Search

We propose a new high-performance activation function, Moderate Adaptive Linear Units (MoLU), for the deep neural network. The MoLU is a simple, beautiful and powerful activation function that can be a good main activation function among hundreds of activation functions. Because the MoLU is made up of the elementary functions, not only it is a infinite diffeomorphism (i.e. smooth and infinitely differentiable over whole domains), but also it decreases training time., Comment: 4 pages, 5 figures

Neural Ordinary Differential Equations (NeuralODEs) present an attractive way to extract dynamical laws from time series data, as they bridge neural networks with the differential equation-based modeling paradigm of the physical sciences. However, these models often display long training times and suboptimal results, especially for longer duration data. While a common strategy in the literature imposes strong constraints to the NeuralODE architecture to inherently promote stable model dynamics, such methods are ill-suited for dynamics discovery as the unknown governing equation is not guaranteed to satisfy the assumed constraints. In this paper, we develop a new training method for NeuralODEs, based on synchronization and homotopy optimization, that does not require changes to the model architecture. We show that synchronizing the model dynamics and the training data tames the originally irregular loss landscape, which homotopy optimization can then leverage to enhance training. Through benchmark experiments, we demonstrate our method achieves competitive or better training loss while often requiring less than half the number of training epochs compared to other model-agnostic techniques. Furthermore, models trained with our method display better extrapolation capabilities, highlighting the effectiveness of our method., Comment: 10 pages, 5 figures, accepted at NeurIPS2023 (https://neurips.cc/virtual/2023/poster/70313)

Optical precision spectroscopy of isotope shifts can be used to test for new forces beyond the Standard Model, and to determine basic properties of atomic nuclei. We measure isotope shifts on the highly forbidden ${}^2S_{1/2} \rightarrow {}^2F_{7/2}$ octupole transition of trapped $^{168,170,172,174,176}$Yb ions. When combined with previous measurements in Yb$^+$ and very recent measurements in Yb, the data reveal a King plot nonlinearity of up to 240$\sigma$. The trends exhibited by experimental data are explained by nuclear density functional theory calculations with the Fayans functional. We also find, with 4.3$\sigma$ confidence, that there is a second distinct source of nonlinearity, and discuss its possible origin.

The characterization of drug-protein interactions is crucial in the high-throughput screening for drug discovery. The deep learning-based approaches have attracted attention because they can predict drug-protein interactions without trial-and-error by humans. However, because data labeling requires significant resources, the available protein data size is relatively small, which consequently decreases model performance. Here we propose two methods to construct a deep learning framework that exhibits superior performance with a small labeled dataset. At first, we use transfer learning in encoding protein sequences with a pretrained model, which trains general sequence representations in an unsupervised manner. Second, we use a Bayesian neural network to make a robust model by estimating the data uncertainty. As a result, our model performs better than the previous baselines for predicting drug-protein interactions. We also show that the quantified uncertainty from the Bayesian inference is related to the confidence and can be used for screening DPI data points.

Surface tension is a key parameter for understanding nucleation from the very initial stage of phase transformation. Although surface tension has been predicted to vary with the curvature of the liquid-vapor interface, particularly at the large curvature of, e.g., the subnanometric critical nucleus, experimental study still remains challenging due to inaccessibility to such a small cluster. Here, by directly measuring the critical size of a single capillary-condensed nanomeniscus using atomic force microscopy, we address the curvature dependence of surface tension of alcohols and observe the surface tension is doubled for ethanol and n-propanol with the radius-of-curvature of ~ -0.46 nm. We also find that the interface of larger negative (positive) curvature exhibits the larger (smaller) surface tension, which evidently governs nucleation at ~ 1 nm scale, indicating more facilitated nucleation than normally expected. Such well characterized curvature effects contribute to better understanding and accurate analysis of nucleation occurring in various fields including material science and atmospheric science., Comment: 5 pages, 4 figures and Supplementary Material(pdf)

Understanding phases of water molecules based on local structure is essential for understanding their anomalous properties. However, due to complicated structural motifs formed via hydrogen bonds, conventional order parameters represent the water molecules incompletely. In this paper, we develop a GCIceNet, which automatically generates machine-based order parameters for classifying the phases of the water molecules via supervised and unsupervised learning. Multiple graph convolutional layers in the GCIceNet can learn topological informations of the complex hydrogen bond networks. It shows a substantial improvement of accuracy for predicting the phase of water molecules in the bulk system and the ice/vapor interface system. A relative importance analysis shows that the GCIceNet can capture the structural features of the given system hidden in the input data. Augmented with the vast amount of data provided by molecular dynamics simulations, the GCIceNet is expected to serve as a powerful tool for the fields of glassy liquids and hydration layers around biomolecules.

