arXiv subjects

Zhenda Xie

Publications and source records attributed to Zhenda Xie.

At least 55 records · Page 3Linked to original sources

arXiv2022

Spatiotemporal mode-locking and photonic flywheel in multimode microresonators

Dissipative Kerr soliton (DKS) frequency combs - also known as microcombs - have arguably created a new field in cavity nonlinear photonics, with a strong cross-fertilization between theoretical, experimental, and technological research. Spatiotemporal mode-locking (STML) not only add new degrees of freedom to ultrafast laser technology, but also provide new insights for implementing analogue computers and heuristic optimizers with photonics. Here, we combine the principles of DKS and STML for the first time to demonstrate the STML DKS by developing an unexplored ultrahigh-quality-factor Fabry-Perot microresonator based on graded index multimode fiber (GRIN-MMF). Using the intermodal stimulated Brillouin scattering, we can selectively excite either the eigenmode DKS or the STML DKS. Furthermore, we demonstrate an ultralow noise microcomb that enhances the photonic flywheel performance in both the fundamental comb linewidth and DKS timing jitter. The demonstrated fundamental comb linewidth of 400 mHz and DKS timing jitter of 500 attosecond represent improvements of 25x and 2.5x, respectively, from the state-of-the-art. Our results show the potential of GRIN-MMF FP microresonators as an ideal testbed for high-dimensional nonlinear cavity dynamics and photonic flywheel with ultrahigh coherence and ultralow timing jitter.

physics.optics↗

arXiv2022

Arbitrary coherent distributions in a programmable quantum walk

The coherent superposition of position states in a quantum walk (QW) can be precisely engineered towards the desired distributions to meet the need of quantum information applications. The coherent distribution can make full use of quantum parallel in computation and simulation. Particularly, the uniform superposition provides the robust non-locality, which has wide applications such as the generation of genuine multi-bit random numbers without post-processing. We experimentally demonstrate that the rich dynamics featured with arbitrary coherent distributions can be obtained by introducing different sets of the time- and position-dependent operations. Such a QW is realized by a resource-constant and flexible optical circuit, in which the variable operation is executed based on a Sagnac interferometer in an intrinsically stable and precisely controlled way. Our results contribute to the practical realization of quantum-walk-based quantum computation, quantum simulations and quantum information protocols.

quant-ph↗

arXiv2022

Maximal coin-walker entanglement in a ballistic quantum walk

We report the position-inhomogeneous quantum walk (IQW) can be utilized to produce the maximal high dimensional entanglement while maintaining the quadratic speedup spread of the wave-function. Our calculations show that the maximal coin-walker entanglement can be generated in any odd steps or asymptotically in even steps, and the nearly maximal entanglement can be obtained in even steps after $2$. We implement the IQW by a stable resource-saving time-bin optical network, in which a polarization Sagnac loop is employed to realize the precisely tunable phase shift. Our approach opens up an efficient way for high-dimensional entanglement engineering as well as promotes investigations on the role of coin-walker interactions in QW based applications.

quant-ph↗

arXiv2021

Breaking Shortcut: Exploring Fully Convolutional Cycle-Consistency for Video Correspondence Learning

Previous cycle-consistency correspondence learning methods usually leverage image patches for training. In this paper, we present a fully convolutional method, which is simpler and more coherent to the inference process. While directly applying fully convolutional training results in model collapse, we study the underline reason behind this collapse phenomenon, indicating that the absolute positions of pixels provide a shortcut to easily accomplish cycle-consistence, which hinders the learning of meaningful visual representations. To break this absolute position shortcut, we propose to apply different crops for forward and backward frames, and adopt feature warping to establish correspondence between two crops of a same frame. The former technique enforces the corresponding pixels at forward and back tracks to have different absolute positions, and the latter effectively blocks the shortcuts going between forward and back tracks. In three label propagation benchmarks for pose tracking, face landmark tracking and video object segmentation, our method largely improves the results of vanilla fully convolutional cycle-consistency method, achieving very competitive performance compared with the self-supervised state-of-the-art approaches. Our trained model and code are available at \url{https://github.com/Steve-Tod/STFC3}.

