Dynamical correlation enhanced orbital magnetization in VI$_{3}$

The effect of electronic correlations on the orbital magnetization in real materials has not been explored beyond a static mean-field level. Based on the dynamical mean-field theory, the effect of electronic correlations on the orbital magnetization in layered ferromagnet VI$_3$ has been studied. A comparison drawn with the results obtained from density functional theory calculations robustly establishes the crucial role of dynamical correlations in this case. In contrast to the density functional theory that leads to negligible orbital magnetization in VI$_3$, in dynamical mean-field approach the orbital magnetization is greatly enhanced. Further analysis show that this enhancement is mainly due to the enhanced local circulations of electrons, which can be attributed to a better description of the localization behavior of correlated electrons in VI$_3$. The conclusion drawn in our study could be applicable to a wide range of layered materials in this class.

Neural network representation of electronic structure from ab initio molecular dynamics

Despite their rich information content, electronic structure data amassed at high volumes in ab initio molecular dynamics simulations are generally under-utilized. We introduce a transferable high-fidelity neural network representation of such data in the form of tight-binding Hamiltonians for crystalline materials. This predictive representation of ab initio electronic structure, combined with machine-learning boosted molecular dynamics, enables efficient and accurate electronic evolution and sampling. When applied to a one-dimension charge-density wave material, carbyne, we are able to compute the spectral function and optical conductivity in the canonical ensemble. The spectral functions evaluated during soliton-antisoliton pair annihilation process reveal significant renormalization of low-energy edge modes due to retarded electron-lattice coupling beyond the Born-Oppenheimer limit. The availability of an efficient and reusable surrogate model for the electronic structure dynamical system will enable calculating many interesting physical properties, paving way to previously inaccessible or challenging avenues in materials modeling.

Soft Gradient Boosting Machine

Gradient Boosting Machine has proven to be one successful function approximator and has been widely used in a variety of areas. However, since the training procedure of each base learner has to take the sequential order, it is infeasible to parallelize the training process among base learners for speed-up. In addition, under online or incremental learning settings, GBMs achieved sub-optimal performance due to the fact that the previously trained base learners can not adapt with the environment once trained. In this work, we propose the soft Gradient Boosting Machine (sGBM) by wiring multiple differentiable base learners together, by injecting both local and global objectives inspired from gradient boosting, all base learners can then be jointly optimized with linear speed-up. When using differentiable soft decision trees as base learner, such device can be regarded as an alternative version of the (hard) gradient boosting decision trees with extra benefits. Experimental results showed that, sGBM enjoys much higher time efficiency with better accuracy, given the same base learner in both on-line and off-line settings.

Learning to Confuse: Generating Training Time Adversarial Data with Auto-Encoder

In this work, we consider one challenging training time attack by modifying training data with bounded perturbation, hoping to manipulate the behavior (both targeted or non-targeted) of any corresponding trained classifier during test time when facing clean samples. To achieve this, we proposed to use an auto-encoder-like network to generate the pertubation on the training data paired with one differentiable system acting as the imaginary victim classifier. The perturbation generator will learn to update its weights by watching the training procedure of the imaginary classifier in order to produce the most harmful and imperceivable noise which in turn will lead the lowest generalization power for the victim classifier. This can be formulated into a non-linear equality constrained optimization problem. Unlike GANs, solving such problem is computationally challenging, we then proposed a simple yet effective procedure to decouple the alternating updates for the two networks for stability. The method proposed in this paper can be easily extended to the label specific setting where the attacker can manipulate the predictions of the victim classifiers according to some predefined rules rather than only making wrong predictions. Experiments on various datasets including CIFAR-10 and a reduced version of ImageNet confirmed the effectiveness of the proposed method and empirical results showed that, such bounded perturbation have good transferability regardless of which classifier the victim is actually using on image data.

