|
Number 496, July 24, 2000 by Phillip F. Schewe and Ben Stein
TWO-DIMENSIONAL TURBULENT FLOWS LEAK SIGNIFICANT ENERGY to their surroundings, new experiments have confirmed, providing insights that may improve models for the atmosphere and other fluids that move predominantly in two dimensions. As noted fluid dynamicist--and former Einstein postdoc--Robert Kraichnan pointed out in the 1960s, 2-D turbulent flows are remarkably analogous to Bose-Einstein condensates. That's because the vortices in these flat fluids can organize themselves and act in a coherent manner. Perhaps the clearest manifestation of this coherence is the "inverse cascade" effect: small vortices coalesce into larger and larger vortices. This effect partly contributes to the formation of large-scale circulation patterns in the atmosphere, which can be considered a 2-D fluid. The inverse-cascade process would continue unchecked, until a single vortex enveloped the entire fluid, if not for the fact that the fluid dissipates a lot of its energy. But researchers were unsure if this energy primarily went to the fluid's internal viscosity (resistance to flow), or to its surroundings, such as air molecules which exert an external friction. In a new experiment, University of Pittsburgh researchers (X.L. Wu, University of Pittsburgh, 412-624-0873, xlwu+@pitt.edu) create turbulence in a salt-based soap film, by sending electric and magnetic fields through it. (See movies at www.aip.org/png.) By adding a dash of lycopodium (mushroom spores) to the film, they could monitor the turbulence by tracking the lycopodium particles. They found that the energy leakage to the air molecules was at least equal to and in many cases greater than the energy dissipated to internal viscosity. These experiments will give researchers a better idea of the "energy budget" in 2-D turbulent flows. They also underline the fact that there are no ideal 2-D systems in our observable world--they often interact significantly with objects in the third spatial dimension. Such an interaction is essential for maintaining a steady state in the system. (Rivera and Wu, Physical Review Letters, 31 July 2000; movie at Phys. News Graphics.)
SINGLE-ELECTRON DETECTOR AND COUNTER. Two scientists at Cambridge (Haroon Ahmed, ha10@phy.cam.ac.uk, and N.J. Stone) have built a tiny multiple tunnel junction device (or MTJ, consisting of a series of barriers and islands which force electrons to reside on one side or another) which can move electrons one at a time. But unlike other single-electron transistors, the Cambridge circuit also uses a second MTJ to count the electronic comings and goings. Electron detection is achieved by registering the change in a detector circuit when an electron enters or leaves. Counting electrons, however, is subtler since each additional electron arrival must be registered and added to the previous output stored in the detector. Possible applications? Multi-level memory, logic functions, current standards, and high sensitivity detection. An important consideration for all micro-circuits is ease of fabrication; in this case no advance in lithography is required since the islands and barriers are formed naturally along the length of a silicon nanowire using presently available silicon technology. (Applied Physics Letters, 31 July; Select Articles.)
ROLLING DICE AND FLIPPING BURGERS. Undergraduate researchers are combining mathematical versions of these tasks to slash dramatically the computational times for difficult simulations of physical phenomena. At the upcoming summer meeting of the American Association of Physics Teachers in Ontario, Shafiq Rahman of Allegheny College in Pennsylvania (814-332-5331, srahman@alleg.edu) will discuss this computational approach. Simulating even small systems in physics can seem impossible. Suppose that researchers want to study the magnetic properties of a ten-by-ten grid of electrons. Assume further that each electron has two possible values of spin, a property that helps determine its magnetic properties. The ten-by-ten grid of those electrons has a whopping 2^100 possible combinations of spin states. It would take a computer far longer than the age of the universe to go through all of those possibilities. So researchers turn to the Monte Carlo simulation. Through the use of random number generators, Monte Carlo simulations determine a small but representative fraction of these possible configurations. This in turn enables researchers to calculate the average properties of the system. To get a good representation, researchers use the Metropolis algorithm, which flips the spin of some electrons, in the attempt to generate a wide variety of spin configurations. However, Metropolis fails in systems on the verge of a phase transition, such as water turning to ice or a metal losing its magnetism. Such systems have to go through many, many changes before arriving at another "independent" configuration. So researchers instead identify clusters or islands of identical spin states. In this approach, researchers flip whole islands rather than individual members to generate completely distinct states. This "cluster-flipping Monte Carlo" approach, says Rahman, can cut computational time by orders of magnitude, especially near phase transitions. He and some undergraduates at Allegheny are using this approach to study complex anti-ferromagnetic materials. (Meeting Talk AB4, July 31; more information at www.aapt.org)
|