cs.CV↗

arXiv2021

Self-Supervised Learning with Swin Transformers

We are witnessing a modeling shift from CNN to Transformers in computer vision. In this work, we present a self-supervised learning approach called MoBY, with Vision Transformers as its backbone architecture. The approach basically has no new inventions, which is combined from MoCo v2 and BYOL and tuned to achieve reasonably high accuracy on ImageNet-1K linear evaluation: 72.8% and 75.0% top-1 accuracy using DeiT-S and Swin-T, respectively, by 300-epoch training. The performance is slightly better than recent works of MoCo v3 and DINO which adopt DeiT as the backbone, but with much lighter tricks. More importantly, the general-purpose Swin Transformer backbone enables us to also evaluate the learnt representations on downstream tasks such as object detection and semantic segmentation, in contrast to a few recent approaches built on ViT/DeiT which only report linear evaluation results on ImageNet-1K due to ViT/DeiT not tamed for these dense prediction tasks. We hope our results can facilitate more comprehensive evaluation of self-supervised learning methods designed for Transformer architectures. Our code and models are available at https://github.com/SwinTransformer/Transformer-SSL, which will be continually enriched.

cs.CV↗

arXiv2021

Propagate Yourself: Exploring Pixel-Level Consistency for Unsupervised Visual Representation Learning

Contrastive learning methods for unsupervised visual representation learning have reached remarkable levels of transfer performance. We argue that the power of contrastive learning has yet to be fully unleashed, as current methods are trained only on instance-level pretext tasks, leading to representations that may be sub-optimal for downstream tasks requiring dense pixel predictions. In this paper, we introduce pixel-level pretext tasks for learning dense feature representations. The first task directly applies contrastive learning at the pixel level. We additionally propose a pixel-to-propagation consistency task that produces better results, even surpassing the state-of-the-art approaches by a large margin. Specifically, it achieves 60.2 AP, 41.4 / 40.5 mAP and 77.2 mIoU when transferred to Pascal VOC object detection (C4), COCO object detection (FPN / C4) and Cityscapes semantic segmentation using a ResNet-50 backbone network, which are 2.6 AP, 0.8 / 1.0 mAP and 1.0 mIoU better than the previous best methods built on instance-level contrastive learning. Moreover, the pixel-level pretext tasks are found to be effective for pre-training not only regular backbone networks but also head networks used for dense downstream tasks, and are complementary to instance-level contrastive methods. These results demonstrate the strong potential of defining pretext tasks at the pixel level, and suggest a new path forward in unsupervised visual representation learning. Code is available at \url{https://github.com/zdaxie/PixPro}.

cs.CV↗

arXiv2020

Ultra-bright multiplexed energy-time entangled photon generation from lithium niobate on insulator chip

High-flux entangled photon source is the key resource for quantum optical study and application. Here it is realized in a lithium niobate on isolator (LNOI) chip, with 2.79*10^11 Hz/mW photon pair rate and 1.53*10^9 Hz/nm/mW spectral brightness. These data are boosted by over two orders of magnitude compared to existing technologies. A 130-nm broad bandwidth is engineered for 8-channel multiplexed energy-time entanglement. Harnessed by high-extinction frequency correlation and Franson interferences up to 99.17% visibility, such energy-time entanglement multiplexing further enhances high-flux data rate, and warrants broad applications in quantum information processing on a chip.

quant-ph↗

arXiv2020

Spatially Adaptive Inference with Stochastic Feature Sampling and Interpolation

In the feature maps of CNNs, there commonly exists considerable spatial redundancy that leads to much repetitive processing. Towards reducing this superfluous computation, we propose to compute features only at sparsely sampled locations, which are probabilistically chosen according to activation responses, and then densely reconstruct the feature map with an efficient interpolation procedure. With this sampling-interpolation scheme, our network avoids expending computation on spatial locations that can be effectively interpolated, while being robust to activation prediction errors through broadly distributed sampling. A technical challenge of this sampling-based approach is that the binary decision variables for representing discrete sampling locations are non-differentiable, making them incompatible with backpropagation. To circumvent this issue, we make use of a reparameterization trick based on the Gumbel-Softmax distribution, with which backpropagation can iterate these variables towards binary values. The presented network is experimentally shown to save substantial computation while maintaining accuracy over a variety of computer vision tasks.

cs.CV↗

arXiv2020

Photonic flywheel in a monolithic fiber resonator

We demonstrate the first compact photonic flywheel with sub-fs time jitter (averaging times up to 10 μs) at the quantum-noise limit of a monolithic fiber resonator. Such quantum-limited performance is accessed through novel two-step pumping scheme for dissipative Kerr soliton (DKS) generation. Controllable interaction between stimulated Brillouin lasing and Kerr nonlinearity enhances the DKS coherence and mitigate the thermal instability challenge, achieving a remarkable 22-Hz intrinsic comb linewidth and an unprecedented phase noise of -180 dBc/Hz at 945 MHz carrier at free running. The scheme can be generalized to various device platforms for field-deployable precision metrology.

