Two stellar components in the halo of the Milky Way.

The halo of the Milky Way provides unique elemental abundance and kinematic information on the first objects to form in the Universe, and this information can be used to tightly constrain models of galaxy formation and evolution. Although the halo was once considered a single component, evidence for its dichotomy has slowly emerged in recent years from inspection of small samples of halo objects. Here we show that the halo is indeed clearly divisible into two broadly overlapping structural components--an inner and an outer halo--that exhibit different spatial density profiles, stellar orbits and stellar metallicities (abundances of elements heavier than helium).

Discovery of young, isolated planetary mass objects in the final sigma Orionis star cluster.

We present the discovery by optical and near-infrared imaging of an extremely red, low-luminosity population of isolated objects in the young, nearby stellar cluster around the multiple, massive star final sigma Orionis. The proximity (352 parsecs), youth (1 million to 5 million years), and low internal extinction make this cluster an ideal site to explore the substellar domain from the hydrogen mass limit down to a few Jupiter masses. Optical and near-infrared low-resolution spectroscopy of three of these objects confirms the very cool spectral energy distribution (atmospheric effective temperatures of 1700 to 2200 kelvin) expected for cluster members with masses in the range 5 to 15 times that of Jupiter.

A recurrent neural network for modelling dynamical systems.

We introduce a recurrent network architecture for modelling a general class of dynamical systems. The network is intended for modelling real-world processes in which empirical measurements of the external and state variables are obtained at discrete time points. The model can learn from multiple temporal patterns, which may evolve on different timescales and be sampled at non-uniform time intervals.