Nature rarely presents tidy, manageable figures. Take the hundreds of billions of stars in our Milky Way galaxy or the hundreds of thousands of individual atoms in a single virus. Scientists endeavor to describe all the properties and behaviors of every component within these complex assemblages.

That is why supercomputers are revolutionizing our understanding of complex physical systems. Simulations running for days on room-size machines have given visual form to the previously unimaginable, from boundless cosmic realms to the infinitesimal constituents of matter. Such simulations are more than intellectual amusements. They are helping researchers study how drugs work against the swine flu virus and how space and time warp around colliding black holes. Here, then, is a look at supercomputer-generated visualizations, to better see the secrets and beauty of our world.

This image shows a supercomputer simulation run at the Department of Energy's SLAC National Accelerator Laboratory in Menlo Park, California, which revealed that double stars with relatively low masses might have formed very early in cosmic history, just 200 million years after the Big Bang. (Prior simulations of the primordial universe had suggested that the first stars were mostly lone giants, up to 300 times as massive as our sun.)

The simulation contains 8*10^52 cubic miles of gas- and dark matter-strewn space. Lighter colors represent higher densities, marking the locations of two developing low-mass stars.

When an aging star about 10 times as massive as our sun runs low on light, easy-to-fuse elements, its nonfusing iron core collapses. The star's interior then rebounds like a spherical piston, exploding into a type II supernova. Using a supercomputer at the Leadership Computing Facility at Oak Ridge National Laboratory in Tennessee, astrophysicists have shown that neutrinos play a key role in powering the shock wave that tears through the star. This rendering shows the hydrodynamic flow of matter in a giant star's core. Tan areas show higher entropy, or heat; blues and greens are comparatively cooler surges of matter.

Last spring, with fears rising over the H1N1 swine flu virus, researchers from the University of Illinois at Urbana-Champaign and the University of Utah were granted emergency supercomputing time at the Texas Advanced Computing Center in Austin. Their goal: to model the virus's possible resistance to medications such as Tamiflu (white molecule, center of image). These antiviral drugs bind to a surface protein, called neuraminidase, that influenza uses to infect host cells.

Two weeks of continuous calculations on the supercomputer showed that the neuraminidase-binding site has a mostly negative charge (red); it is surrounded by positively charged regions (purple), with a narrow negative-charge pathway for the drug to follow. Mutations to proteins along that delivery route could make the H1N1 virus resistant to drugs, although for now it remains vulnerable.

For decades scientists have sought to generate clean energy by instigating the kind of sustained nuclear fusion reactions that power the sun. On Earth this feat requires taming plasma (electrically charged gas) at temperatures around 150 million degrees Celsius, 10 times as hot as the inferno at the sun's core. The most ambitious attempt: the 10 billion euro International Thermonuclear Experimental Reactor (ITER), being built in France.

Recent computer simulations incorporating the geometry of ITER's doughnut-shaped reaction chamber and its powerful, plasma-confining magnetic fields have helped researchers predict the behavior of turbulent plasma inside the machine. The long fibers seen here represent the concentration of electrons in the plasma flowing through the chamber, with red and orange showing higher-density regions. Because plasma makes up more than 99 percent of the visible universe, such simulations also offer insight into a wide range of astrophysical phenomena, including stellar formation and the energetic events that create cosmic rays.

Among the most powerful but as yet undetected events in the universe are the mergers of supermassive black holes. These billion-solar-mass monsters, residing in the centers of large galaxies, can glom together during galactic collisions after a spiraling dance, as shown here. The colorful bands represent propagating gravitational fields, while the gray spheres indicate the black holes' event horizons, the boundary from within which not even light can escape.

Albert Einstein's general theory of relativity predicts that black hole mergers should send out intense blasts of gravitational waves, ripples in space-time. Scientists are learning how to detect and recognize those waves by studying supercomputer models run at two NASA campuses, the Ames Research Center at Moffett Field, California, and the Goddard Space Flight Center in Greenbelt, Maryland. The simulations reveal that the recoil from the combining of black holes could shoot the resulting merged supermassive black hole right out of its galaxy.

People with Parkinson's disease experience muscle tremors and rigidity because something kills off the brain cells that normally make the movement-coordinating chemical dopamine. The prime suspect is a protein molecule called alpha-synuclein, which can bunch into long, destructive fibers. Precisely how it does so is not clear, however.

To investigate, scientists at the University of California at San Diego ran supercomputer simulations using 962,757 processor hours to explore myriad possible shapes that alpha-synuclein molecules can take. The simulations revealed that the protein aggregates into ringlike structures that perforate cell membranes (the green areas in the image), leading to cell death and triggering Parkinson's. These studies have paved the way for the development of new aggregation-prevention drugs.