Surface tension plays a ubiquitous role in phase transitions including condensation or evaporation of atmospheric liquid droplets. Especially, understanding of interfacial thermodynamics of the critical nucleus of 1 nm scale is important for molecular characterization of the activation energy barrier of nucleation. Here, we investigate surface tension of spherical nanodroplets with both molecular dynamics and density functional theory, and find that surface tension decreases appreciably below 1 nm radius, whose analytic expression is consistently derived from the classic Tolman's equation. In particular, the free energy analysis of nanodroplets shows that the change of surface tension originates dominantly from the configurational energy of interfacial molecules, which is evidenced by the increasingly disrupted hydrogen bond network as the droplet size decreases. Our result can be applied to the interface-related phenomena associated with molecular fluctuations such as biomolecule adsorption at sub-nm scale where the macroscopic thermodynamic quantities are ill-defined., Comment: 17 pages, 6 figures

We measure isotope shifts for five Yb$^+$ isotopes with zero nuclear spin on two narrow optical quadrupole transitions ${}^2S_{1/2} \rightarrow {}^2D_{3/2}$, ${}^2S_{1/2} \rightarrow {}^2D_{5/2}$ with an accuracy of $\sim 300$ Hz. The corresponding King plot shows a $3 \times 10^{-7}$ deviation from linearity at the 3 $\sigma$ uncertainty level. Such a nonlinearity can indicate physics beyond the Standard Model (SM) in the form of a new bosonic force carrier, or arise from higher-order nuclear effects within the SM. We identify the quadratic field shift as a possible contributor to the nonlinearity at the observed scale, and show how the nonlinearity pattern can be used in future, more accurate measurements to separate a new-boson signal from nuclear effects.

Solid-state transformation is often accompanied by mechanical expansion/compression, due to their volume change and structural evolution at interfaces at the atomic scale. However, these two types of dynamics are usually difficult to monitor in the same time. In this work, we use in-situ transmission electron microscopy to directly study the reduction transformation at the AgCl-Ag interface. Three stages of lattice fluctuations were identified and correlated to the structural evolution. During the steady state, a quasi-layered growth mode of Ag in both vertical and lateral directions were observed due to the confinement of AgCl lattices. The development of planar defects and depletion of AgCl are respectively associated with lattice compression and relaxation. Topography and structure of decomposing AgCl was further monitored by in-situ scanning transmission electron microscopy. Silver species are suggested to originate from both the surface and the interior of AgCl, and be transported to the interface. Such mass transport may have enabled the steady state and lattice compression in this volume-shrinking transformation., Comment: 21 pages, 5 figures; accepted for publication in J. Phys. Chem. Lett

The physics of critical phenomena in a many-body system far from thermal equilibrium is an interesting and important issue to be addressed both experimentally and theoretically. The trapped cold atoms have been actively used as a clean and versatile simulator for classical and quantum-mechanical systems, deepening understanding of the many-body physics behind. Here we review the nonlinear and collective dynamics in a periodically modulated magneto-optically trapped cold atoms. By temporally modulating the intensity of the trapping lasers with the controlled phases, one can realize two kinds of nonlinear oscillators, the parametrically driven oscillator and the resonantly driven Duffing oscillator, which exhibit the dynamical bistable states. Cold atoms behave not only as the single-particle nonlinear oscillators, but also as the coupled oscillators by the light-induced inter-atomic interaction, which leads to the phase transitions far from equilibrium in a way similar to the phase transition in equilibrium. The parametrically driven cold atoms show the ideal mean-field symmetry-breaking transition, and the symmetry is broken with respect to time translation by the modulation period. Such a phase transition results from the cooperation and competition between the inter-particle interaction and the fluctuations, which lead to the nonlinear switching of atoms between the vibrational states, and the experimentally measured critical characteristics prove it as the ideal mean-field transition class. On the other hand, the resonantly driven cold atoms that possess the coexisting periodic attractors exhibit the kinetic phase transition analogous to the discontinuous gas-liquid phase transition in equilibrium, and interestingly the global interaction between atoms causes the shift of the phase-transition boundary., Comment: 57 pages, 25 figures

