Scalable Inference for Sparse Deep Neural Networks using Kokkos Kernels.
Abstract not provided.
Engineering topics
Publications and source records attributed to Ellis, John.
Abstract not provided.
Abstract not provided.
Abstract not provided.
We analyze nucleon decay modes in a no-scale supersymmetric flipped SU(5) GUT model, and contrast them with the predictions for proton decays via dimension-6 operators in a standard unflipped supersymmetric SU(5) GUT model. We find that these GUT models make very different predictions for the ratios Γ(p → π 0 µ + )/Γ(p → π 0 e + ), Γ(p → π + ν - )/Γ(p → π 0 e + ), Γ(p → K 0 e + )/Γ(p → π 0 e + ) and Γ(p → K 0 µ + )/Γ(p → π 0 µ + ), and that predictions for the ratios Γ(p → π 0 µ + )/Γ(p → π 0 e + ) and Γ(p → π + ν - )/Γ(p → π 0 e + ) also differ in variants of the flipped SU(5) model with normal- or inverse-ordered light neutrino masses. Upcoming large neutrino experiments may have interesting opportunities to explore both GUT and flavour physics in proton and neutron decays.
Encouraged by the advent of a new generation of underground detectors – JUNO, DUNE and Hyper-Kamiokande – that are projected to improve significantly on the present sensitivities to various baryon decay modes, we revisit baryon decay in the minimal supersymmetric SU(5) GUT. We discuss the phenomenological uncertainties associated with hadronic matrix elements and the value of the strong coupling αs – which are the most important – the weak mixing angle θ W , quark masses including one-loop renormalization effects, quark mixing and novel GUT phases that are not visible in electroweak interaction processes. We apply our analysis to a variety of CMSSM, super- and sub-GUT scenarios in which soft supersymmetry-breaking parameters are assumed to be universal at, above and below the GUT scale, respectively. In many cases, we find that the next generation of underground detectors should be able to probe models with sparticle masses that are O(10) TeV, beyond the reach of the LHC.
We propose in this White Paper a concept for a space experiment using cold atoms to search for ultra-light dark matter, and to detect gravitational waves in the frequency range between the most sensitive ranges of LISA and the terrestrial LIGO/Virgo/KAGRA/INDIGO experiments. This interdisciplinary experiment, called Atomic Experiment for Dark Matter and Gravity Exploration (AEDGE), will also complement other planned searches for dark matter, and exploit synergies with other gravitational wave detectors. We give examples of the extended range of sensitivity to ultra-light dark matter offered by AEDGE, and how its gravitational-wave measurements could explore the assembly of super-massive black holes, first-order phase transitions in the early universe and cosmic strings. AEDGE will be based upon technologies now being developed for terrestrial experiments using cold atoms, and will benefit from the space experience obtained with, e.g., LISA and cold atom experiments in microgravity. This paper is based on a submission (v1) in response to the Call for White Papers for the Voyage 2050 long-term plan in the ESA Science Programme. ESA limited the number of White Paper authors to 30. However, in this version (v2) we have welcomed as supporting authors participants in the Workshop on Atomic Experiments for Dark Matter and Gravity Exploration held at CERN: ({\tt https://indico.cern.ch/event/830432/}), as well as other interested scientists, and have incorporated additional material.
miniGAN is a python-based machine learning proxy application for generative adversarial networks, developed through the Exascale Computing Project's (ECP) ExaLearn project. It will be included in the main ECP proxy application and the machine learning proxy application suite. It is a proxy for ECP cosmological(CosmoFlow, ExaGAN) and wind energy(ExaWind) applications. miniGAN will be distributed to ECP hardware vendors as part of hardware codesign. miniGAN uses the Numpy/PyTorch/TensorFlow/Keras/Horovod frameworks and libraries. It also relies on the Kokkos and Kokkos-Kernels packages developed here at Sandia Labs. SAND2020-2038 M Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy's National Nuclear Security Administration under contract DE-NA0003525.
