• We investigate the theoretical framework of the $\cos 2\phi$ azimuthal asymmetry contributed by the coupling of two Boer-Mulders functions in the dilepton production unpolarized $\pi p$ Drell-Yan process by applying the transverse momentum dependent factorization at leading order. We adopt the model calculation results of the unpolarized distribution function $f_1$ and Boer-Mulders function $h_1^\perp$ of pion meson from the light-cone wave functions. We take into account the transverse momentum evolution effects for both the distribution functions of pion and proton by adopting the existed extraction of the nonperturbative Sudakov form factor for the pion and proton distribution functions. An approximate kernel is included to deal with the energy dependence of the Boer-Mulders function related twist-3 correlation function $T_{q,F}^{(\sigma)}(x,x)$ needed in the calculation. We numerically estimate the Boer-Mulders asymmetry $\nu_{BM}$ as the functions of $x_p$, $x_\pi$, $x_F$ and $q_T$ considering the kinematics at COMPASS Collaboration.
  • In this paper, we propose a novel object detection framework named "Deep Regionlets" by establishing a bridge between deep neural networks and conventional detection schema for accurate generic object detection. Motivated by the advantages of regionlets on modeling object deformation and multiple aspect ratios, we incorporate regionlet into an end-to-end trainable deep learning framework. The deep regionlets framework consists of a region selection network and a deep regionlet learning module. Specifically, given a detection bounding box proposal, the region selection network serves as a guidance on where to select regions to learn the features from. The regionlet learning module focuses on local feature selection and transformation to alleviate local variations. To this end, we first realize non-rectangular region selection within the detection framework to accommodate variations in object appearance. Moreover, we further design a "gating network" within the regionlet leaning module to enable soft regionlet selection and pooling. The Deep Regionlets framework is trained end-to-end without additional efforts. We perform ablation studies on its behavior and conduct extensive experiments on the PASCAL VOC and Microsoft COCO dataset. The proposed framework outperforms state-of-the-art algorithms, such as RetinaNet and Mask R-CNN, even without additional segmentation labels.
  • We use a 200 $h^{-1}Mpc$ a side N-body simulation to study the mass accretion history (MAH) of dark matter halos to be accreted by larger halos, which we call infall halos. We define a quantity $a_{\rm nf}\equiv (1+z_{\rm f})/(1+z_{\rm peak})$ to characterize the MAH of infall halos, where $z_{\rm peak}$ and $z_{\rm f}$ are the accretion and formation redshifts, respectively. We find that, at given $z_{\rm peak}$, their MAH is bimodal. Infall halos are dominated by a young population at high redshift and by an old population at low redshift. For the young population, the $a_{\rm nf}$ distribution is narrow and peaks at about $1.2$, independent of $z_{\rm peak}$, while for the old population, the peak position and width of the $a_{\rm nf}$ distribution both increases with decreasing $z_{\rm peak}$ and are both larger than those of the young population. This bimodal distribution is found to be closely connected to the two phases in the MAHs of halos. While members of the young population are still in the fast accretion phase at $z_{\rm peak}$, those of the old population have already entered the slow accretion phase at $z_{\rm peak}$. This bimodal distribution is not found for the whole halo population, nor is it seen in halo merger trees generated with the extended Press-Schechter formalism. The infall halo population at $z_{\rm peak}$ are, on average, younger than the whole halo population of similar masses identified at the same redshift. We discuss the implications of our findings in connection to the bimodal color distribution of observed galaxies and to the link between central and satellite galaxies.
