The significant neutrino flux at high rapidity at the LHC motivates dedicated forward detectors to study the properties of neutrinos at TeV energies. We investigate magnetic dipole interactions between the active neutrinos and new sterile states at e
mulsion and liquid argon experiments that could be located in a future Forward Physics Facility (FPF) downstream of the ATLAS interaction point. The up-scattering of neutrinos off electrons produces an electron recoil signature that can probe new regions of parameter space at the High Luminosity LHC (HL-LHC), particularly for liquid argon detectors due to low momentum thresholds. We also consider the decay of the sterile neutrino through the dipole operator, which leads to a photon that could be displaced from the production vertex. FPF detectors can test sterile neutrino states as heavy as 1 GeV produced through the dipole portal, highlighting the use of high energy LHC neutrinos as probes of new physics.
Supernova remnants are known to accelerate cosmic rays (CRs) on account of their non-thermal emission of radio waves, X-rays, and gamma rays. However, the ability to accelerate CRs up to PeV-energies has yet to be demonstrated. The presence of cut-of
fs in the gamma-ray spectra of several young SNRs led to the idea that PeV energies might only be achieved during the very initial stages of a remnants evolution. We use the time-dependent acceleration code RATPaC to study the acceleration of cosmic rays in supernovae expanding into dense environments around massive stars, where the plentiful target material might offer a path to the detection of gamma-rays by current and future experiments. We performed spherically symmetric 1-D simulations in which we simultaneously solve the transport equations for cosmic rays, magnetic turbulence, and the hydrodynamical flow of the thermal plasma in the test-particle limit. We investigated typical parameters of the circumstellar medium (CSM) in the freely expanding winds around red supergiant (RSG) and luminous blue variable (LBV) stars. The maximum achievable energy might be limited to sub-PeV energies despite strong magnetic fields close to the progenitor star that enhance turbulence-damping by cascading: we find a maximum CR energy of 100-200 TeV, reached within one month after explosion. The peak luminosity for a LBV progenitor is 1e43 erg/s (1e42 erg/s) at GeV (TeV) energies and, for a RSG progenitor, 1e41 erg/s (1e40 erg/s). All calculated SNe reach their peak gamma-ray luminosity after <~1 month and then fade at a rate ~1/t as long as the SN shock remains in the freely expanding wind of the progenitor. Potentially detectable gamma-ray signals can be expected in the Fermi-LAT waveband weeks to months after an explosion into a freely expanding wind.
We introduce Amortized Neural Networks (AmNets), a compute cost- and latency-aware network architecture particularly well-suited for sequence modeling tasks. We apply AmNets to the Recurrent Neural Network Transducer (RNN-T) to reduce compute cost an
d latency for an automatic speech recognition (ASR) task. The AmNets RNN-T architecture enables the network to dynamically switch between encoder branches on a frame-by-frame basis. Branches are constructed with variable levels of compute cost and model capacity. Here, we achieve variable compute for two well-known candidate techniques: one using sparse pruning and the other using matrix factorization. Frame-by-frame switching is determined by an arbitrator network that requires negligible compute overhead. We present results using both architectures on LibriSpeech data and show that our proposed architecture can reduce inference cost by up to 45% and latency to nearly real-time without incurring a loss in accuracy.
We study the non-equilibrium thermodynamics of a heat engine operating between two finite-sized reservoirs with well-defined temperatures. Within the linear response regime, it is discovered that there exists a power-efficiency trade-off depending on
the ratio of heat capacities ($gamma$) of the reservoirs for the engine; the uniform temperature of the two reservoirs at final time $tau$ is bounded from below by the entropy production $sigma_{mathrm{min}}propto1/tau$. We further obtain a universal efficiency at maximum power of the engine for arbitrary $gamma$. Our findings can be used to develop an optimization scenario for thermodynamic cycles with finite-sized reservoirs in practice.
The LHC may produce light, weakly-interacting particles that decay to dark matter, creating an intense and highly collimated beam of dark matter particles in the far-forward direction. We investigate the prospects for detecting this dark matter in tw
o far-forward detectors proposed for a future Forward Physics Facility: FASER$ u$2, a 10-tonne emulsion detector, and FLArE, a 10- to 100-tonne LArTPC. We focus here on nuclear scattering, including elastic scattering, resonant pion production, and deep inelastic scattering, and devise cuts that efficiently remove the neutrino-induced background. In the invisibly-decaying dark photon scenario, DM-nuclear scattering probes new parameter space for dark matter masses 5 MeV $lesssim m_{chi} lesssim$ 500 MeV. When combined with the DM-electron scattering studied previously, FASER$ u$2 and FLArE will be able to discover dark matter in a large swath of the cosmologically-favored parameter space with MeV $lesssim m_{chi} lesssim $ GeV.
