arXiv subjects

Junjie Yang

Publications and source records attributed to Junjie Yang.

At least 109 records · Page 6Linked to original sources

arXiv2020

Kinetically-stabilized Ferroelectricity in Bulk Singlecrystalline HfO2:Y without Wake-up Effects

HfO2, a simple binary oxide, holds ultra-scalable ferroelectricity integrable into silicon technology. Polar orthorhombic (Pbc21) form in ultra-thin-films ascribes as the plausible root-cause of the astonishing ferroelectricity, which has thought not attainable in bulk crystals. Though, perplexities remain primarily due to the polymorphic nature and the characterization challenges at small-length scales. Herein, utilizing a state-of-the-art Laser-Diode-heated Floating Zone technique, we report ferroelectricity in bulk single-crystalline HfO2:Y as well as the presence of anti-polar Pbca phase at different Y concentrations. Neutron diffraction and atomic imaging demonstrate (anti-)polar crystallographic signatures and abundant 90o/180o ferroelectric domains in addition to the switchable polarization with little wake-up effects. Density-functional theory calculations suggest that the Yttrium doping and rapid cooling are the key factors for the desired phase. Our observations provide new insights into the polymorphic nature and phase controlling of HfO2, remove the upper size limit for ferroelectricity, and also pave a new road toward the next-generation ferroelectric devices.

cond-mat.mtrl-sci↗

arXiv2020

Neural Network Training Techniques Regularize Optimization Trajectory: An Empirical Study

Modern deep neural network (DNN) trainings utilize various training techniques, e.g., nonlinear activation functions, batch normalization, skip-connections, etc. Despite their effectiveness, it is still mysterious how they help accelerate DNN trainings in practice. In this paper, we provide an empirical study of the regularization effect of these training techniques on DNN optimization. Specifically, we find that the optimization trajectories of successful DNN trainings consistently obey a certain regularity principle that regularizes the model update direction to be aligned with the trajectory direction. Theoretically, we show that such a regularity principle leads to a convergence guarantee in nonconvex optimization and the convergence rate depends on a regularization parameter. Empirically, we find that DNN trainings that apply the training techniques achieve a fast convergence and obey the regularity principle with a large regularization parameter, implying that the model updates are well aligned with the trajectory. On the other hand, DNN trainings without the training techniques have slow convergence and obey the regularity principle with a small regularization parameter, implying that the model updates are not well aligned with the trajectory. Therefore, different training techniques regularize the model update direction via the regularity principle to facilitate the convergence.

cs.LG↗

arXiv2020

Second-order nonlinear optical and linear UV-VIS absorption properties of type-II multiferroic candidates RbFe(AO4)2 (A = Mo, Se, S)

Motivated by the search for type-II multiferroics, we present a comprehensive optical study of a complex oxide family of type-II multiferroic candidates: RbFe(MoO4)2, RbFe(SeO4)2, and RbFe(SO4)2. We employ rotational-anisotropy second harmonic generation spectroscopy (RA SHG), a technique sensitive to point symmetries, to address discrepancies in literature-assigned point/space groups and to identify the correct crystal structures. At room temperature we find that our RA SHG patterns rotate away from the crystal axes in RbFe(AO4)2 (A = Se, S), which identifies the lack of mirror symmetry and in-plane two-fold rotational symmetry. Also, the SHG efficiency of RbFe(SeO4)2 is two orders of magnitude stronger than RbFe(AO4)2 (A = Mo, S), which suggests broken inversion symmetry. Additionally, we present temperature-dependent linear optical characterizations near the band edge of this family of materials using ultraviolet-visible (UV-VIS) absorption spectroscopy. Included is experimental evidence of the band gap energy and band gap transition type for this family. Previously unreported sub-band gap absorption is also presented, which reveals prominent optical transitions, some with an unusual central energy temperature dependence. Furthermore, we find that by substituting the A-site in RbFe(AO4)2 (A = Mo, Se, S), the aforementioned transitions are spectrally tunable. Finally, we discuss the potential origin and impact of these tunable transitions.