  • We study the Collins function $H^\perp_{1}$ of the $\Lambda$ hyperon, which describes the fragmentation of a transversely polarized quark into an unpolarized $\Lambda$ hyperon. We calculate $H^\perp_{1}$ for light quarks of the $\Lambda$ hyperon, in the diquark spectator model with a Gaussian form factor for the hyperon-quark-diquark vertex. The model calculation includes contributions from both the scalar diquark and vector diquark spectators. Using the model result, we estimate the weighted $\cos2\phi_0$ asymmetry in the process $e^+e^- \to \Lambda\bar{\Lambda}X$ contributed by the coupling of two Collins functions. The QCD evolution effects for the first $k_T$-moment of the Collins function and the unpolarized fragmentation function $D_1(z)$ are also included. The results show that asymmetry is sizable and measurable at the kinematical configurations of Belle and BaBar experiments. We also find that the evolution effects play an important role in the phenomenological analysis.
  • We propose a novel method for real-time face alignment in videos based on a recurrent encoder-decoder network model. Our proposed model predicts 2D facial point heat maps regularized by both detection and regression loss, while uniquely exploiting recurrent learning at both spatial and temporal dimensions. At the spatial level, we add a feedback loop connection between the combined output response map and the input, in order to enable iterative coarse-to-fine face alignment using a single network model, instead of relying on traditional cascaded model ensembles. At the temporal level, we first decouple the features in the bottleneck of the network into temporal-variant factors, such as pose and expression, and temporal-invariant factors, such as identity information. Temporal recurrent learning is then applied to the decoupled temporal-variant features. We show that such feature disentangling yields better generalization and significantly more accurate results at test time. We perform a comprehensive experimental analysis, showing the importance of each component of our proposed model, as well as superior results over the state of the art and several variations of our method in standard datasets.
  • We investigate the Sivers asymmetry in the pion-induced single polarized Drell-Yan process in the theoretical framework of the transverse momentum dependent factorization up to next-to-leading logarithmic of QCD. Within the TMD evolution formalism of parton distribution functions, the recently extracted nonperturbative Sudakov form factor for the pion distribution functions as well as the one for the Sivers function of the proton are applied to numerically estimate the Sivers asymmetry in the $\pi^- p$ Drell-Yan at the kinematics of the COMPASS at CERN. The effect of the energy-scale dependence of the Qiu-Sterman function $T_{q,F}(x,x,\mu)$ to the asymmetry is also studied. We find that our prediction on the Sivers asymmetries as functions of $x_p$, $x_\pi$, $x_F$ and $q_\perp$ is consistent with the recently COMPASS measurement.
  • We investigate the interplay between charge order and superconductivity near an antiferromagnetic quantum critical point using sign-problem-free Quantum Monte Carlo simulations. We establish that, when the electronic dispersion is particle-hole symmetric, the system has an emergent SU(2) symmetry that implies a degeneracy between $d$-wave superconductivity and charge order with $d$-wave form factor. Deviations from particle-hole symmetry, however, rapidly lift this degeneracy, despite the fact that the SU(2) symmetry is preserved at low energies. As a result, we find a strong suppression of charge order caused by the competing, leading superconducting instability. Across the antiferromagnetic phase transition, we also observe a shift in the charge order wave-vector from diagonal to axial. We discuss the implications of our results to the universal phase diagram of antiferromagnetic quantum-critical metals and to the elucidation of the charge order experimentally observed in the cuprates.
  • We study the transverse momentum spectrum of dilepton produced in the unpolarized $\pi^- N$ Drell-Yan process, using transverse momentum dependent factorization up to next-to-logarithmic order of QCD. We extract the nonperturbative Sudakov form factor for the pion in the evolution formalism of the unpolarized TMD distribution function, by fitting the experimental data collected by the E615 Collaboration at Fermilab. With the extracted Sudakov factor, we calculate the normalized differential cross section with respect to transverse momentum of the dimuon and compare it with the recent measurement by the COMPASS Collaboration.