physics.optics↗

arXiv2020

Brillouin-Kerr soliton frequency combs in an optical microresonator

By generating a Brillouin laser in an optical microresonator, we realize a soliton Kerr microcomb through exciting the Kerr frequency comb using the generated Brillouin laser in the same cavity. The intracavity Brillouin laser pumping scheme enables us to access the soliton states with a blue-detuned input pump. Due to the ultra-narrow linewidth and the low-noise properties of the generated Brillouin laser, the observed soliton microcomb exhibits narrow-linewidth comb lines and stable repetition rate even by using a diode laser with relatively broad linewidth. Also, we demonstrate a low-noise microwave signal with phase noise levels of -24 dBc/Hz at 10 Hz, -111 dBc/Hz at 10 kHz, and -147 dBc/Hz at 1 MHz offsets for a 11.14 GHz carrier with only a free-running input pump. The easy operation of the Brillouin-Kerr soliton microcomb with excellent performance makes our scheme promising for practical applications.

physics.optics↗

arXiv2020

Parametric Instance Classification for Unsupervised Visual Feature Learning

This paper presents parametric instance classification (PIC) for unsupervised visual feature learning. Unlike the state-of-the-art approaches which do instance discrimination in a dual-branch non-parametric fashion, PIC directly performs a one-branch parametric instance classification, revealing a simple framework similar to supervised classification and without the need to address the information leakage issue. We show that the simple PIC framework can be as effective as the state-of-the-art approaches, i.e. SimCLR and MoCo v2, by adapting several common component settings used in the state-of-the-art approaches. We also propose two novel techniques to further improve effectiveness and practicality of PIC: 1) a sliding-window data scheduler, instead of the previous epoch-based data scheduler, which addresses the extremely infrequent instance visiting issue in PIC and improves the effectiveness; 2) a negative sampling and weight update correction approach to reduce the training time and GPU memory consumption, which also enables application of PIC to almost unlimited training images. We hope that the PIC framework can serve as a simple baseline to facilitate future study.

cs.CV↗

arXiv2019

Coherent satellites in multi-spectral regenerative frequency microcombs

Multi-spectral frequency combs provide frontier architectures for laser standoff spectroscopy, optical clockwork, and high-capacity optical communications. Frequency microcombs, aided by their high-quality resonances and inherent third-order nonlinear susceptibility, have demonstrated remarkable impact in frequency metrology and synthesis. However, microcombs are often with limited spectral bandwidth bounded by the intrinsic second-order chromatic dispersion and the consequently low intensities at the spectral edges. To enhance the applications of frequency combs, a spectrally-broad comb generation scheme is often desired. Here we report coherent satellite clusters in multi-spectral regenerative frequency microcombs with enhanced intensities at the octave points and engineered frequency span. Beyond the conventional bandwidth of parametric oscillation, the regenerative satellites are facilitated by higher-order dispersion control allowing for multi-phase-matching in the microcavities. The frequency span of the multi-spectral regenerative satellites is deterministically controlled from 57 THz to 126 THz by pumping at C- and L- bands. We demonstrate that the regenerative satellite preserves the coherence with the central comb through the nonlinear parametric process. We further show the mirrored appearance of the satellite transition dynamics including each comb state that are simultaneously observed at the central comb. These multi-spectral regenerative satellites extend the scope of parametric-based frequency combs and provide a unique platform with wide applications.

physics.optics↗

arXiv2019

Narrow-band photonic quantum entanglement with counterpropagating domain engineering

Photonic entanglement is a crucial resource for quantum information technologies, including quantum information processing and long-distance communication. These applications call for narrow-band entanglement sources, being compatible with the photon-electron interfaces and relax the dispersion-induced entanglement-degradation in the network mediums. However, such Giga-hertz-level bandwidth is a challenge for the photonic entanglement based on conventional spontaneous parametric down-conversion. Here we introduce a state-of-art domain-engineering technology for the counterpropagating phase matching in the polarization-entanglement generation, which results in an inherent bandwidth of 7.1 GHz at telecom wavelength. This geometry also resolves the fundamental challenge for the deterministic separation of the entangled photon pairs at frequency degeneracy, so that efficient collinear interaction can be achieved in a periodically-poled KTP waveguide and results in a spectral brightness of 3.4*10^3(GHz*mW*s)^-1. The 155 ps base-to-base Hong-Ou-Mandel dip also confirms the source bandwidth, with and a high interference visibility of 97.1%. The entanglement is measured to violate the Bell inequality by up to 18.5 standard deviations, with Clauser-Horne-Shimony-Holt S-parameter of 2.720. The quantum state tomography further characterizes the entanglement, with fidelity F=95.71%. These unique features establish a cornerstone for the photonic entanglement sources development and warrant broad practical quantum information applications.

