As theory goes, less than a billionth of a billionth of a second after the Universe formed, it fractured.
Dubbed cosmic strings, those fractures are infinitesimally thin — a fraction of the size of a proton — and exceedingly long, spanning the cosmos, interwoven into the fabric of spacetime itself as topological defects.
Predicted in 1976 by Tom Kibble, the defects have long been beyond the reach of observational technology. Now, however, nascent gravitational-wave observatories like NANOGrav are providing new opportunity to look for them. Yet, so far, they’ve had no success.
Part of the difficulty in finding cosmic strings lays in first understanding what their signatures would look like now, from Earth, after billions of years of growing, straightening, crossing, emitting radiation, and dissipating energy.
To better understand these signals, PhD student José Correia and physicist Carlos Martins from the Centre for Astrophysics of the University of Porto simulate the evolution of spacetime following the Big Bang, looking for cosmic strings among hundreds of thousands of simulated configurations of the Universe.
Searching for signals
After forming in the early Universe, cosmic strings would have continued to evolve. They lose energy from their kinks, which straighten out over time under the strings' massive tensions. And, from time to time, they cross themselves and other strings, forming closed loops which detach, shrink, and radiate energy — sometimes in the form of gravitational waves.
It is primarily through these signatures that they become detectable.
“Their gravitational waves generate, let's say, voices,” Correia says. “The sort of noise a full network of strings will generate depends on the properties of the strings themselves — so how they move, the number of loops they form, and their small-scale structure, like kinks and cusps.”
This noise makes up only a small portion of the Universe’s background noise.
“There are a myriad of possible sources which can contribute to such backgrounds [i.e. the cosmic microwave background and the stochastic gravitational wave background]. Searching for strings is then much like trying to hear specific voices in a crowded street,” Correia says.
The various theories that expect cosmic strings give very different predictions on the strings’ properties and, thus, what their signatures could look like.
Using the laws of motion and mathematics of these theories, the simulations evolve various universes and output the string networks’ characteristics, like their mean velocity and string separation.
These simulated signals can be used to generate a description of the background noise that strings produce, giving astronomers an idea of what they should look for in their observational data.
Simulating the loops, kinks, and cusps of the strings brings unprecedented detail to cosmic string simulations. But, doing so was a challenge, as the researchers are not simply simulating entities throughout spacetime. They’re simulating spacetime itself.
“As you might imagine, it’s a bit awkward because I'm evolving fields, not strings. So, first, I have to find some way to understand where on the simulation’s cells the strings are, and then I output only string positions. If I try to output the entire lattice, it wouldn’t work; it's an insane amount of IO,” says Correia.
By outputting only unique cells, the researchers drastically reduce the amount of data written to disk. Their GPU-extension, presented at PRACE’s 2021 summit week, further improves the state of current cosmic string simulations. Making a memory tradeoff to maximize its throughput and bandwidth, the GPU-extension codes approximately 30 times faster than conventional CPUs.
As physics has focused on increasingly distant times and spaces, supercomputers have become increasingly essential to physicists, says Correia.
On the theoretical side, they help conceptualize and predict systems like warped spacetime — doing so over longer timescales and at higher precisions than ever before. And, on the experimentation side, they’re needed to process the petabytes — even exabytes — of data produced to test those theories.
“People talk about the interplay in physics between theory and observation, but in reality, it’s not a duality. It’s a trinity,” Correia says. “You also need supercomputing in there. Without next generation computational facilities, we’re basically stuck in ground zero. We've seen amazing discoveries in my lifetime: I never expected to see the Higgs boson. I never expected we would see gravitational waves from black hole mergers or a photograph of a black hole, which is a fantastic wallpaper by the way. And if you look, none of this extraordinary science is possible without next-gen HPC facilities."