cond-mat.mtrl-sci↗

arXiv2020

Theoretical Convergence of Multi-Step Model-Agnostic Meta-Learning

As a popular meta-learning approach, the model-agnostic meta-learning (MAML) algorithm has been widely used due to its simplicity and effectiveness. However, the convergence of the general multi-step MAML still remains unexplored. In this paper, we develop a new theoretical framework to provide such convergence guarantee for two types of objective functions that are of interest in practice: (a) resampling case (e.g., reinforcement learning), where loss functions take the form in expectation and new data are sampled as the algorithm runs; and (b) finite-sum case (e.g., supervised learning), where loss functions take the finite-sum form with given samples. For both cases, we characterize the convergence rate and the computational complexity to attain an $ε$-accurate solution for multi-step MAML in the general nonconvex setting. In particular, our results suggest that an inner-stage stepsize needs to be chosen inversely proportional to the number $N$ of inner-stage steps in order for $N$-step MAML to have guaranteed convergence. From the technical perspective, we develop novel techniques to deal with the nested structure of the meta gradient for multi-step MAML, which can be of independent interest.

cs.LG↗

arXiv2020

Evolution of the structural transition in Mo$_{1-x}$W$_{x}$Te$_{2}$

The composition dependence of the structural transition between the monoclinic 1T$^{\prime}$ and orthorhombic T$_{d}$ phases in the Mo$_{1-x}$W$_{x}$Te$_{2}$ Weyl semimetal was investigated by elastic neutron scattering on single crystals up to $x \approx 0.54$. First observed in MoTe$_{2}$, the transition from T$_{d}$ to 1T$^{\prime}$ is accompanied by an intermediate pseudo-orthorhombic phase, T$_{d}^{*}$. Upon doping with W, the T$_{d}^{*}$ phase vanishes by $x \approx 0.34$. Above this concentration, a phase coexistence behavior with both T$_{d}$ and 1T$^{\prime}$ is observed instead. The interlayer in-plane positioning parameter $δ$, which relates to the 1T$^{\prime}$ $β$ angle, decreases with temperature as well as with W substitution, likely due to strong anharmonicity in the interlayer interactions. The temperature width of the phase coexistence remains almost constant up to $x \approx 0.54$, in contrast to the broadening reported under pressure.

cond-mat.mtrl-sci↗

arXiv2020

The Dirac oscillator: an alternative basis for nuclear structure calculations

Background: The isotropic harmonic oscillator supplemented by a strong spin-orbit interaction has been the cornerstone of nuclear structure since its inception more than seven decades ago. In this paper we introduce---or rather re-introduced---the "Dirac Oscillator", a fully relativistic basis that has all the desired attributes of the ordinary harmonic oscillator while naturally incorporating a strong spin-orbit coupling. Purpose: To assess---to our knowledge for the first time---the power and flexibility of the Dirac Oscillator basis in the solution of nuclear structure problems within the framework of covariant density functional theory. Methods: Self-consistent calculations of binding energies and ground-state densities for a selected set of doubly-magic magic are performed using the Dirac-oscillator basis and are then compared against results obtained with the often-used Runge-Kutta method. Results: Results obtained using the Dirac oscillator basis reproduced with high accuracy those derived using the Runge-Kutta method and suggest a clear path for a generalization to systems with axial symmetry. Conclusions: Although the three-dimensional harmonic oscillator with spin-orbit corrections has been the staple of the nuclear shell model since the beginning, the Dirac oscillator is practically unknown among the nuclear physics community. In this paper we illustrate the power and flexibility of the Dirac oscillator and suggest extensions to the study of systems without spherical symmetry as required in constrained calculations of nuclear excitations.

