster that object is receding from us.
About 50 supernovae, most with low redshifts and from the relatively nearby universe, form the basis for the 1998 discovery of dark energy. Probing whether and how dark energy might change across the eons, however, demands pegging the distances and velocities of supernovae from a much bigger swath of spacetime. With DES, Davis says, “we wanted to see whether dark energy has been the same throughout the universe’s history, and the only way to do that is by seeing what it was in the past.” Ultimately, the project found and studied more than 1,500 of the telltale stellar explosions, many of them with high redshifts from the far depths of the cosmos.
New Analysis, Same Perplexing Results
To spot fresh type Ia supernovae, “we had to play the game of ‘find the differences in the photos,’” says Maria Vincenzi, a cosmologist at Duke University and a DES collaborator, who co-led the group’s supernova cosmology effort. “You look at images from one night to another to see if you see something pop.”
DES researchers had to weed through about 19,000 “pops” that they identified in search of the right type. To comb through such a large dataset, the team developed new machine-learning techniques that made the process roughly 100 times faster.
But even that was not enough: although clever automation saves time in the initial selection of candidates, determining each one’s redshift typically requires getting more data. This can be an urgent, fraught task because gathering a supernova’s diagnostic spectrum must occur before its light fades away and requires securing oodles of fiercely competed-for observing time on a large telescope. “It only takes about five minutes to take a picture of a supernova versus several hours to get its spectrum,” Davis says.
The DES team’s solution was to measure redshifts for each computer-flagged candidate’s presumed host galaxy, rather than the candidate itself, using dedicated time on the Anglo-Australian Telescope in Coonabarabran, Australia. “So that’s the thing that allows you to sit back, relax and take the spectrum at your leisure whenever you like because you’re not in a rush to catch the supernova’s light,” Davis says.
And from this––the biggest, best survey of distant supernovae ever––what did they find?
“The DES paper explores whether there is a compelling need for additional complexity in the [standard] cosmological model and finds that the answer is no,” says Charles Bennett, a cosmologist at Johns Hopkins University, who was not involved in DES. “This does not mean that nature is not more complex but rather that we don’t have sufficiently compelling evidence for more complicated models.”
Yet, as Vincenzi points out, “the results are literally borderline between supporting the standard model and suggesting instead that the universe’s acceleration hasn’t been constant over time.”
Essentially, the team used more and better data to come to the same vexing conclusion that motivated its quest in the first place. Despite being the very best supernova-based estimate of dark energy yet performed, DES’s result is almost uncannily placed in the shrunken liminal space where certainty remains elusive. And so, even now, no one can say whether our laws of gravity need revision.
The Path to Illumination
Many of the same machine-learning techniques developed by the DES team could, in coming years, be applied to far larger datasets, which would put dark energy to an even more stringent test. Rather than finding fresh type Ia supernovae by the hundreds or thousands, facilities such as the Vera C. Rubin Observatory in Chile will likely find millions, which will demand more efficient ways to parse all those data.
“It wouldn’t be possible to follow up every single supernova live; there just are not enough telescope resources in the world,” Davis says.
Astronomers will also use observatories such as NASA’s upcoming Nancy Grace Roman Space Telescope to probe the universe’s expansion history even further back in time by studying supernovae and taking a host of other approaches.
“We have to measure the history of the universe’s expansion and structure growth in great detail to understand the nature of dark energy and potential modifications to Einstein’s theory of gravity,” says Yun Wang, a senior research scientist and dark energy expert at the Infrared Processing and Analysis Center at the California Institute of Technology, who was not involved in DES. Indeed, early results of various far-seeing surveys have already found one possible hint that dark energy is somehow more complex than the simple constant that cosmologists have long assumed. It’s “a problem called the Hubble tension,” Wang says. “Different ways of measuring the present-day expansion of the universe give very different answers. The modification of our laws of gravity is a possible solution to this tension, which may also be the origin of dark energy.” Sleuthing out the cause behind the strangely inconsistent measurements of the current cosmic expansion rate, which the DES analysis did not address, could offer clues to the dark energy puzzle.
“I find it very strange that the standard model works so well and explains a great diversity of precise data with only a few parameters but also has one and only one area of failure,” Bennett says. Many attempts to adjust the standard model so that it resolves the Hubble tension end up violating other physical laws that are well supported by observations, he notes. “It’s hard to change the model of the universe without affecting multiple measurable properties.”
For now, “the mystery of dark energy persists,” Wang says, “and the ultimate fate of the universe hangs in the balance.” If cosmic acceleration has a constant strength, the universe will expand forever; its fate is sealed. But if dark energy can vary over time, that opens up, well, a universe of other possibilities. “To me, it’s the most exciting problem in physics and astronomy today.”