  • While going deeper has been witnessed to improve the performance of convolutional neural networks (CNN), going smaller for CNN has received increasing attention recently due to its attractiveness for mobile/embedded applications. It remains an active and important topic how to design a small network while retaining the performance of large and deep CNNs (e.g., Inception Nets, ResNets). Albeit there are already intensive studies on compressing the size of CNNs, the considerable drop of performance is still a key concern in many designs. This paper addresses this concern with several new contributions. First, we propose a simple yet powerful method for compressing the size of deep CNNs based on parameter binarization. The striking difference from most previous work on parameter binarization/quantization lies at different treatments of $1\times 1$ convolutions and $k\times k$ convolutions ($k>1$), where we only binarize $k\times k$ convolutions into binary patterns. The resulting networks are referred to as pattern networks. By doing this, we show that previous deep CNNs such as GoogLeNet and Inception-type Nets can be compressed dramatically with marginal drop in performance. Second, in light of the different functionalities of $1\times 1$ (data projection/transformation) and $k\times k$ convolutions (pattern extraction), we propose a new block structure codenamed the pattern residual block that adds transformed feature maps generated by $1\times 1$ convolutions to the pattern feature maps generated by $k\times k$ convolutions, based on which we design a small network with $\sim 1$ million parameters. Combining with our parameter binarization, we achieve better performance on ImageNet than using similar sized networks including recently released Google MobileNets.
  • We calculate the two leading-twist transverse momentum dependent distribution functions of the pion meson, the unpolarized distribution $f_{1\pi}(x,\bm{k}^2_T)$ and the Boer-Mulders function $h_{1\pi}^\perp (x,\bm{k}^2_T)$, using the pion wave functions derived from a light-cone approach. The evolution effect of the first $\bm k_T$-moment of the pion Boer-Mulders function is studied by employing an approximate evolution kernel. Using the model resulting distributions, we predict the transverse momentum weighted $\cos2\phi$ azimuthal asymmetry in the unpolarized $\pi^- p$ Drell-Yan process which can be measured at COMPASS in the near future.
  • In several unconventional superconductors, the highest superconducting transition temperature $T_{c}$ is found in a region of the phase diagram where the antiferromagnetic transition temperature extrapolates to zero, signaling a putative quantum critical point. The elucidation of the interplay between these two phenomena - high-$T_{c}$ superconductivity and magnetic quantum criticality - remains an important piece of the complex puzzle of unconventional superconductivity. In this paper, we combine sign-problem-free Quantum Monte Carlo simulations and field-theoretical analytical calculations to unveil the microscopic mechanism responsible for the superconducting instability of a general low-energy model, called spin-fermion model. In this approach, low-energy electronic states interact with each other via the exchange of quantum critical magnetic fluctuations. We find that even in the regime of moderately strong interactions, both the superconducting transition temperature and the pairing susceptibility are governed not by the properties of the entire Fermi surface, but instead by the properties of small portions of the Fermi surface called hot spots. Moreover, $T_{c}$ increases with increasing interaction strength, until it starts to saturate at the crossover from hot-spots dominated to Fermi-surface dominated pairing. Our work provides not only invaluable insights into the system parameters that most strongly affect $T_{c}$, but also important benchmarks to assess the origin of superconductivity in both microscopic models and actual materials.
  • Image captioning is a challenging problem owing to the complexity in understanding the image content and diverse ways of describing it in natural language. Recent advances in deep neural networks have substantially improved the performance of this task. Most state-of-the-art approaches follow an encoder-decoder framework, which generates captions using a sequential recurrent prediction model. However, in this paper, we introduce a novel decision-making framework for image captioning. We utilize a "policy network" and a "value network" to collaboratively generate captions. The policy network serves as a local guidance by providing the confidence of predicting the next word according to the current state. Additionally, the value network serves as a global and lookahead guidance by evaluating all possible extensions of the current state. In essence, it adjusts the goal of predicting the correct words towards the goal of generating captions similar to the ground truth captions. We train both networks using an actor-critic reinforcement learning model, with a novel reward defined by visual-semantic embedding. Extensive experiments and analyses on the Microsoft COCO dataset show that the proposed framework outperforms state-of-the-art approaches across different evaluation metrics.
