Modal Loss Research Articles

Low-loss weakly coupled four-mode hollow-core antiresonant fiber based on centrosymmetric nested structure

Abstract This paper designs a novel centrosymmetric double-layer nested antiresonant fiber (CDNAF) with few-mode transmission characteristics. The cladding structure of the CDNAF incorporates glass plates and multilayer capillaries, effectively mitigating the coupling between various core modes and cladding modes. Simulation results show that the CDNAF can support four modes of low-loss transmission from LP01 to LP02 at 1550 nm. The confinement loss (CL) of the four modes is less than 0.008 dB/km, 0.07 dB/km, 0.42 dB/km, and 4.8 dB/km, respectively, and the CDNAF has a sizeable differential group delay (DGD) and weak bending sensitivity. The loss of high-order modes is much greater than that of the other four transmission modes, ensuring high-purity transmission of the four modes. In the wavelength range of 1200–1700 nm, the CDNAF ensures few-mode characteristics with at least two low-loss transmission modes. Furthermore, the effective refractive index difference (Δneff) between adjacent modes is much greater than 10^(-4), eliminating or significantly reducing the need for multiple-input multiple-output digital signal processing (MIMO-DSP) techniques in optical communication systems. Additionally, the results demonstrate that CDNAF can transmit three modes at an ultimate bending radius of 5 cm and can transmit four modes like a straight fiber at a bending radius of 18 cm, which demonstrates that CDNAF has excellent bending performance. These excellent characteristics give the CDNAF great potential for application in large-capacity optical communication systems.

Evaluating and Enhancing Face Anti-Spoofing Algorithms for Light Makeup: A General Detection Approach

Makeup modifies facial textures and colors, impacting the precision of face anti-spoofing systems. Many individuals opt for light makeup in their daily lives, which generally does not hinder face identity recognition. However, current research in face anti-spoofing often neglects the influence of light makeup on facial feature recognition, notably the absence of publicly accessible datasets featuring light makeup faces. If these instances are incorrectly flagged as fraudulent by face anti-spoofing systems, it could lead to user inconvenience. In response, we develop a face anti-spoofing database that includes light makeup faces and establishes a criterion for determining light makeup to select appropriate data. Building on this foundation, we assess multiple established face anti-spoofing algorithms using the newly created database. Our findings reveal that the majority of these algorithms experience a decrease in performance when faced with light makeup faces. Consequently, this paper introduces a general face anti-spoofing algorithm specifically designed for light makeup faces, which includes a makeup augmentation module, a batch channel normalization module, a backbone network updated via the Exponential Moving Average (EMA) method, an asymmetric virtual triplet loss module, and a nearest neighbor supervised contrastive module. The experimental outcomes confirm that the proposed algorithm exhibits superior detection capabilities when handling light makeup faces.

The transformation of sensory to perceptual braille letter representations in the visually deprived brain.

Experience-based plasticity of the human cortex mediates the influence of individual experience on cognition and behavior. The complete loss of a sensory modality is among the most extreme such experiences. Investigating such a selective, yet extreme change in experience allows for the characterization of experience-based plasticity at its boundaries. Here, we investigated information processing in individuals who lost vision at birth or early in life by probing the processing of braille letter information. We characterized the transformation of braille letter information from sensory representations depending on the reading hand to perceptual representations that are independent of the reading hand. Using a multivariate analysis framework in combination with functional magnetic resonance imaging (fMRI), electroencephalography (EEG), and behavioral assessment, we tracked cortical braille representations in space and time, and probed their behavioral relevance. We located sensory representations in tactile processing areas and perceptual representations in sighted reading areas, with the lateral occipital complex as a connecting 'hinge' region. This elucidates the plasticity of the visually deprived brain in terms of information processing. Regarding information processing in time, we found that sensory representations emerge before perceptual representations. This indicates that even extreme cases of brain plasticity adhere to a common temporal scheme in the progression from sensory to perceptual transformations. Ascertaining behavioral relevance through perceived similarity ratings, we found that perceptual representations in sighted reading areas, but not sensory representations in tactile processing areas are suitably formatted to guide behavior. Together, our results reveal a nuanced picture of both the potentials and limits of experience-dependent plasticity in the visually deprived brain.

The transformation of sensory to perceptual braille letter representations in the visually deprived brain

Experience-based plasticity of the human cortex mediates the influence of individual experience on cognition and behavior. The complete loss of a sensory modality is among the most extreme such experiences. Investigating such a selective, yet extreme change in experience allows for the characterization of experience-based plasticity at its boundaries. Here, we investigated information processing in individuals who lost vision at birth or early in life by probing the processing of braille letter information. We characterized the transformation of braille letter information from sensory representations depending on the reading hand to perceptual representations that are independent of the reading hand. Using a multivariate analysis framework in combination with functional magnetic resonance imaging (fMRI), electroencephalography (EEG), and behavioral assessment, we tracked cortical braille representations in space and time, and probed their behavioral relevance. We located sensory representations in tactile processing areas and perceptual representations in sighted reading areas, with the lateral occipital complex as a connecting ‘hinge’ region. This elucidates the plasticity of the visually deprived brain in terms of information processing. Regarding information processing in time, we found that sensory representations emerge before perceptual representations. This indicates that even extreme cases of brain plasticity adhere to a common temporal scheme in the progression from sensory to perceptual transformations. Ascertaining behavioral relevance through perceived similarity ratings, we found that perceptual representations in sighted reading areas, but not sensory representations in tactile processing areas are suitably formatted to guide behavior. Together, our results reveal a nuanced picture of both the potentials and limits of experience-dependent plasticity in the visually deprived brain.

