ACADEMIA

3D simulations at TACC force scientists to reevaluate the Sun's composition

What would happen if the yardstick that astronomers used to measure the universe were too long?

This is what Carlos Allende Prieto, with colleagues David Lambert (director of the University of Texas McDonald Observatory), and Martin Asplund (director of the Max Planck Institute for Astrophysics), proposed when they published their 2001 paper, "The Forbidden Abundance of Oxygen in the Sun” in the Astrophysical Journal, stirring up a significant controversy in the world of astronomy.

Allende Prieto’s investigation of the chemical abundance of the Sun, based on emerging 3D models of the solar atmosphere, suggested that the amount of carbon and oxygen in the Sun was 30 to 40 percent lower than previously believed.

Since the chemical make-up of the Sun is frequently taken as a reference point in formulating interpretations of measurements for the composition of other objects in the Universe, many models that relied on the higher abundances were put into question by Allende Prieto’s assertion. A dozen rebuttals appeared in scientific journals.

Recently, more comprehensive simulations, performed at the Texas Advanced Computing Center (TACC) using a more powerful parallel computing code, proved that Allende Prieto’s measurements were accurate. This discovery is leading to new notions about our cosmic evolution, as well as a reevaluation of the distinctiveness of the Sun.

“Everything we know from objects in the universe comes from the analysis of light,” said Lars Koesterke, Allende Prieto’s collaborator and a research associate at TACC. “We analyze the light of stars to figure out what they’re made of, what their temperature is, and how much energy they emit.”

What that light means isn’t exactly self-evident. Certain characteristics, like an object’s color or intensity, give us clues about the source of the light, if we interpret them correctly. To do so, astronomers developed a method called spectral analysis, whereby visible light is refracted and analyzed to assess the concentration of a given chemical species in an astronomical object.

By breaking down light into optical bands signifying different chemical elements — hydrogen, helium, carbon, oxygen, etc. — and comparing this spectrum with models of the Sun, astronomers can accurately determine the solar abundance.

Or so they believed.

For decades, scientists had been using one-dimensional models of the Sun to perform spectral analyses.

“In a one-dimensional atmosphere of a star, the models assume that everything is static, frozen,” said Allende Prieto, a researcher at the Institute of Astrophysics of the Canary Islands (IAC). “In reality, because too much energy is being emitted to be quietly radiated away, everything is moving around and you have this boiling at the surface. That changes the dynamics, the energy balance, and the appearance of the spectrum.”

By integrating a new three-dimensional model of the Sun into his spectral analysis, Allende Prieto obtained a spectrum that indicated lower amounts of carbon and oxygen than those determined by solar seismologists, who infer the interior of the Sun by observing the oscillations in the atmosphere, and by earlier studies. It appeared to Allende Prieto that, with the 1D model, one of the most important spectral lines in the Sun, representing oxygen, was entangled with a line representing a different chemical element, nickel — a proposition first suggested by Lambert.

“The strength looked just fine with 1D models. But because the 3D models can better reproduce the observed shape of the line profiles, it was obvious that there was something wrong,” said Allende Prieto.

This huge change in chemical abundance is altering prevailing theories about the structure and evolution of stars. Since the chemical composition of galaxies, other stars, and planets are compared to the Sun, a change to the Sun’s chemical abundance means the composition of the entire universe needs to be recalibrated.

Furthermore, chemical compositions are one of the main pieces of evidence used in determining the narrative of our galaxy’s evolution: the cycle of birth and destruction that led to the creation of Earth and its heavy chemical elements.

“If you believe that there’s now less carbon and oxygen, then our view of the chemical evolution of the galaxy has to be changed,” said Koesterke. “The total amount of metals each generation of stars produces might be lower than previously thought.”

Allende Prieto’s conclusions led to much opposition in the field. The evidence, critics suggested, was based on a small fraction of the spectrum and used unproven models and codes. It was therefore too limited and fragmentary to be definitive. In the following years, additional evidence was gathered from many more spectral lines, as criticism concentrated on the lack of a thorough test of the structure of the 3D atmospheric models.

Allende Prieto would have liked to present more proof, but the simulations had required all of the computer processing power available to produce just a few lines of the spectrum, and these were the chief pieces of evidence buttressing the new hypothesis. The analysis of the full spectrum, using 3D models, was predicted to require several million times more computing power than the analysis of equivalent 1D models.

Which is to say: impossible.

It was at this time, in 2004, that the McDonald Observatory hired Lars Koesterke to assist with Allende Prieto’s problem. Working over a period of two years, Koesterke created ASSET (Advanced Spectral Synthesis 3D Tool), a tool that models the solar atmosphere in three dimensions and takes snapshots from various time-evolutions to find the average spectral signal.

The serial version of ASSET was good, but by parallelizing the algorithms and optimizing its structure, Koesterke made the code much faster and able to use thousands of CPUs at the same time. Applying the vast resources at TACC, full calculations in 3D are now accomplished in just a day, the same amount of time it takes a desktop to perform 1D calculations.

“Suddenly, we were able to compute the whole solar spectrum, something that seemed utterly impossible five years ago,” Koesterke said.

In 2008, as a consequence of this speed-up, Koesterke and Allende Prieto proved definitively that the initial assessment of chemical abundances was no fluke. The pair published several papers describing this work [see list below] and increasingly, the new abundances are being accepted and integrated into solar models.

“There are still researchers who are totally opposed to the new solar abundances because they cannot match them with their own results,” said Koesterke. “But many scientists now agree with our work assessing the abundance to be a lot lower than previously believed.”

This debunks the once prevailing idea that the Sun is significantly different in composition from other stars. The reduced carbon and oxygen abundances align the Sun more closely with the rest of the stars in the vicinity, and suggest a common galactic history.

Today, with fast and powerful methods in place, the astronomers are extending their research in new directions. They have simulated the spectra of 73 additional stars using models developed by a team in Germany, and are now evaluating the effects that their 3D geometry modeling imprints in the light.

In collaboration with Ivan Ramirez of the Max Planck Institute, they have applied these techniques to study stars that are cooler than the Sun. Not surprisingly, researchers have found that the 3D spectra computed with ASSET match observations of stars much better than the 1D calculations.

Starting in 2012, the ASSET code will be used for the European Space Agency’s Gaia mission. The mission will map a billion stars, or nearly one percent of the galaxy. Data from Koesterke’s 3D spectral synthesis tool will inform researchers about the observed stars’ velocities and compositions, giving them precise knowledge about the overall makeup of the galaxy, which is, after all, the goal of astronomical measurements.

“A good fraction of astrophysics relies on getting the chemical composition of the stars right. If the huge revisions to carbon and oxygen abundances we’ve seen with the Sun are waiting for us with other stars, then there will be exciting surprises,” said Allende Prieto.

“We now have the right computing power, and the right code, so we’re going to be able to do better science.”

Further reading:

"The Forbidden Abundance of Oxygen in the Sun” (2001), Carlos Allende Prieto and David L. Lambert

"Granulation in K-type Dwarf Stars. II. Hydrodynamic simulations and 3D spectrum synthesis" (2009), I. Ramirez, C. Allende Prieto, L. Koesterke, D. L. Lambert, M. Asplund

Koesterke’s slideshow presentation from the ”Recent Directions in Astrophysical Quantitative
Spectrosopy and Radiation Hydrodynamics” conference (2009)

"Quantitative Spectroscopy in 3D" (2009), L. Koesterke