CHARLES: A C++ fixed-point library for Photonic-Aware Neural Networks.

In this paper we present CHARLES (C++ pHotonic Aware neuRaL nEtworkS), a C++ library aimed at providing a flexible tool to simulate the behavior of Photonic-Aware Neural Network (PANN). PANNs are neural network architectures aware of the constraints due to the underlying photonic hardware, mostly in terms of low equivalent precision of the computations. For this reason, CHARLES exploits fixed-point computations for inference, while it supports both floating-point and fixed-point numerical formats for training.

Energy Level Alignment at the Cobalt Phosphate/Electrolyte Interface: Intrinsic Stability vs Interfacial Chemical Reactions in 5 V Lithium Ion Batteries.

The intrinsic stability of the 5 V LiCoPO-LiCoPO thin-film (carbon-free) cathode material coated with MoO thin layer is studied using a comprehensive synchrotron electron spectroscopy in situ approach combined with first-principle calculations. The atomic-molecular level study demonstrates fully reversible electronic properties of the cathode after the first electrochemical cycle. The polyanionic oxide is not involved in chemical reactions with the fluoroethylene-containing liquid electrolyte even when charged to 5.

Ab Initio Electron-Phonon Interactions in Correlated Electron Systems.

Electron-phonon (e-ph) interactions are pervasive in condensed matter, governing phenomena such as transport, superconductivity, charge-density waves, polarons, and metal-insulator transitions. First-principles approaches enable accurate calculations of e-ph interactions in a wide range of solids. However, they remain an open challenge in correlated electron systems (CES), where density functional theory often fails to describe the ground state.

The Different Story of Bonds.

We revisit "classical" issues in multiply bonded systems between main groups elements, namely the structural distortions that may occur at the multiple bonds and that lead, e.g., to trans-bent and bond-length alternated structures.

Fast Approximations of Activation Functions in Deep Neural Networks when using Posit Arithmetic.

With increasing real-time constraints being put on the use of Deep Neural Networks (DNNs) by real-time scenarios, there is the need to review information representation. A very challenging path is to employ an encoding that allows a fast processing and hardware-friendly representation of information. Among the proposed alternatives to the IEEE 754 standard regarding floating point representation of real numbers, the recently introduced Posit format has been theoretically proven to be really promising in satisfying the mentioned requirements.