Testing the Multiverse

August 3, 2011
by Miriam Frankel
Testing the Multiverse
Does the CMB sky show signatures of other universes?
by Miriam Frankel
FQXi Awardees: Hiranya Peiris
August 3, 2011
It sounds like a surreal dream, but Hiranya Peiris spends her days pondering being trapped inside a "Swiss cheese." It’s an analogy she uses to describe the multiverse theory, which states that our universe is just one of many—one hollow amongst many others, each separated by a cheesy spacetime. The question that Peiris, a cosmologist at University College London, is now attempting to tackle is: Housed as we are within our hollow, could we ever find scientific evidence that other hollows—or other universes—exist?

As fanciful as it may seem, the idea of the multiverse stems from one of the most established theories in cosmology: inflation. First proposed in the 1980s, inflation explains why distant parts of the universe look so similar. The idea is that the very early universe went through a period of extremely rapid expansion, flinging apart regions that were once in close contact—and hence appear similar—so that today they are vastly separated. Soon after the theory was developed, cosmologists realized that inflation may not have been a one-off event. Rather, different patches of spacetime could suddenly inflate into fully-fledged universes in their own right. Though inflation would eventually end within each patch, it would continue in the spacetime between them—hurling these universes so far apart that they will never be able to communicate with each other.

Peiris has spent much of her career painstakingly building up a dossier of observational support for inflation. In particular, she has scrutinized patterns in the temperature map of the first observable light of the universe, known as the cosmic microwave background (CMB), to verify that they match up with inflation’s predictions. Yet, it was not until fairly recently that she began to seriously search through CMB data for evidence of the multiverse. She had dismissed the idea before, not because she thought the model was wrong, but because she believed it was untestable. Like many physicists, she had assumed that any neighbouring universes lay too far away to leave any directly detectable signatures on the CMB sky.

Cosmic Collisions

That changed in 2009, when Peiris met string theorist Matthew Johnson, based at the Perimeter Institute in Canada. Johnson, along with FQXi director Anthony Aguirre and others, had been toying with the idea that a neighboring universe could leave readable signatures on the CMB, if it collided with our cosmos soon after it was born and before it was hurled away from us. (See "When Universes Collide" for more on cosmic crashes.) If correct, the multiverse theory would potentially be testable and Peiris, Johnson realized, would be the perfect person to help test it. "There are so many untested ideas out there," says Johnson, adding that most theorists—himself included—have no experience of working with observational data. "You’ve got to wait around for the people who actually know what they’re doing," he says.

The discussions sparked a strong partnership between Johnson and Peiris, who have been working together ever since. With the help of a $112,331 grant from FQXi, they have now developed an algorithm to search for the scars left by collisions between neighboring cosmoses.

There’s a lot of work being carried out on the multiverse and it’s important to know whether it’s testable or not.
- Hiranya Peiris
Alexander Vilenkin, an FQXi-member and cosmologist at Tufts University, who laid the foundations of much of the multiverse theory, praises the test as "the first of its kind." Sceptics have long dismissed the multiverse question as lying outside of the bounds of science because the theory made no directly testable predictions. But that has now changed. "There is nothing like a direct test," he says. (See "Measuring up the Multiverse" for more about indirect tests.)

Johnson and others had worked out that the collision imprints should look like disks on the sky, formed where two spherical universes temporarily intersected. But knowing what to look for is just the first step. ESA’s Planck satellite, and NASA’s Wilkinson Microwave Anisotropy Probe (WMAP) before it, offer cosmologists a wealth of CMB data to sift through. So where do you begin to look?

"It’s a very hard statistical and computational problem to search for all possible radii of this disk-like signature at any possible place in the sky," says Peiris. "But that’s what picked up my curiosity."

Tantalizing Hints

The team ran simulations of what the sky would look like with and without cosmic collisions and developed an algorithm to determine which fit better with actual WMAP data (see image, below). They found four tantalizing features in the data that are better explained by collisions than without them. Their work appears in Physical Review Letters. (For a detailed description of the analysis, see Matthew Johnson’s post, "New Year, New Universe.")


Simulated Collision Signatures
A collision (top left) induces a temperature modulation in the CMB temperature
map (top right). The "blob" associated with the collision is identified by
a large needlet response (bottom left), and the presence of an edge
is highlighted by a large response from the edge detection algorithm (bottom right).
In parallel with the edge-detection step, the team perform a Bayesian parameter
estimation and model selection analysis.

Credit: Feeney et al., arXiv:1012.1995v3
Peiris is well aware that the human eye is very good at cherry-picking patterns in the CMB data and stresses that, by contrast, the team’s algorithm is harder to fool. "The algorithm includes the possibility that it’s all just a coincidence," she explains. "It implements Occam’s razor—the idea that a theory with more complexity, like collisions, should be penalised for its complexity if it doesn’t fit the data." In essence, the algorithm demands a much better fit to the data for a complex model than for a simple one.

Peiris and Johnson also emphasize that their findings are not strong enough to rule out the possibility that the features were not created by collisions. But they have high hopes that new data from Planck will help settle the issue. Johnson estimates that once the required the data comes in, hopefully in 2013, it should only take a few weeks to get an answer. The team also plans to investigate how a collision would affect the polarisation of the CMB in the impact region.

If the Planck satellite confirms the vague features indicated by WMAP, it could eventually persuade the bulk of physicists to favor the collision model, says Vilenkin. "After a few years, if it is solid data and solid evidence, people will get convinced," he says.

Peiris is more reserved, noting that even if the results are corroborated, "one dataset" may not be enough to convince everyone that the multiverse exists. "The sceptics are right to be sceptical," she says. "They would probably need to apply their own methods or reproduce the analysis."

But what about a negative result?

Peiris argues that the credibility of the multiverse as a scientific theory is raised just by the very fact that cosmologists can now observationally test its predictions—regardless of the final outcome: "There’s a lot of work being carried out on the multiverse and it’s important to know whether it’s testable or not," she says. "I think it will be a very important and interesting result either way."