Nearly 400 years ago, Galileo found the sun to be more than a hot disk on a flat sky. Continuing in that tradition, the University of Chicago's Andrea Malagoli and colleagues use supercomputers to simulate the sun's behavior, which comes out in active and quiet personalities.

This article is the ninth in a series on NASA High-Performance
Computing and Communications (HPCC) research teams.

Galileo Galilei, full-time astronomer, offered this observation about the sun in the 1600s: "Spots are on the surface of the solar body where they are produced and also dissolved, some in shorter and others in longer periods. They are carried around the sun, an important occurrence in itself."

As Galileo knew better than most, things are not always what they appear to be. The sun is more than a hot disk on a flat sky. Instead, Galileo found a fiery ball spinning in space with spots dancing across it.

His subsequent campaign for a sun orbited by the planets challenged authorities of the day, and Galileo, who some call "the first scientist," lived his last years under house arrest.

History may be repeating itself as another struggle ensues between astrophysicists who rely on supercomputer simulations to analyze the sun and old-school astronomers who rest on empirical data alone.

Making sense of data

The problem with empirical data is it can only take you so far. Generations of astronomers vindicated Galileo as they built upon his work. Today, ground-based telescopes and spacecraft capture the sun's basic make-up and highly dynamic surface.

Astronomy's dilemma is that these observations say little about what happens inside the sun, the source of the energy and magnetism that permeates the solar system.

Financed by NASA, HPCC research leader Andrea Malagoli studies mechanisms that create these forces and motions that pull them through and out of the sun. "You need to understand the physics behind the observations," says Malagoli. The University of Chicago astrophysicist insists that "supercomputer simulations are the only tool observational people have to make sense of data and go on to plan new missions."

Malagoli and his collaborators have confidence in their simulations because they reproduce observed physical features. These include a granular surface pattern and variations in temperature and magnetic fields.

Despite these encouraging results, the research community has long resisted accepting simulations as scientific discoveries. Resistance to new methods, as we know, is as old as the scientific method itself and dates back to when Galileo was collecting the first data on the sun.

"The problem is when you do simulations, typically you work on well-known equations and physical principles," Malagoli explains. "You are not developing any new equation, nor are you deriving rigorous mathematical solutions to existing equations."

"In many ways computation is actually closer to an experimental science," he adds. "So, the more conservative physics environment is resisting this approach because they think you're much better off doing some back-of-the-envelope calculations or some very abstract model that often tells you little about reality. There is no question that all the recent advances in understanding solar physics and other parts of astrophysics come from simulations."

The active sun

Simulating the sun is no simple matter. Its sheer enormity presents one challenge. Nearly 110 Earths could line up across its equator and 1.3 million Earths could fit inside. Complicating things further is that simulations must treat motions ranging in size from the sun's 1.4 million-kilometer diameter down to less than one kilometer.

Add all the physics equations to be solved, and "current supercomputers are almost like little pocket calculators compared to what you would really need to accomplish the task," Malagoli says.

To get around that problem, NASA's HPCC Program supports Malagoli's research team in developing multiple simulation codes to analyze different aspects of the sun's behavior, which comes out in two distinct personalities.

One personality is active, producing sunspots and spectacular outbursts such as solar flares over an 11-year cycle of boom and bust. The other personality is quiet, a day-to-day, workhorse production of heat and light.

Playing a clear role in the active sun is magnetism. X-ray observations show large magnetic fields accompanying dramatic surface phenomena. Some of these events inject the solar wind with magnetic and charged particles that can knock out satellites, and occasionally power grids, when blowing past Earth.

Observations stop at the magnetic fields' presence, however, and don't tell how they are created or how they move to the sun's surface. Understanding these issues will ultimately contribute to predicting when potentially destructive events will occur.

In work destined for coupling with solar weather forecasts, Malagoli's University of Colorado colleagues Juri Toomre and Nic Brummell are simulating the large-scale magnetic field involved in the active sun.

They have been testing the widely held theory that portions of the sun rotating at varying speeds pump its magnetic field from a big storage area in the middle. While the basic mechanism has been known for 40 years, the computations are the only means to study specific effects.

Malagoli, who hails from the same northern Italian village as opera star Luciano Pavarotti, likens this solar process to making pasta. "The way you produce a magnetic field is how you produce noodles. You stretch your dough," he says. "Stretching and folding creates strong lines of magnetic field. We haven't gotten the details right. That is where most of our work comes in and is why we need these big simulations."

