Oak Ridge Leadership Computing Facility

When German physicist Max Planck created quantum theory in 1900, he was not trying to revolutionize the world. He was just trying to provide a theoretical foundation for the way a heated object radiates energy and, thereby, to improve the efficiency of light bulbs.

Modern scientists, too, cannot know which research will change the world, but history suggests the next scientific revolution will focus on complex systems rather than isolated bits of matter. If so, it may stem from computer simulations such as those performed by Tommaso Roscilde of France’s École Normale Supérieure de Lyon and recently of Germany’s Max-Planck Institute. A team led by Roscilde is using Oak Ridge National Laboratory’s (ORNL’s) Cray XT4 Jaguar supercomputer to explore the quantum mechanical phenomena that give us superconductors and superfluids.

Roscilde and his teammates—Stephan Haas of the University of Southern California and Rong Yu of ORNL—are using Jaguar through the Department of Energy’s INCITE [Innovative and Novel Computational Impact on Theory and Experiment] program. A grant for 800,000 processor hours in 2007 allowed the team to simulate a lattice of atoms in a quantum magnet to examine two extraordinary quantum phases, or states of matter.

In the first, called Bose-Einstein condensation, atoms throughout the material occupy the same state, with the same momentum, range of probable locations, and spin. In a quantum system, this is the closest they can get to being in the same place at the same time. By introducing impurities into the material, the team is also able to create the second phase, known as Bose glass. In a Bose glass, the impurities force the condensation into separate islands throughout the lattice, with atoms sharing the same state only with other nearby atoms.

These two images illustrate the phase transition from a Bose glass to a superfluid in a diluted quantum magnet that has been exposed to an increasingly powerful external magnetic field. The size of each ball represents the local magnetization of that lattice site, which corresponds to the local density of bosonic quasiparticles. In the first image (Figure 1), the system is in the Bose glass phase at a field H = 2.79 tesla, where the quasiparticles are rare and localized in the yellow areas. At the higher field H = 2.90 tesla (Figure 2), these quasiparticles can move all over the lattice through a connected network shown in the yellow-to-red areas, giving the system a superfluid nature. Image source: Oak Ridge National Laboratory

Roscilde’s team is using a technique called Quantum Monte Carlo to simulate disorder in a quantum magnet and thereby create Bose glass. The team hopes its efforts will allow its collaborators—Vivien Zapf and Marcelo Jaime at the National High Magnetic Field Laboratory at Los Alamos National Laboratory—to perform the first experimental confirmation of Bose glass.

The work is at the cutting edge of condensed-matter science.

“I find that in this particular instance of a study of a solid-state system, you’re really trying to tailor matter to a level of control that was unthinkable a few decades ago or even a few years ago,” Roscilde said. “What you have is a system where, in principle, you can tune the system among completely exotic phases that have no analog in classical systems.”

Roscilde, like Planck a century before, cannot say whether his work will produce benefits beyond a deeper understanding of the universe. He noted also that a device based on quantum systems would likely be very different from existing technologies.

“You have to think hard what to make out of these systems,” he said. “It’s not just that once you know them, you know what to make out of them. You have to totally think of new functionalities.”