Electron-phonon coupling and the Coexistence of Superconductivity and Charge-Density Wave in Monolayer NbSe2

Monolayer 2H-NbSe2 has recently been shown to be a 2-dimensional superconductor, with a coexisting charge-density wave (CDW). As both phenomena are intimately related to electron-lattice interaction, a natural question is how superconductivity and CDW are interrelated through electron-phonon coupling (EPC), which is important to the understanding of 2-dimensional superconductivity. This work investigates the superconductivity of monolayer NbSe2 in CDW phase using the anisotropic Migdal-Eliashberg formalism based on first principles calculations. The mechanism of the competition between and coexistence of the superconductivity and CDW is studied in detail by analyzing EPC. It is found that the intra-pocket scattering is related to superconductivity, leading to almost constant value of superconducting gaps on parts of the Fermi surface. The inter-pocket scattering is found to be responsible for CDW, leading to partial or full bandgap on the remaining Fermi surface. Recent experiment indicates that there is transitioning from regular superconductivity in thin-film NbSe2 to two-gap superconductivity in the bulk, which is shown here to have its origin in the extent of Fermi surface gapping of K and K pockets induced by CDW. Overall blue shifts of the phonons and sharp decrease of Eliashberg spectrum are found when the CDW forms.

Multi-Layered Gradient Boosting Decision Trees

Multi-layered representation is believed to be the key ingredient of deep neural networks especially in cognitive tasks like computer vision. While non-differentiable models such as gradient boosting decision trees (GBDTs) are the dominant methods for modeling discrete or tabular data, they are hard to incorporate with such representation learning ability. In this work, we propose the multi-layered GBDT forest (mGBDTs), with an explicit emphasis on exploring the ability to learn hierarchical representations by stacking several layers of regression GBDTs as its building block. The model can be jointly trained by a variant of target propagation across layers, without the need to derive back-propagation nor differentiability. Experiments and visualizations confirmed the effectiveness of the model in terms of performance and representation learning ability.

Geometric effects in the effective-mass theory and topological optical superlattices

Cold atoms tailored by an optical lattice have become a fascinating arena for simulating quantum physics. In this area, one important and challenging problem is creating effective spin-orbit coupling (SOC), especially for fashioning a cold atomic gas into a topological phase, for which prevailing approaches mainly rely on the Raman coupling between the atomic internal states and a laser field. Herein, a strategy for realizing effective SOC is proposed by exploiting the geometric effects in the effective-mass theory, without resorting to internal atomic states. It is shown that the geometry of Bloch states can have nontrivial effects on the wave-mechanical states under external fields, leading to effective SOC and an effective Darwin term, which have been neglected in the standard effective-mass approximation. It is demonstrated that these relativisticlike effects can be employed to introduce effective SOC in a two-dimensional optical superlattice, and induce a nontrivial topological phase.

Extend natural neighbor: a novel classification method with self-adaptive neighborhood parameters in different stages

Various kinds of k-nearest neighbor (KNN) based classification methods are the bases of many well-established and high-performance pattern-recognition techniques, but both of them are vulnerable to their parameter choice. Essentially, the challenge is to detect the neighborhood of various data sets, while utterly ignorant of the data characteristic. This article introduces a new supervised classification method: the extend natural neighbor (ENaN) method, and shows that it provides a better classification result without choosing the neighborhood parameter artificially. Unlike the original KNN based method which needs a prior k, the ENaNE method predicts different k in different stages. Therefore, the ENaNE method is able to learn more from flexible neighbor information both in training stage and testing stage, and provide a better classification result.

Computational Study of the Magnetic Structure of Na$_2$IrO$_3$

The magnetic structure of honeycomb iridate Na$_2$IrO$_3$ is of paramount importance to its exotic properties. The magnetic order is established experimentally to be zigzag antiferromagnetic. However, the previous assignment of ordered moment to the $bm{a}$-axis is tentative. We examine the magnetic structure of Na$_{2}$IrO$_{3}$ using first-principles methods. Our calculations reveal that total energy is minimized when the zigzag antiferromagnetic order is magnetized along $bm{g}approxbm{a}+bm{c}$. Such a magnetic configuration is explained by adding anisotropic interactions to the nearest-neighbor Kitaev-Heisenberg model. Spin-wave spectrum is also calculated, where the calculated spin gap of $10.4$ meV can in principle be measured by future inelastic neutron scattering experiments. Finally we emphasize that our proposal is consistent with all known experimental evidence, including the most relevant resonant x-ray magnetic scattering measurements [X. Liu emph{et al.} {Phys. Rev. B} textbf{83}, 220403(R) (2011)].