Friction is the basic, ubiquitous mechanical interaction between two surfaces that results in resistance to motion and energy dissipation. In spite of its technological and economic significance, our ability to control friction remains modest, and our understanding of the microscopic processes incomplete. At the atomic scale, mismatch between the two contacting crystal lattices can lead to a reduction of stick-slip friction (structural lubricity), while thermally activated atomic motion can give rise to a complex velocity dependence, and nearly vanishing friction at sufficiently low velocities (thermal lubricity). Atomic force microscopy has provided a wealth of experimental results, but limitations in the dynamic range, time resolution, and control at the single-atom level have hampered a full quantitative description from first principles. Here, using an ion-crystal friction emulator with single-atom, single substrate-site spatial resolution and single-slip temporal resolution, we measure the friction force over nearly five orders of magnitude in velocity, and contiguously observe four distinct regimes, while controlling temperature and dissipation. We elucidate the interplay between thermal and structural lubricity in a system of two coupled atoms, and provide a simple explanation in terms of the Peierls-Nabarro potential. This extensive control at the atomic scale paves the way for fundamental studies of the interaction of many-atom surfaces, as for example in the Frenkel-Kontorova model, and possibly into the quantum regime., Comment: 3 pages, 4 figures

We show that an atomic system in a periodically modulated optical trap displays an ideal mean-field symmetry-breaking transition. The symmetry is broken with respect to time translation by the modulation period. The transition is due to the interplay of the long-range interatomic interaction and nonequilibrium fluctuations. The observed critical dynamics, including anomalous fluctuations in the symmetry broken phase, are fully described by the proposed microscopic theory.

We report on high-resolution photoluminescence (PL) spectroscopic and microscopic study of laterally coupled InAs/GaAs self-assembled quantum dots by using a low-temperature near-field scanning optical microscope. We have observed slightly split PL spectra, which are associated with the bonding (symmetric) and antibonding (antisymmetric) energy states between two coupled quantum dots, closely located each other as confirmed by spatial mapping of the PL intensity. The experimental results are in qualitative agreement with the simple theoretical calculations based on a two-dimensional potential model. This work may open the way to a simultaneous spectroscopy and microscopy study of laterally coupled quantum dots in a high-density quantum dot sample without any articulate sample fabrication., Comment: 10 pages, 4 figures

We have observed spontaneous symmetry breaking of atomic populations in the dynamic phase-space double-potential system, which is produced in the parametrically driven magneto-optical trap of atoms. We find that the system exhibits similar characteristics of the Ising-class phase transition and the critical value of the control parameter, which is the total atomic number, can be calculated. In particular, the collective effect of the laser shadow becomes dominant at large atomic number, which is responsible for the population asymmetry of the dynamic two-state system. This study may be useful for investigation of dynamic phase transition and temporal behaviour of critical phenomena., Comment: 4 pages, 4 figures

We report on high-resolution photoluminescence (PL) spectroscopy of spatial structure of InAs/AlGaAs quantum dots (QDs) by using a near-field scanning optical microscope (NSOM). The double-peaked distribution of PL spectra is clearly observed, which is associated with the bimodal size distribution of single QDs. In particular, the size difference of single QDs, represented by the doublet spectral distribution, can be directly observed by the NSOM images of PL., Comment: 3pages, 3figues

We have performed near-field spectroscopy and microscopy of the InAs/AlGaAs quantum-dot-based nanoclusters. It is observed that the photoluminescence spectra of spatially confined excitons in the nanoclusters is blue-shifted up to 20 meV as the power density is increased. In particular, the near-field photoluminescence images have shown that excitons became spatially confined gradually from lower energy state (1.4150 eV) to higher energy state (1.4392 eV) as the excitation power is increased, which is indicative of the band-filling effect of semiconductor nanostructures., Comment: 3pages, 5 figures

We have investigated noise-induced transition of atoms between double or triple phase-space attractors that are produced in the parametrically driven magneto-optical trap. The transition rates between two or three dynamic attractors, measured for various modulation frequencies and amplitudes, are in good agreement with theoretical calculations and Monte-Carlo simulations based on the Langevin equations. Our experiment may be useful to study nonlinear dynamic problems involving transitions between states far from equilibrium., Comment: 4 pages, 4 figures

The theory of Bose-Einstein condensation in a two-dimensional(2D) harmonic trap is developed from 2D Gross-Pitaevskii equation. The 2D interaction strength is obtained from a 2D collision theory. We show the realization of 2D condensation of trapped Bose atoms directly by obtaining the stable solutions for the condensate wave function from the 2D Gross-Pitaevskii equation. We calculate the ground-state energy of the 2D system, and also the wave function of the 2D vortex state. In particular, the 2D energy state becomes less stable than the 3D case with the number of trapped atoms. The results of the 2D Bose condensation are also compared with those of the well-known 3D case., Comment: 6 pages, 4 figures

We derive the Gross-Pitaevskii equation in two-dimension from the first principles of two-dimensional scattering theory. Numerical calculation of the condensate wave function shows that atoms in a two-dimensional harmonic trap can be condensed into its ground state. Moreover, the ground state energy and the wave function of two-dimensional vortex state are also obtained. Quantitative comparisons between 2D results and 3D ones are made in detail.