Keele University

About: Keele University is a education organization based out in Newcastle-under-Lyme, United Kingdom. It is known for research contribution in the topics: Population & Stars. The organization has 11318 authors who have published 26323 publications receiving 894671 citations. The organization is also known as: Keele University.

The viscosity of liquid iron at the physical conditions of the Earth's core

Abstract: It is thought that the Earth's outer core consists mainly of liquid iron and that the convection of this metallic liquid gives rise to the Earth's magnetic field. A full understanding of this convection is hampered, however, by uncertainty regarding the viscosity of theouter core. Viscosity estimates from various sources span no less than 12 orders of magnitude1,2, and it seems unlikely that thisuncertainty will be substantially reduced by experimental measurements in the near future. Here we present dynamical first-principles simulations of liquid iron which indicate that the viscosity of iron at core temperatures and pressures is at the low end of the range of previous estimates — roughly 10 times that of typical liquid metals at ambient pressure. This estimate supports the approximation commonly made in magnetohydrodynamic models that the outer core is an inviscid fluid3,4,5 undergoing small-scale circulation and turbulent convection6, rather than large-scale global circulation.

A Spitzer view of protoplanetary disks in the gamma Velorum cluster

Abstract: We present new Spitzer Space Telescope observations of stars in the young ~5 Myr gamma Velorum stellar cluster. Combining optical and 2MASS photometry, we have selected 579 stars as candidate members of the cluster. With the addition of the Spitzer mid-infrared data, we have identified 5 debris disks around A-type stars, and 5-6 debris disks around solar-type stars, indicating that the strong radiation field in the cluster does not completely suppress the production of planetesimals in the disks of cluster members. However, we find some evidence that the frequency of circumstellar primordial disks is lower, and the IR flux excesses are smaller than for disks around stellar populations with similar ages. This could be evidence for a relatively fast dissipation of circumstellar dust by the strong radiation field from the highest mass star(s) in the cluster. Another possibility is that gamma Velorum stellar cluster is slightly older than reported ages and the the low frequency of primordial disks reflects the fast disk dissipation observed at ~5 Myr.

THE r-PROCESS NUCLEOSYNTHESIS IN THE VARIOUS JET-LIKE EXPLOSIONS OF MAGNETOROTATIONAL CORE-COLLAPSE SUPERNOVAE

Abstract: The r-process nucleosynthesis in core-collapse supernovae (CC-SNe) is studied, with a focus on the explosion scenario induced by rotation and strong magnetic fields. Nucleosynthesis calculations are conducted based on magneto-hydrodynamical explosion models with a wide range of parameters for initial rotation and magnetic fields. The explosion models are classified in two different types: i.e., prompt-magnetic-jet and delayed-magnetic-jet, for which the magnetic fields of proto-neutron stars (PNSs) during collapse and the core-bounce are strong and comparatively moderate, respectively. Following the hydrodynamical trajectories of each explosion model, we confirmed that r-processes successfully occur in the prompt-magnetic-jets, which produce heavy nuclei including actinides. On the other hand, the r-process in the delayed-magnetic-jet is suppressed, which synthesizes only nuclei up to the second peak (A∼130). Thus, the r-process in the delayed-magnetic-jets could explain only "weak r-process" patterns observed in metal-poor stars rather than the "main r-process", represented by the solar abundances. Our results imply that core-collapse supernovae are possible astronomical sources of heavy r-process elements if their magnetic fields are strong enough, while weaker magnetic explosions may produce "weak r-process" patterns (A≲130). We show the potential importance and necessity of magneto-rotational supernovae for explaining the galactic chemical evolution, as well as abundances of r-process enhanced metal-poor stars. We also examine the effects of the remaining uncertainties in the nature of PNSs due to weak interactions that determine the final neutron-richness of ejecta. Additionally, we briefly discuss radioactive isotope yields in primary jets (e.g., 56Ni), with relation to several optical observation of SNe and relevant high-energy astronomical phenomena.

Surveying the Agents of Galaxy Evolution in the Tidally Stripped, Low Metallicity Small Magellanic Cloud (SAGE-SMC). I. Overview

Abstract: The Small Magellanic Cloud (SMC) provides a unique laboratory for the study of the lifecycle of dust given its low metallicity (~1/5 solar) and relative proximity (~60 kpc). This motivated the SAGE-SMC (Surveying the Agents of Galaxy Evolution in the Tidally Stripped, Low Metallicity Small Magellanic Cloud) Spitzer Legacy program with the specific goals of studying the amount and type of dust in the present interstellar medium, the sources of dust in the winds of evolved stars, and how much dust is consumed in star formation. This program mapped the full SMC (30 deg^2) including the body, wing, and tail in seven bands from 3.6 to 160 μm using IRAC and MIPS on the Spitzer Space Telescope. The data were reduced and mosaicked, and the point sources were measured using customized routines specific for large surveys. We have made the resulting mosaics and point-source catalogs available to the community. The infrared colors of the SMC are compared to those of other nearby galaxies and the 8 μm/24 μm ratio is somewhat lower than the average and the 70 μm/160 μm ratio is somewhat higher than the average. The global infrared spectral energy distribution (SED) shows that the SMC has approximately 1/3 the aromatic emission/polycyclic aromatic hydrocarbon abundance of most nearby galaxies. Infrared color-magnitude diagrams are given illustrating the distribution of different asymptotic giant branch stars and the locations of young stellar objects. Finally, the average SED of H II/star formation regions is compared to the equivalent Large Magellanic Cloud average H II/star formation region SED. These preliminary results will be expanded in detail in subsequent papers.

Federated learning with hierarchical clustering of local updates to improve training on non-IID data

Abstract: Federated learning (FL) is a well established method for performing machine learning tasks over massively distributed data. However in settings where data is distributed in a non-iid (not independent and identically distributed) fashion - as is typical in real world situations - the joint model produced by FL suffers in terms of test set accuracy and/or communication costs compared to training on iid data. We show that learning a single joint model is often not optimal in the presence of certain types of non-iid data. In this work we present a modification to FL by introducing a hierarchical clustering step (FL+HC) to separate clusters of clients by the similarity of their local updates to the global joint model. Once separated, the clusters are trained independently and in parallel on specialised models. We present a robust empirical analysis of the hyperparameters for FL+HC for several iid and non-iid settings. We show how FL+HC allows model training to converge in fewer communication rounds (significantly so under some non-iid settings) compared to FL without clustering. Additionally, FL+HC allows for a greater percentage of clients to reach a target accuracy compared to standard FL. Finally we make suggestions for good default hyperparameters to promote superior performing specialised models without modifying the the underlying federated learning communication protocol.