  • We study the longitudinal-transverse double-spin asymmetry with a $\cos\phi_S$ modulation in semi-inclusive deep inelastic scattering for charged and neutral pions production. We consider the particular case in which the transverse momentum of the final state hadron is integrated out. The corresponding asymmetry may be contributed by two parts: one is the convolution of the twist-3 distribution function $g_{T}(x)$ and the unpolarized fragmentation function $D_1(z)$, the other is related to the coupling of the transversity distribution function $h_1(x)$ and the collinear twist-3 fragmentation function $\tilde{E}(z)$. We take into account both contributions to predict the $\cos\phi_S$ asymmetry at the kinematics of CLAS12 and a future Electron Ion Collider. We find that the asymmetry of pion production at CLAS12 is sizable, and $\tilde{E}(z)$ can play an important role in the asymmetry in the large-$z$ region.
  • We propose a novel recurrent encoder-decoder network model for real-time video-based face alignment. Our proposed model predicts 2D facial point maps regularized by a regression loss, while uniquely exploiting recurrent learning at both spatial and temporal dimensions. At the spatial level, we add a feedback loop connection between the combined output response map and the input, in order to enable iterative coarse-to-fine face alignment using a single network model. At the temporal level, we first decouple the features in the bottleneck of the network into temporal-variant factors, such as pose and expression, and temporal-invariant factors, such as identity information. Temporal recurrent learning is then applied to the decoupled temporal-variant features, yielding better generalization and significantly more accurate results at test time. We perform a comprehensive experimental analysis, showing the importance of each component of our proposed model, as well as superior results over the state-of-the-art in standard datasets.
  • We study the transverse single-spin asymmetry with a $\sin\phi_S$ modulation in semi-inclusive deep inelastic scattering. Particularly, we consider the case in which the transverse momentum of the final state hadron is integrated out. Thus, the asymmetry is merely contributed by the coupling of the transversity distribution function $h_1(x)$ and the twist-3 collinear fragmentation function $\tilde{H}(z)$. Using the available parametrization of $h_1(x)$ from SIDIS data and the recent extracted result for $\tilde{H}(z)$, we predict the $\sin\phi_S$ asymmetry for charged and neutral pion production at an Electron Ion Collider. We find that the asymmetry is sizable and could be measured. We also study the impact of the leading-order QCD evolution effect and find that it affects the $\sin\phi_S$ asymmetry at EIC considerably.
  • We study the $\cos{\phi_h}$ azimuthal asymmetry in double polarized semi-inclusive pion production by considering dynamical twist-3 effects. In particular, we evaluate the role of the transverse momentum dependent distributions $e_L(x, k_T^2)$ and $g_L^\perp(x, k_T^2)$ on the asymmetry. Using two different sets of spectator model results for these distributions, we predict the $\cos{\phi_h}$ asymmetry of $\pi^+$, $\pi^-$, and $\pi^0$ at the kinematic configuration available at CLAS, HERMES and COMPASS. Our estimates show that the asymmetries are positive for all the pions and could be accessed by CLAS and HERMES. We also find that $g_L^\perp$ gives the dominant contribution to the $\cos\phi_h$ asymmetry, while the contribution of $e_L$ is almost negligible.