quant-ph↗

arXiv2019

Drone-based all-weather entanglement distribution

The quantum satellite is a cornerstone towards practical free-space quantum network and overcomes the photon loss over large distance. However, challenges still exist including real-time all-location coverage and multi-node construction, which may be complemented by the diversity of modern drones. Here we demonstrate the first drone-based entanglement distribution at all-weather conditions over 200 meters (test field limited), and the Clauser-Horne-Shimony-Holt S-parameter exceeds 2.49, within 35 kg take-off weight. With symmetric transmitter and receiver beam apertures and single-mode-fiber-coupling technology, such progress is ready for future quantum network with multi-node expansion. This network can be further integrated in picture-drone sizes for plug-and-play local-area coverage, or loaded onto high-altitude drones for wide-area coverage, which adds flexibility while connecting to the existing satellites and ground fiber-based quantum network.

quant-ph↗

arXiv2019

Local Relation Networks for Image Recognition

The convolution layer has been the dominant feature extractor in computer vision for years. However, the spatial aggregation in convolution is basically a pattern matching process that applies fixed filters which are inefficient at modeling visual elements with varying spatial distributions. This paper presents a new image feature extractor, called the local relation layer, that adaptively determines aggregation weights based on the compositional relationship of local pixel pairs. With this relational approach, it can composite visual elements into higher-level entities in a more efficient manner that benefits semantic inference. A network built with local relation layers, called the Local Relation Network (LR-Net), is found to provide greater modeling capacity than its counterpart built with regular convolution on large-scale recognition tasks such as ImageNet classification.

cs.CV↗

arXiv2019

Mid-infrared optical frequency comb generation from a chi-2 optical superlattice box resonator

Optical frequency combs (OFCs) at Mid-Infrared (MIR) wavelengths are essential for applications in precise spectroscopy, gas sensing and molecular fingerprinting, because of its revolutionary precision in both wavelength and frequency domain. The microresonator-based OFCs make a further step towards practical applications by including such high precision in a compact and cost-effective package. However, dispersion engineering is still a challenge for the conventional chi-3 micro-ring resonators and a MIR pump laser is required. Here we develop a different platform of a chi-2 optical superlattice box resonator to generate MIR OFC by optical parametric down conversion. With near-material-limited quality factor of 2.0*10^7, broadband MIR OFC can be generated with over 250 nm span around 2060 nm, where only a common near-infrared laser is necessary as pump. The fine teeth spacing corresponds to a measurable radio frequency beat note at 1.566 GHz, and also results in a fine spectroscopy resolution. Its linewidth is measured to be 6.1 kHz, which reveals a low comb noise that agrees well with the clean temporal waveforms. With high output power of over 370 mW, such MIR OFC is capable for long distance sensing and ranging applications.

physics.optics↗

arXiv2018

Generation of optical frequency comb in a chi-2 sheet micro optical parametric oscillator via cavity phase matching

Chi-3 micro resonators have enabled compact and portable frequency comb generation, but require sophisticated dispersion control. Here we demonstrate an alternative approach using a chi-2 sheet cavity, where the dispersion requirement is relaxed by cavity phase matching. 21.2 THz broadband comb generation is achieved with uniform line spacing of 133.0 GHz, despite a relatively large dispersion of 275.4 fs^2/mm around 1064nm. With 22.6 % high slope efficiency and 14.9 kW peak power handling, this chi-2 comb can be further stabilized for navigation, telecommunication, astronomy, and spectroscopy applications.

physics.optics↗

arXiv2017

Broadband gate-tunable THz plasmons in graphene heterostructures

Graphene, a unique two-dimensional material of carbon in a honeycomb lattice, has brought remarkable breakthroughs across the domains of electronics, mechanics, and thermal transport, driven by the quasiparticle Dirac fermions obeying a linear dispersion. Here we demonstrate a counter-pumped all-optical difference frequency process to coherently generate and control THz plasmons in atomic layer graphene with an octave tunability and high efficiency. We leverage the inherent surface asymmetry of graphene for a strong second-order nonlinear polarizability chi(2), which together with tight plasmon field confinement, enables a robust difference frequency signal at THz frequencies. The counter-pumped resonant process on graphene uniquely achieves both energy and momentum conservation. Consequently we demonstrate a dual-layer graphene heterostructure that achieves the charge- and gate-tunability of the THz plasmons over an octave, from 9.4 THz to 4.7 THz, bounded only by the pump amplifier optical bandwidth. Theoretical modeling supports our single-volt-level gate tuning and optical-bandwidth-bounded 4.7 THz phase-matching measurements, through the random phase approximation with phonon coupling, saturable absorption, and below the Landau damping, to predict and understand the graphene carrier plasmon physics.

physics.optics↗