The startling announcement in 1998 that the universe is expanding at an accelerating rate — followed not long after by evidence that a mysterious "dark energy" filling the universe is responsible for this acceleration — was based on comparing the redshifts and magnitudes of numerous Type Ia supernovae, exploding stars of extreme and remarkably uniform brightness.

	A Type Ia supernova discovered in 2001 (Photo: Ted Dobosz).

The quality and quantity of data on these and other types of supernovae, both at high and low redshifts, will increase exponentially in the next few years because of two experiments based at Berkeley Lab: the well known Supernova Cosmology Project and a newer program, the Nearby Supernova Factory. Their goal is to determine the underlying physics behind these catastrophic events, so that they can be used to understand dark energy and other cosmological questions.

"The only way to fully exploit the power of the amazing data now being collected will be to make an equally dramatic improvement in computational studies of supernovae," says Peter Nugent, an astrophysicist in the Scientific Computing Group at the National Energy Research Scientific Computing Center (NERSC). "Unfortunately, current computational resources only allow us to fully model supernova atmospheres in the spherically symmetric regime."

Researchers have done their calculations as if supernovae explode spherically, that is, the same way in every direction, in order to simplify computer modeling. Recently, however, Nugent was awarded a three-year, $500,000 grant by NASA's Astrophysics Theory Program to help expand supernova modeling to the third dimension. "The more we look at supernovae," he says, "the more we realize that assuming they are 1D explosion events is a horrible oversimplification, which might lead us down some very wrong paths concerning the nature of their progenitors and their explosion mechanisms."

Nugent will take advantage of coming increases in the computational power of NERSC's massively parallel supercomputers, as well as improvements in the algorithms used to calculate supernova hydrodynamics and processes like radiative transport.

"We'll be able to put tight constraints on the characteristics of supernova progenitor stars, including their luminosities, explosion mechanisms, and nucleosynthesis products, once we can compare computed spectra based on different progenitor models with high-quality observations from the Supernova Cosmology Project and Nearby Supernova Factory," Nugent says.

Modeling will also address other effects, such as the possibility that different compositions of the progenitor star, owing to the evolutionary stage of the universe when the star formed, might somehow bias the determination of cosmological parameters.

Type Ia supernovae in particular are complex events. They begin as white dwarf stars made of carbon and oxygen, accreting a companion star's material. Eventually they accumulate a critical mass, 1.4 times the mass of the sun: deep within the white dwarf, carbon fuses, setting off a runaway thermonuclear explosion. Despite whatever differences there may be among progenitor systems, the critical mass is the same in all cases, so Type Ia supernovae have very similar spectra and rising and falling light curves.

Type Ia supernova explosion creates an abundance of intermediate-mass elements, requiring a transition from subsonic deflagration to supersonic detonation.

A Type Ia ejects material at velocities approaching 20 percent of the speed of light and, unlike other types of supernova, produces a great many elements of intermediate mass between carbon and nickel, which are created at different densities. To account for this, different models invoke scenarios of explosions involving different rates at which a relatively slow-burning deflagration front makes the transition to a supersonic detonation wave.

An important new source of information that could provide insight into these processes is the supernova polarization data taken by Lifan Wang and Andy Howell of Berkeley Lab's Physics Division. Different regions of an exploding star's atmosphere may be moving at different rates or in different directions; some regions may be richer or poorer in various elements; viewed from a particular angle, the remnants of a companion star or an accretion disk surrounding the progenitor may block or filter some of the supernova's light. These details leave their mark as different orientations of the polarized light in different spectral lines of the exploding star.

"One of our basic goals is to start making models in 3D that exploit the new polarization data," Nugent says. "Polarization allows you to probe the geometry of the event."

While polarization data show that the geometries of other kinds of supernovae vary more than Type Ia supernovae and that their differences persist longer, a Type Ia can still reveal much about itself through polarization if it is caught early. Despite their apparent similarity, Type Ia's are not identical.

Already Nugent and graduate student Daniel Kasen have begun constructing models that use polarization to explore supernova geometries, including determining the angle at which these ancient cataclysms have been observed.

"For the first time we are learning details about supernova progenitors from the data," Nugent says. "If what we're seeing proves out, it may even create shockwaves right here in the supernova community."