  • Most iron-based superconductors undergo a transition to a magnetically ordered state characterized by staggered stripes of parallel spins. With ordering vectors $(\pi,0)$ or $(0,\pi)$, this magnetic state breaks the high-temperature tetragonal symmetry of the system, which is manifested by a splitting of the lattice Bragg peaks. Remarkably, recent experiments in hole-doped iron arsenides reported an ordered state that displays magnetic Bragg peaks at $(\pi,0)$ and $(0,\pi)$ but remains tetragonal. Despite being inconsistent with a magnetic stripe configuration, this unusual magnetic phase can be described in terms of a double-$\mathbf{Q}$ magnetic structure consisting of an equal-weight superposition of the ordering vectors $(\pi,0)$ and $(0,\pi)$. Here we show that a non-collinear double-$\mathbf{Q}$ magnetic configuration, dubbed \emph{orthomagnetic}, arises naturally within an itinerant three-band microscopic model for the iron pnictides. In particular, we find that strong deviations from perfect nesting and residual interactions between the electron pockets favor the orthomagnetic over the stripe magnetic state. Using an effective low-energy model, we also calculate the spin-wave spectrum of the orthomagnetic state. In contrast to the stripe state, there are three Goldstone modes, manifested in all diagonal and one off-diagonal component of the spin-spin correlation function. The total magnetic structure factor displays two anisotropic spin-wave branches emerging from both $(\pi,0)$ and $(0,\pi)$ momenta, in contrast to the case of domains of stripe order, where only one spin-wave branch emerges from each momentum. We propose that these unique features of the orthomagnetic state can be used to unambiguously distinguish it from the stripe state via neutron scattering experiments, and discuss the implications of its existence to the nature of the magnetism of the iron arsenides.
  • Motivated by recent experiments in Ba$_{1-x}$K$_{x}$Fe$_{2}$As$_{2}$ {[}A. E. B\"ohmer \textit{et al}, to be published{]}, we analyze the type of spin-density wave (SDW) order in doped iron-pnictides and the discontinuities of the superconducting transition temperature $T_{c}$ in the coexistence phase with SDW magnetism. By tracking the magnetic transition line $T_{N}(x)$ towards optimal doping within an itinerant fermionic model, we find a sequence of transitions from the stripe-orthorhombic ($C_{2}$) SDW order to the tetragonal ($C_{4}$) order and then back to the $C_{2}$ order. We argue that the superconducting $T_{c}$ has two discontinuities -- it jumps to a smaller value upon entering the coexistence region with the $C_{4}$ magnetic phase, and then jumps to a larger value inside the SDW state when it crosses the boundary between the $C_{4}$ and $C_{2}$ SDW orders. The full agreement with the experimental phase diagram provides a strong indication that the itinerant approach is adequate to describe the physics of weakly/moderately doped iron-pnictides.
  • This paper proposes to go beyond the state-of-the-art deep convolutional neural network (CNN) by incorporating the information from object detection, focusing on dealing with fine-grained image classification. Unfortunately, CNN suffers from over-fiting when it is trained on existing fine-grained image classification benchmarks, which typically only consist of less than a few tens of thousands training images. Therefore, we first construct a large-scale fine-grained car recognition dataset that consists of 333 car classes with more than 150 thousand training images. With this large-scale dataset, we are able to build a strong baseline for CNN with top-1 classification accuracy of 81.6%. One major challenge in fine-grained image classification is that many classes are very similar to each other while having large within-class variation. One contributing factor to the within-class variation is cluttered image background. However, the existing CNN training takes uniform window sampling over the image, acting as blind on the location of the object of interest. In contrast, this paper proposes an \emph{object-centric sampling} (OCS) scheme that samples image windows based on the object location information. The challenge in using the location information lies in how to design powerful object detector and how to handle the imperfectness of detection results. To that end, we design a saliency-aware object detection approach specific for the setting of fine-grained image classification, and the uncertainty of detection results are naturally handled in our OCS scheme. Our framework is demonstrated to be very effective, improving top-1 accuracy to 89.3% (from 81.6%) on the large-scale fine-grained car classification dataset.
  • A revised Incremental Conductance (IncCond) maximum power point tracking (MPPT) algorithm for PV generation systems is proposed in this paper. The commonly adopted traditional IncCond method uses a constant step size for voltage adjustment and is difficult to achieve both a good tracking performance and quick elimination of the oscillations, especially under the dramatic changes of the environment conditions. For the revised algorithm, the incremental voltage change step size is adaptively adjusted based on the slope of the power-voltage (P-V) curve. An accelerating factor and a decelerating factor are further applied to adjust the voltage step change considering whether the sign of the P-V curve slope remains the same or not in a subsequent tracking step. In addition, the upper bound of the maximum voltage step change is also updated considering the information of sign changes. The revised MPPT algorithm can quickly track the maximum power points (MPPs) and remove the oscillation of the actual operation points around the real MPPs. The effectiveness of the revised algorithm is demonstrated using a simulation.