nucl-th↗

arXiv2020

Deepening Hidden Representations from Pre-trained Language Models

Transformer-based pre-trained language models have proven to be effective for learning contextualized language representation. However, current approaches only take advantage of the output of the encoder's final layer when fine-tuning the downstream tasks. We argue that only taking single layer's output restricts the power of pre-trained representation. Thus we deepen the representation learned by the model by fusing the hidden representation in terms of an explicit HIdden Representation Extractor (HIRE), which automatically absorbs the complementary representation with respect to the output from the final layer. Utilizing RoBERTa as the backbone encoder, our proposed improvement over the pre-trained models is shown effective on multiple natural language understanding tasks and help our model rival with the state-of-the-art models on the GLUE benchmark.

cs.CL↗

arXiv2019

Covariant Density Functional Theory in Nuclear Physics and Astrophysics

How does subatomic matter organize itself? Neutron stars are cosmic laboratories uniquely poised to answer this fundamental question that lies at the heart of nuclear science. Newly commissioned rare isotope facilities, telescopes operating across the entire electromagnetic spectrum, and ever more sensitive gravitational wave detectors will probe the properties of neutron-rich matter with unprecedented precision over an enormous range of densities. Yet, a coordinated effort between observation, experiment, and theoretical research is of paramount importance for realizing the full potential of these investments. Theoretical nuclear physics provides valuable insights into the properties of neutron-rich matter in regimes that are not presently accessible to experiment or observation. In particular, nuclear density functional theory is likely the only tractable framework that can bridge the entire nuclear landscape by connecting finite nuclei to neutron stars. This compelling connection is the main scope of the present review.

nucl-th↗

arXiv2019

Electroweak probes of ground state densities

Elastic electron scattering has been used to paint the most accurate picture of the proton distribution in atomic nuclei. Spurred by new experimental developments, it is now possible to gain valuable insights into the neutron distribution using exclusively electroweak probes. Our goal is to assess the information content and complementarity of the following three electroweak experiments in constraining the neutron distribution of atomic nuclei: (a) parity violating elastic electron scattering, (b) coherent elastic neutrino nucleus scattering, and (c) elastic electron scattering of unstable nuclei. Relativistic mean-field models informed by the properties of finite nuclei and neutron stars are used to compute ground state densities and form factors of a variety of nuclei. All the models follow the same fitting protocol, except for the assumed and presently unknown value of the neutron skin thickness of 208Pb. This enables one to tune the density dependence of the symmetry energy without compromising the success in reproducing well known physical observables. We found that the ongoing PREX-II and upcoming CREX campaigns at Jefferson Lab will play a vital role in constraining the weak form factor of xenon and argon, liquid noble gases that are used for the detection of both neutrinos and dark matter particles. We concluded that remarkable new advances in experimental physics have opened a new window into ground state densities of atomic nuclei using solely electroweak probes. The diversity and versatility of these experiments reveal powerful correlations that impose important nuclear structure constraints. In turn, these constraints provide quantitative theoretical uncertainties that are instrumental in searches for new physics and insights into the behavior of dense matter.

nucl-th↗

arXiv2019

Emergence of topologically protected states in MoTe$_{2}$ Weyl semimetal with layer stacking order

Electronic tunability in crystals with weakly-bound layers can be achieved through layer stacking order. One such example is MoTe$_2$, where the low-temperature orthorhombic T$_d$ phase is topological and host to Weyl quasiparticles. The transition mechanism to the non-trivial topology is elucidated by single crystal neutron diffraction. Upon cooling from the monoclinic 1T$\prime$ to the T$_d$ phase, diffuse scattering accompanies the transition, arising from random, in-plane layer displacements, and dissipates upon entering the T$_d$ phase. Diffuse scattering is observed only in the H0L plane due to irreversible layer shifts along the c-axis that break the centrosymmetry of the monoclinic lattice.