Emittance exchange beamlines employ transverse masks to create drive and witness beams of variable longitudinal profile and bunch spacing. Recently, this approach has been used to create advanced driver profiles and demonstrate record-breaking plasma
wakefield transformer ratios [Roussel, R., et al., Phys. Rev. Lett. 124, 044802 (2020)], a crucial advancement for efficient witness acceleration. However, since the transverse masks are individually laser cut and installed into the UHV beamline, refinement of the beam profiles is not possible without replacing masks. Instead, this work proposes the use of a UHV compatible multileaf collimator as a beam mask. Such a device permits real-time adjustment of the electron distribution, permitting greater refinement in a manner highly synergistic with machine learning. Beam dynamics simulations have shown that a practically realizable multileaf collimator can offer resolution that is functionally equivalent to that offered by laser cut masks.
The Deep Extragalactic VIsible Legacy Survey (DEVILS) is an ongoing high-completeness, deep spectroscopic survey of $sim$60,000 galaxies to Y$<$21.2 mag, over $sim$6 deg2 in three well-studied deep extragalactic fields: D10 (COSMOS), D02 (XMM-LSS) an
d D03 (ECDFS). Numerous DEVILS projects all require consistent, uniformly-derived and state-of-the-art photometric data with which to measure galaxy properties. Existing photometric catalogues in these regions either use varied photometric measurement techniques for different facilities/wavelengths leading to inconsistencies, older imaging data and/or rely on source detection and photometry techniques with known problems. Here we use the ProFound image analysis package and state-of-the-art imaging datasets (including Subaru-HSC, VST-VOICE, VISTA-VIDEO and UltraVISTA-DR4) to derive matched-source photometry in 22 bands from the FUV to 500{mu}m. This photometry is found to be consistent, or better, in colour-analysis to previous approaches using fixed-size apertures (which are specifically tuned to derive colours), but produces superior total source photometry, essential for the derivation of stellar masses, star-formation rates, star-formation histories, etc. Our photometric catalogue is described in detail and, after internal DEVILS team projects, will be publicly released for use by the broader scientific community.
Theoretical calculations of core electron binding energies are important for aiding the interpretation of experimental core level photoelectron spectra. In previous work, the $Delta$-Self-Consistent-Field ($Delta$-SCF) method based on density functio
nal theory has been shown to yield highly accurate 1s and 2p binding energies in free molecules. However, most experimental work is concerned with solids, not gases. In this study, we demonstrate the application of the all-electron $Delta$-SCF method to periodic systems. A consideration of the experimentally accessible points of reference leads to the definition of a core electron binding energy in a solid as the difference between the total energies of two $N-1$ electron systems: one with an explicit, localized core hole, and one with an electron removed from the highest occupied state. The calculation of each of these quantities is addressed. In addition, the analogy between a localized core hole and a charged defect in a solid is highlighted, and the extrapolation of calculated core electron binding energies to the infinite supercell limit is discussed. It is found that the extrapolated values of the core electron binding energies from periodic $Delta$-SCF calculations agree well with experimental results for both metallic and insulating systems, with a mean absolute error of 0.24 eV for the 15 core levels considered in this study.
The present work concerns the transferability of coarse-grained (CG) modeling in reproducing the dynamic properties of the reference atomistic systems across a range of parameters. In particular, we focus on implicit-solvent CG modeling of polymer so
lutions. The CG model is based on the generalized Langevin equation, where the memory kernel plays the critical role in determining the dynamics in all time scales. Thus, we propose methods for transfer learning of memory kernels. The key ingredient of our methods is Gaussian process regression. By integration with the model order reduction via proper orthogonal decomposition and the active learning technique, the transfer learning can be practically efficient and requires minimum training data. Through two example polymer solution systems, we demonstrate the accuracy and efficiency of the proposed transfer learning methods in the construction of transferable memory kernels. The transferability allows for out-of-sample predictions, even in the extrapolated domain of parameters. Built on the transferable memory kernels, the CG models can reproduce the dynamic properties of polymers in all time scales at different thermodynamic conditions (such as temperature and solvent viscosity) and for different systems with varying concentrations and lengths of polymers.
Modeling a high-dimensional Hamiltonian system in reduced dimensions with respect to coarse-grained (CG) variables can greatly reduce computational cost and enable efficient bottom-up prediction of main features of the system for many applications. H
owever, it usually experiences significantly altered dynamics due to loss of degrees of freedom upon coarse-graining. To establish CG models that can faithfully preserve dynamics, previous efforts mainly focused on equilibrium systems. In contrast, various soft matter systems are known out of equilibrium. Therefore, the present work concerns non-equilibrium systems and enables accurate and efficient CG modeling that preserves non-equilibrium dynamics and is generally applicable to any non-equilibrium process and any observable of interest. To this end, the dynamic equation of a CG variable is built in the form of the non-stationary generalized Langevin equation (nsGLE) to account for the dependence of non-equilibrium processes on the initial conditions, where the two-time memory kernel is determined from the data of the two-time auto-correlation function of the non-equilibrium trajectory-averaged observable of interest. By embedding the non-stationary non-Markovian process in an extended stochastic framework, an explicit form of the non-stationary random noise in the nsGLE is introduced, and the cost is significantly reduced for solving the nsGLE to predict the non-equilibrium dynamics of the CG variable. To prove and exploit the equivalence of the nsGLE and extended dynamics, the memory kernel is parameterized in a two-time exponential expansion. A data-driven hybrid optimization process is proposed for the parameterization, a non-convex and high-dimensional optimization problem.