  • This paper addresses the challenge of establishing a bridge between deep convolutional neural networks and conventional object detection frameworks for accurate and efficient generic object detection. We introduce Dense Neural Patterns, short for DNPs, which are dense local features derived from discriminatively trained deep convolutional neural networks. DNPs can be easily plugged into conventional detection frameworks in the same way as other dense local features(like HOG or LBP). The effectiveness of the proposed approach is demonstrated with the Regionlets object detection framework. It achieved 46.1% mean average precision on the PASCAL VOC 2007 dataset, and 44.1% on the PASCAL VOC 2010 dataset, which dramatically improves the original Regionlets approach without DNPs.
  • We investigate the fate of the orthorhombic stripe-type magnetic state (ordering vectors $\left(\pi,0\right)$/$\left(0,\pi\right)$), observed in most iron-pnictide superconductors, in the presence of localized magnetic moments that tend to form a Neel state (ordering vector $\left(\pi,\pi\right)$). We show that before long-range Neel order sets in, the coupling between the conduction electrons and the fluctuations of the local moments favors an unusual magnetic state consisting of a coherent superposition of the $\left(\pi,0\right)$ and $\left(0,\pi\right)$ orders that preserves tetragonal symmetry. The magnetization of this state is non-uniform and induces a simultaneous checkerboard charge order. We discuss signatures of this magnetic configuration on the electronic spectrum and its impact on the superconducting state, showing that its phase space for coexistence with the $s^{+-}$ state is smaller than the stripe-type state. Our results shed light on recent experimental observations on $\mathrm{Ba(Fe_{1-x}Mn_{x})_{2}As_{2}}$ compounds, where the Neel-type local Mn moments interact with the Fe conduction electrons.
  • Patient respiratory signal associated with the cone beam CT (CBCT) projections is important for lung cancer radiotherapy. In contrast to monitoring an external surrogate of respiration, such signal can be extracted directly from the CBCT projections. In this paper, we propose a novel local principle component analysis (LPCA) method to extract the respiratory signal by distinguishing the respiration motion-induced content change from the gantry rotation-induced content change in the CBCT projections. The LPCA method is evaluated by comparing with three state-of-the-art projection-based methods, namely, the Amsterdam Shroud (AS) method, the intensity analysis (IA) method, and the Fourier-transform based phase analysis (FT-p) method. The clinical CBCT projection data of eight patients, acquired under various clinical scenarios, were used to investigate the performance of each method. We found that the proposed LPCA method has demonstrated the best overall performance for cases tested and thus is a promising technique for extracting respiratory signal. We also identified the applicability of each existing method.
  • Considering model Hamiltonians that respect the symmetry properties of the pnictides, it is argued that pairing interactions that couple electrons at different orbitals with an orbital-dependent pairing strength inevitably lead to interband pairing matrix elements, at least in some regions of the Brillouin zone. Such interband pairing has not been considered of relevance in multiorbital systems in previous investigations. It is also observed that if, instead, a purely intraband pairing interaction is postulated, this requires that the pairing operator has the form $\Delta^+(k)=f(k) \sum_{\alpha} d^+_{k,\alpha,\uparrow}d^+_{-k, \alpha,\downarrow}$ where $\alpha$ labels the orbitals considered in the model and f(k) arises from the spatial location of the coupled electrons or holes. This means that the gaps at two different Fermi surfaces involving momenta $k_F$ and $k'_F$ can only differ by the ratio $f(k_F)/ f(k'_F)$ and that electrons in different orbitals must be subject to the same pairing attraction, thus requiring fine tuning. These results suggest that previously neglected interband pairing tendencies could actually be of relevance in a microscopic description of the pairing mechanism in the pnictides.