We develop a string-inspired model for particle cosmology, based on a flipped SU(5)×U(1) gauge group formulated in a no-scale supergravity framework. The model realizes Starobinsky-like inflation, which we assume to be followed by strong reheating, with the GUT symmetry being broken subsequently by a light ‘flaton’ field whose decay generates a second stage of reheating. We discuss the production of gravitinos and the non-thermal contribution made by their decays to the density of cold dark matter, which is assumed to be provided by the lightest neutralino. We also discuss the masses of light and heavy neutrinos and leptogenesis. As discussed previously, a key rôle is played by a superpotential coupling between the inflaton, matter and GUT Higgs fields, called λ 6 . We scan over possible values of λ 6 , exploring the correlations between the possible values of observables. We emphasize that the release of entropy during the GUT transition allows large regions of supersymmetry-breaking parameter space that would otherwise lead to severe overdensity of dark matter. Furthermore, we find that the Big Bang nucleosynthesis lower limit on the reheating temperature of ~ 1 MeV restricts the supersymmetry-breaking scale to a range O(10) TeV that is consistent with the absence of supersymmetric particles at the LHC.
We show the uniqueness of superpotentials leading to Minkowski vacua of single-field no-scale supergravity models, and the construction of dS/AdS solutions using pairs of these single-field Minkowski superpotentials. We then extend the construction to two- and multifield no-scale supergravity models, providing also a geometrical interpretation. We also consider scenarios with additional twisted or untwisted moduli fields, and discuss how inflationary models can be constructed in this framework.
Not Available
We have presented previously a general treatment of Starobinsky-like inflation in no-scale supergravity where the tensor-to-scalar ratio r = 3(1–n s ) 2 , and n s is the tilt of the scalar perturbations. In particular, we have shown how this scenario can be unified with modulus fixing, supersymmetry breaking and a small cosmological constant. In this study we extend these constructions to inflationary models based on generalized no-scale structures. In particular, we consider alternative values of the curvature parameter, α < 1, as may occur if not all the complex Kähler moduli contribute to driving inflation, as well as α > 1, as may occur if complex structure moduli also contribute to driving inflation. In all cases, we combine these α-Starobinsky inflation models with supersymmetry breaking and a present-day cosmological constant, allowing for additional contributions to the vacuum energy from stages of gauge symmetry breaking.
Not Available
No-scale supergravity and the flipped SU(5)×U(1) gauge group provide an ambitious prototype string-inspired scenario for physics below the string scale, which can accommodate the Starobinsky-like inflation favoured by observation when the inflaton is associated with one of the singlet fields associated with neutrino mass generation. During inflation, the vacuum remains in the unbroken GUT phase, and GUT symmetry breaking occurs later when a field with a flat direction (the flaton) acquires a vacuum expectation value. Inflaton decay and the reheating process depend crucially on GUT symmetry breaking, as decay channels open and close, depending on the value of the flaton vacuum expectation value. Here, we consider the simultaneous cosmological evolution of both the inflaton and flaton fields after inflation. We distinguish weak, moderate and strong reheating regimes, and calculate in each case the entropy produced as all fields settle to their global minima. These three reheating scenarios differ in the value of a Yukawa coupling that introduces mass mixing between the singlets and the 10s of SU(5). Here, the dynamics of the GUT transition has an important impact on the production of gravitinos, and we also discuss the pattern of neutrino masses we expect in each of the three cases. Finally, we use recent CMB limits on neutrino masses to constrain the reheating models, finding that neutrino masses and the cosmological baryon asymmetry can both be explained if the reheating is strong.
Measurements of the cosmic microwave background (CMB) favour models of inflation with a small tensor-to-scalar ratio r, as predicted by the Starobinsky R + R2 model. It has been shown previously that various models based on no-scale supergravity with different forms of superpotential make predictions similar to those of the Starobinsky model. In this paper we present a unified and general treatment of Starobinsky avatars of no-scale supergravity, using the underlying non-compact SU(2,1)/SU(2)×U(1) symmetry to demonstrate equivalences between different models, exhibiting 6 specific equivalence classes.