The WMAP determination of the baryon−to−photon ratio implies, through Big Bang nucleosynthesis, a cosmological Li abundance larger, by a factor of 2 to 3, than the Li abundance plateau observed in the oldest Pop II stars. It is however inescapable that there be a reduction by a factor of at least 1.6 to 2.0 of the surface Li abundance during the evolution of Pop II field stars with [Fe/H] ≤ −1.5. That the observed Li be lower than cosmologically produced Li is expected from stellar evolution models. Since at turnoff most of the Li abundance reduction is caused by gravitational settling, the presence of 6 Li in some turnoff stars is also understood. Given that the WMAP implications for Li cosmological abundance and the Li Spite plateau can be naturally explained by gravitational settling in the presence of weak turbulence, there appears little need for exotic physics as suggested by some authors. Instead, there is a need for a better understanding of turbulent transport in the radiative zones of stars. This requires simulations from first principles. Rather strict upper limits to turbulent transport are determined for the Sun and Pop II stars.

The measurement of the cosmic microwave background has strongly constrained the cosmological parameters of the Universe 1 . When the measured density of baryons (ordinary matter) is combined with standard Big Bang nucleosynthesis calculations 2,3 , the amounts of hydrogen, helium and lithium produced shortly after the Big Bang can be predicted with unprecedented precision 1,4 . The predicted primordial lithium abundance is a factor of two to three higher than the value measured in the atmospheres of old stars 5,6 . With estimated errors of 10 to 25 %, this cosmological lithium discrepancy seriously challenges our understanding of stellar physics, Big Bang nucleosynthesis or both. Certain modifications to nucleosynthesis have been proposed 7 , but found experimentally not to be viable 8 . Diffusion theory, however, predicts atmospheric abundances of stars to vary with time 9 , which offers a possible explanation of the discrepancy. Here we report spectroscopic observations of stars in the metal-poor globular cluster NGC 6397 that reveal trends of atmospheric abundance with evolutionary stage for various elements. These element-specific trends are reproduced by stellar-evolution models with diffusion and turbulent mixing 10 . We thus conclude that diffusion is predominantly responsible for the low apparent stellar lithium abundance in the atmospheres of old stars by transporting the lithium deep into the star. Diffusive processes altering the elemental composition in stars have been studied for decades 9,11 . Evidence for their importance comes from helioseismology 12 and the study of hot stars with peculiar abundance patterns 13 . Among solar-type stars, the effects of diffusion are expected to be more pronounced in old, very metal-poor stars. Given their greater age, diffusion has had more time to produce sizeable effects than in younger stars like the Sun. Detailed element-by-element predictions from models including effects of atomic diffusion and radiative accelerations became available a few years ago 14 , but these early models produced strong abundance trends that are incompatible with measurements of, in particular, the abundance of lithium common among stars of the Galactic halo over a wide range of metallicities (the so-called Spite plateau of lithium). However, the recent inclusion of turbulent mixing 10 brings model predictions into better agreement with observations. According to the predictions from such model calculations, stars leaving the main-sequence (turn-off stars) are expected to show the largest variations relative to the composition of the gas from which the stars originated. Giant stars, however, have deep surface convection zones which erase most effects of diffusion and restore the original composition. One notable exception is lithium which disintegrates in layers with T ≥ 2.1 million K. The destruction of lithium inside the star leads to a successive dilution of the surface lithium because the convective envelope expands when the star becomes a red giant. We performed spectroscopic ob...

International audienceLarge scale distributed systems such as Grids are difficult to study from theoretical models and simulators only. Most Grids deployed at large scale are production platforms that are inappropriate research tools because of their limited reconfiguration, control and monitoring capabilities. In this paper, we present Grid'5000, a 5000 CPU nation-wide infrastructure for research in Grid computing. Grid'5000 is designed to provide a scientific tool for computer scientists similar to the large-scale instruments used by physicists, astronomers, and biologists. We describe the motivations, design considerations, architecture, control, and monitoring infrastructure of this experimental platform. We present configuration examples and performance results for the reconfiguration subsystem