cond-mat.mtrl-sci↗

arXiv2019

SGD Converges to Global Minimum in Deep Learning via Star-convex Path

Stochastic gradient descent (SGD) has been found to be surprisingly effective in training a variety of deep neural networks. However, there is still a lack of understanding on how and why SGD can train these complex networks towards a global minimum. In this study, we establish the convergence of SGD to a global minimum for nonconvex optimization problems that are commonly encountered in neural network training. Our argument exploits the following two important properties: 1) the training loss can achieve zero value (approximately), which has been widely observed in deep learning; 2) SGD follows a star-convex path, which is verified by various experiments in this paper. In such a context, our analysis shows that SGD, although has long been considered as a randomized algorithm, converges in an intrinsically deterministic manner to a global minimum.

cs.LG↗

arXiv2017

Elastic and electronic tuning of magnetoresistance in MoTe$_2$

Quasi-two dimensional transition metal dichalcogenides (TMD) exhibit dramatic properties that may transform electronic and photonic devices. We report on how the anomalously large magnetoresistance (MR) observed under high magnetic field in MoTe$_2$, a type II Weyl semimetal, can be reversibly controlled under tensile strain. The MR is enhanced by as much as ~ 30 % at low temperatures and high magnetic fields, when uniaxial strain is applied along the $a$-crystallographic direction and reduced by about the same amount when strain is applied along the $b$-direction. We show that the large in-plane electric anisotropy is coupled with the structural transition from the 1T' monoclinic to the Td orthorhombic Weyl phase. A shift of the Td - 1T' phase boundary is achieved by minimal tensile strain. The sensitivity of the MR to tensile strain suggests the possibility of a nontrivial spin-orbital texture of the electron and hole pockets in the vicinity of Weyl points. Our ab initio calculations indeed show a significant orbital mixing on the Fermi surface, which is modified by the tensile strains.

cond-mat.mtrl-sci↗

arXiv2017

Difference in proton radii of mirror nuclei as a possible surrogate for the neutron skin

It has been recently suggested that differences in the charge radii of mirror nuclei are proportional to the neutron-skin thickness of neutron-rich nuclei and to the slope of the symmetry energy $L$ [B.A. Brown Phys. Rev. Lett. 119, 122502 (2017)]. The determination of the neutron skin has important implications for nuclear physics and astrophysics. Although the use of electroweak probes provides a largely model-independent determination of the neutron skin, the experimental challenges are enormous. Thus, the possibility that differences in the charge radii of mirror nuclei may be used as a surrogate for the neutron skin is a welcome alternative. To test the validity of this assumption we perform calculations based on a set of relativistic energy density functionals that span a wide region of values of $L$. Our results confirm that the difference in charge radii of various neutron-deficient nickel isotopes and their corresponding mirror nuclei is indeed strongly correlated to both the neutron-skin thickness and $L$. Moreover, given that various neutron-star properties are also sensitive to $L$, a data-to-data relation emerges between the difference in charge radii of mirror nuclei and the radius of low-mass neutron stars.

nucl-th↗

arXiv2017

Superconductivity at the vacancy disorder boundary in K$_x$Fe$_{2-y}$Se$_2$

The role of phase separation in the emergence of superconductivity in alkali metal doped iron selenides A$_{x}$Fe$_{2-y}$Se$_{2}$ (A = K, Rb, Cs) is revisited. High energy X-ray diffraction and Monte Carlo simulation were used to investigate the crystal structure of quenched superconducting (SC) and as-grown non-superconducting (NSC) K$_{x}$Fe$_{2-y}$Se$_{2}$ single crystals. The coexistence of superlattice structures with the in-plane $\sqrt{2}\times\sqrt{2}$ K-vacancy ordering and the $\sqrt{5}\times\sqrt{5}$ Fe-vacancy ordering were observed in SC and NSC crystals along side the \textit{I4/mmm} Fe-vacancy free phase. Moreover, in the SC crystal an Fe-vacancy disordered phase is additionally present. It appears at the boundary between the \textit{I4/mmm} vacancy free phase and the \textit{I4/m} vacancy ordered phase ($\sqrt{5}\times\sqrt{5}$). The vacancy disordered phase is most likely the host of superconductivity.