The quiet sun

In the last few years, observational tools have gotten powerful enough to find the quiet sun generating magnetic fields of its own. While the amounts are much smaller, these fields seem to be important in getting heat out of the sun so it can supply, either directly or indirectly, all the Earth's energy.

Simulations are piecing together the elements of this story. The sun creates energy by fusing hydrogen protons within its searing core (approximately 14 million degrees Celsius). Radiation transports the energy to the next, cooler region. About two-thirds of the way to the surface, the physics changes and radiation cannot move all the energy anymore.

Matter throughout the sun is plasma, a very hot gas of freely floating protons and electrons. The plasma in the star's upper third must start boiling, or convecting, to carry energy. "That is why the sun is so interesting, because you deal with so many different issues, depending on what layer you're studying," Malagoli says.

Like a wider telescope mirror, a faster computer can see a bigger portion of a celestial object or focus its resolution on smaller details. Because there are trade-offs in the complexity of physics one can simulate, Malagoli's team pursues both options.

The University of Minnesota's Paul Woodward and David Porter simulate convection throughout the sun's outer ring, techniques they have applied to studying other stars, including red giants. At the University of Chicago, Fausto Cattaneo and Anshu Dubey's simulations take on a smaller chunk to determine how and where these motions produce magnetic fields -- the well-known solar dynamo phenomenon.

"By using a big enough computer, we can set up a system we believe roughly reproduces working processes responsible for generating magnetic fields," says Cattaneo, an astrophysicist originally from Rome.

Cattaneo and Dubey's latest calculation considers a region about 50,000 kilometers across and 6,000 kilometers deep. It represents the plasma as a box of fluid divided into 134 million smaller boxes. The supercomputer solves the physics equations in each box and advances the solutions 100,000 steps in time.

With all that number-crunching, three hours of simulated solar time consumes three weeks of NASA HPCC's 512-processor CRAY T3E at Goddard Space Flight Center. Running at 50 billion mathematical operations per second, it generates more than four trillion bytes of data, an output equal to three million books.

Ten times the size of the team's first NASA-supported simulations in 1993, "this is probably the largest simulation to date in studying magnetic fields," Malagoli says. "That's why it is important to go through all these performance milestones with HPCC."

The simulation begins with the fluid not moving. As if turning up a stove burner, heat from below starts boiling the fluid. The surface becomes a meeting place for rising, hot fluid and sinking, cooler fluid.

One reason the simulation takes so long is that it must continue until the convective motions repeat themselves. After all, the sun has been doing that for millions of years!

Magnetic fields arise because the simulated fluid mimics the plasma's ability to conduct electricity. "It is a certainty that churning a highly conductive fluid will generate a magnetic field," Cattaneo says. Alleys of cooler, sinking fluid concentrate the magnetic fields, which flail around like spaghetti.

As it follows these dynamics, the simulation begins to resolve the paradox of the sun's atmosphere being nearly one million degrees hotter than its surface. "There's something really bizarre going on. It is not conduction or radiation because they move energy down a gradient. You cannot convince heat to go the wrong way!" Cattaneo says.

"We're finding that it's all this evolving of magnetic fields that keeps bumping heat into the atmosphere, and that's why it's so hot," he explains. "The total contribution to this emission doesn't just come from the active regions; it comes from magnetic fields all over the surface of the sun."

One step ahead

The phenomena probed by the quiet sun calculation are beyond observations' limits, both physical (even spacecraft cannot see below 100 kilometers or so) and temporal (convection's magnetic fields last only five minutes).

"I challenge you to find someone who by hand can find a solution of the equations that resembles anything like that," Malagoli stresses.

Simulations keeping ahead of observations also can contribute to future missions, as Cattaneo describes. "You design an instrument to measure things you believe are important. To know what to measure, you have to have some idea of what is going on," he says. "That understanding comes from setting up simulations. If you have no idea what you're looking for, chances are you're not going to find it."

Cementing the relationship between observation and simulation, this HPCC team's results are serving as reference models for planning NASA's Solar-B, a spacecraft slated for launch in 2003.

"As long as there are missions, you will need HPCC," Malagoli says.

Credits for Insights Magazine go to the following people along with the writers and photographers who contribute to each issue and the researchers and specialists whose material is highlighted:

Program manager: Dr. Eugene L. Tu
Deputy program manager: Dr. William R. Van Dalsem
Editor, writer, photographer and designer: Judy Conlon
Associate editor: Sue Reynolds
Web designer and programmer: Sue Cox

Insights was published by the HPCC Program office and produced by Raytheon contractor staff at NASA Ames Research Center.