Published online 2 June 2010 | Nature 465, 534-535 (2010) | doi:10.1038/465534a


Model stars set to explode

Realistic computational models of supernovae might soon solve a long-standing mystery.

Giant, mushroom-shaped blobs of nickel speed out from the blistering core of the supernova at 4,000 kilometres per second, piercing the smooth, ballooning sphere of hydrogen. The effect looks something like a cosmic hernia, but astrophysicist Hans-Thomas Janka couldn't be more pleased with what his computer model is showing.

This three-dimensional model shows the evolution of a supernova in its first half-second. The explosion has a 200 km radius in the first frame; in the last frame, it is 1,900 km.This three-dimensional model shows the evolution of a supernova in its first half-second. The explosion has a 200 km radius in the first frame; in the last frame, it is 1,900 km.H-TH. JANKA, MAX PLANCK INST. FOR ASTROPHYSICS

In a paper published in the 10 May issue of the Astrophysical Journal1, Janka and his colleagues from the Max Planck Institute for Astrophysics in Garching, Germany, used their model to address a long-standing puzzle: how do heavier elements, synthesized in the core of the massive, dying star, get out of the explosion before the lighter stuff that sits in the star's outer shells?

But the simulated explosion, with its colourful protrusions, is significant for another reason. It marks the first published, nearly complete, three-dimensional (3D) model of a supernova. Simpler one-dimensional (1D) and two-dimensional (2D) models, which assume that an explosion unfolds symmetrically, fail to get ticking star bombs to blow. So a handful of computational astrophysics groups, including Janka's, are moving up to the third dimension. The computational demands of tracking a complex and rapidly evolving explosion in which densities, temperatures and velocities are all at physical extremes are themselves astronomical. And the researchers have yet to tackle the toughest challenge: modelling the first milliseconds of the explosion in the innermost core of a supernova.

But when they do, astrophysicists hope their models will finally reveal the root mechanism that triggers supernovae, a problem that has eluded their profession since modelling began 40 years ago. "I think there's overwhelming evidence that going to 3D calculations makes the explosions easier," says Eliot Quataert, who has been following progress in the field as director of the Theoretical Astrophysics Center at the University of California, Berkeley. "That's what everyone is rushing after now."

Supernovae are among the most energetic events in the Universe, briefly outshining entire galaxies as they synthesize and spread the heavy elements that are essential to biological life. One type of supernova detonates when a dense white-dwarf star sucks in enough additional mass for a fusion explosion. The second, and more mysterious, type occurs when a star a few times more massive than the Sun runs out of nuclear fuel. As the star's internal pressure plummets, its massive core collapses into a neutron star just a few tens of kilometres across. At this point the neutrons can be compressed no further, and the core rebounds, sending a shock wave out against the crushing weight of the star's outer layers. But the shock wave by itself isn't enough to blow the star apart, and for a fraction of a second it stalls. The question is what 'revives' the shock, causing the explosion to proceed.

“For a fraction of a second a supernova stalls. The question is, what revives it?”

Some have suggested that energy locked up in the tightly coiled magnetic fields of a spinning core might provide the extra nudge. But many theorists favour a trigger from neutrinos, which are produced in unimaginably large numbers when the neutron star is born. Because of their weak coupling with other forms of matter, the vast majority of the neutrinos simply fly straight through the dying star. But if enough impart their energy to the infalling matter, they could provide the extra kick needed to restart the shock wave and blow up the outer shells of matter. "We're at the point now where we believe the final threshold is being crossed for the neutrino model," says astrophysicist Stan Woosley at the University of California, Santa Cruz, who is leading a 3D modelling effort. "A lot of us think it's going to work when we take this next step."

Multi-dimensional models

But modelling the interactions of the neutrinos with the infalling matter is the most computationally difficult part of the problem; until now, modellers have had to use lower-dimensional short cuts. Not only have these models generally failed to ignite the supernova, but they also preserve unrealistic symmetries, with the progenitor star maintaining its onion layering as it blows up: hydrogen and helium first, followed by heavier elements. The spectacular 1987 supernova in the Large Magellanic Cloud was heralded by a burst of X-rays and γ-rays coming from radioactive heavy elements such as nickel — meaning that the heavy material had somehow punched through the lighter layers. "That was the point when theorists realized that 1D models cannot explain the physics that was observed in the explosion," says Janka.

The rapid increase in available computer power now allows the modellers to run more-ambitious and realistic simulations in a manageable time — currently, Janka needs two to three months to run one supernova in 3D. At some point, researchers say, a digital cataclysm will attain a level of detail adequate to reproduce what occurs in the real Universe. "That is what the big competition is about," says Janka.

Optimistic outlook


Adam Burrows, an astrophysicist at Princeton University in New Jersey, who is working with Woosley, has already analysed the benefits of 3D models. In 2008, he showed that going to 2D made a simulated supernova of a given luminosity blow up 1.4 times as often as the 1D version2. Repeating the analysis, he found that 3D models should make supernovae go off twice as easily as in 1D models. The reason, he says, is that 3D allows the infalling matter to take random walks in all directions — which means that it spends slightly more time interacting with neutrinos and absorbing their energy. Ultimately, the aim is to construct a 3D explosion that is faithful enough to generate a supernova without fail. In the process, the physical forces that govern it should become apparent.

That's why Janka is rushing to fill out the rest of his 3D model to include the neutrino mechanics in the first second of the explosion. He also has to find a computer that can handle the job. He estimates that he needs a computer to perform about 1021 floating point operations, or flops — roughly 50,000 times the computing power behind his latest simulations. This would mean several months of devoted time from a major supercomputer that can perform a petaflop, or 1015 operations, per second. With the growth in worldwide computing infrastructure, Quataert says, that is now not a completely outrageous request. "That's part of the reason for optimism that there will be a breakthrough in this problem," he says. 

  • References

    1. Hammer, N. J., Janka, H.-Th. & Müller, E. Astrophys. J. 714, 1371 (2010). | Article
    2. Murphy, J. W. & Burrows, A. Astrophys. J. 688, 1159 (2008). | Article
Commenting is now closed.