cond-mat.supr-con↗

arXiv2017

Aging, memory, and nonhierarchical energy landscape of spin jam

The notion of complex energy landscape underpins the intriguing dynamical behaviors in many complex systems ranging from polymers, to brain activity, to social networks and glass transitions. The spin glass state found in dilute magnetic alloys has been an exceptionally convenient laboratory frame for studying complex dynamics resulting from a hierarchical energy landscape with rugged funnels. Here, we show, by a bulk susceptibility and Monte Carlo simulation study, that densely populated frustrated magnets in a spin jam state exhibit much weaker memory effects than spin glasses, and the characteristic properties can be reproduced by a nonhierarchical landscape with a wide and nearly flat but rough bottom. Our results illustrate that the memory effects can be used to probe different slow dynamics of glassy materials, hence opening a window to explore their distinct energy landscapes.

cond-mat.dis-nn↗

arXiv2015

Spin Jam: a quantum-fluctuation-induced glassy state of a frustrated magnet

Since the discovery of spin glasses in dilute magnetic systems, their study has been largely focused on understanding randomness and defects as the driving mechanism. The same paradigm has also been applied to explain glassy states found in dense frustrated systems. Recently, however, it has been theoretically suggested that different mechanisms, such as quantum fluctuations and topological features, may induce glassy states in defect-free spin systems, far from the conventional dilute limit. Here we report experimental evidence for the existence of a glassy state, that we call a spin jam, in the vicinity of the clean limit of a frustrated magnet, which is insensitive to a low concentration of defects. We have studied the effect of impurities on SrCr9pGa12-9pO19 (SCGO(p)), a highly frustrated magnet, in which the magnetic Cr3+ (s=3/2) ions form a quasi-two-dimensional triangular system of bi-pyramids. Our experimental data shows that as the nonmagnetic Ga3+ impurity concentration is changed, there are two distinct phases of glassiness: a distinct exotic glassy state, which we call a "spin jam", for high magnetic concentration region (p>0.8) and a cluster spin glass for lower magnetic concentration, (p<0.8). This observation indicates that a spin jam is a unique vantage point from which the class of glassy states in dense frustrated magnets can be understood.

cond-mat.str-el↗

arXiv2015

In-plane Charge Fluctuations in Bismuth Sulfide Superconductors

Evidence for local charge fluctuations linked to a charge disproportionation of the Bi ions in the distorted lattice of superconducting LaO$_{1-x}$F$_{x}$ BiS$_{2}$ is presented. In-plane short-range distortions of sulfur atoms up to 0.3 Å in magnitude break site symmetry and create two distinct environments around Bi. Out-of-plane motion of apical sulfur brings it closer to the La-O/F doping layer with increasing $x$ that may lead to a charge transfer conduit between the doping layers and the superconducting BiS$_{2}$ planes. The mechanism for superconductivity may arise from the interplay between charge density fluctuations and an enhanced spin-orbit coupling suggested theoretically, that induces spin polarization.

cond-mat.supr-con↗

arXiv2014

Structural versus electronic distortions of symmetry-broken IrTe$_2$

We investigate atomic and electronic structures of the intriguing low temperature phase of IrTe2 using high-resolution scanning tunneling microscopy and spectroscopy. We confirm various stripe superstructures such as $\times$3, $\times$5, and $\times$8. The strong vertical and lateral distortions of the lattice for the stripe structures are observed in agreement with recent calculations. The spatial modulations of electronic density of states are clearly identified as separated from the structural distortions. These structural and spectroscopic characteristics are not consistent with the charge-density wave and soliton lattice model proposed recently. Instead, we show that the Ir (Te) dimerization together with the Ir 5d charge ordering can explain these superstructures, supporting the Ir dimerization mechanism of the phase transition.

cond-mat.str-el↗