An adaptive material-field series-expansion method for structural topology optimization

The material-field series-expansion (MFSE) method reduces the dimension of design variables in density-based topology optimization and avoids the checkerboard patterns. However, the constant correlation function of the MFSE method only accounts for the spatial dependency between material field points. Accordingly, the expanded material field using the constant correlation function has an obscure topology, which increases the iterations and computation burden of the MFSE method. This paper proposes an adaptive correlation function that considers the spatial dependency and the values of material field points simultaneously. It describes the correlation between material field points more accurately. As a result, the expanded material field based on the adaptive correlation function has a clearer topology than the constant correlation function. Moreover, the adaptive material-field series-expansion (AMFSE) method with dual-loop optimization is introduced leveraging the adaptive correlation function. The outer loop updates the adaptive correlation functions and expanded material fields. The inner loop searches for the optimal solution of the current topology optimization model based on the expanded material field. Finally, several examples of static compliance and dynamic modal loss factor topology optimization validate the effectiveness and applicability of the AMFSE method. Results show that the AMFSE method converges faster with fewer iterations than the MFSE method, despite it increasing the computing time of matrix decomposition in the outer loops. Therefore, the AMFSE method is suitable for topology optimization with complex responses such as modal loss factors to decrease the iterations and computation burden simultaneously.

Improved fractional-order gradient descent method based on multilayer perceptron

The fractional-order gradient descent (FOGD) method has been employed by numerous scholars in Artificial Neural Networks (ANN), with its superior performance validated both theoretically and experimentally. However, current FOGD methods only apply fractional-order differentiation to the loss function. The application of FOGD based on Autograd to hidden layers leverages the characteristics of fractional-order differentiation, significantly enhancing its flexibility. Moreover, the implementation of FOGD in the hidden layers serves as a necessary foundation for establishing a family of fractional-order deep learning optimizers, facilitating the widespread application of FOGD in deep learning. This paper proposes an improved fractional-order gradient descent (IFOGD) method based on Multilayer Perceptron (MLP). Firstly, a fractional matrix differentiation algorithm and its fractional matrix differentiation solver is proposed based on MLP, ensuring that IFOGD can be applied within the hidden layers. Subsequently, we overcome the issue of incorrect backpropagation direction caused by the absolute value symbol, ensuring that the IFOGD method does not cause divergence in the value of the loss function. Thirdly, fractional-order Autograd (FOAutograd) is proposed based on PyTorch by reconstructing Linear layer and Mean Squared Error Loss module. By combining FOAutograd with first-order adaptive deep learning optimizers, parameter matrices in each layer of ANN can be updated using fractional-order gradients. Finally, we compare and analyze the performance of IFOGD with other methods in simulation experiments and time series prediction tasks. The experimental results demonstrate that the IFOGD method exhibits performances.

Spatial mapping of activity changes across sensory areas following visual deprivation in adults.

Loss of a sensory modality triggers global adaptation across brain areas, allowing the remaining senses to guide behavior more effectively. There are specific synaptic and circuit plasticity observed across many sensory areas, which suggests potential widespread changes in activity. Here we used a cFosTRAP2 mouse line to drive tdTomato (tdT) expression in active cells to spatially map the extent of activity changes in various sensory areas in adult mice of both sexes following two modes of visual deprivation. We found that in the primary visual cortex (V1) both dark exposure (DE) and enucleation (EN) caused an initial loss of active cells followed by a partial rebound, which occurred relatively more in the superficial layers. A similar pattern was observed in the secondary visual cortex, especially in the lateral areas (V2L). The spared primary sensory cortices adapted distinctly. In the primary somatosensory barrel cortex (S1BF) there was a change in the density of active cells dependent on the duration and the mode of visual deprivation. In the primary auditory cortex (A1), there was a relative reduction in the density of active cells in the superficial layers without a significant change in the overall density. There were minimal changes in the active cell density in the secondary cortices of the spared senses and the multisensory retrosplenial cortex (RSP). Our results are consistent with cross-modal recruitment of the deprived visual cortex and compensatory plasticity in the spared primary sensory cortices that can support enhanced processing and refinement of the spared senses.Significance Statement The neural basis for improved processing of the spared senses after losing vision is largely dependent on plasticity mechanisms driven by neural activity. Using a transgenic mouse model to label active neurons in a temporally controlled manner, this study reveals the spatial pattern of widespread activity changes across brain areas following visual deprivation in adult mice. The findings suggest that the deprived visual cortex undergoes an initial reduction in activity followed by a rebound suggestive of cross-modal recruitment. Activity changes in the spared primary sensory cortices conform to compensatory plasticity, which could support improved processing of the spared senses. The results highlight the presence of large-scale plasticity in the adult brain to compensate for loss of vision.

ReMamba: a hybrid CNN-Mamba aggregation network for visible-infrared person re-identification.

Visible-Infrared Person Re-identification (VI-ReID) has been consistently challenged by the significant intra-class variations and cross-modality differences between different cameras. Therefore, the key lies in how to extract discriminative modality-shared features. Existing VI-ReID methods based on Convolutional Neural Networks (CNN) and Vision Transformers (ViT) have shortcomings in capturing global features and controlling computational complexity, respectively. To tackle these challenges, we propose a hybrid network framework called ReMamba. Specifically, we first use a CNN as the backbone network to extract multi-level features. Then, we introduce the Visual State Space (VSS) model, which is responsible for integrating the local features output by the CNN from lower to higher levels. These local features serve as a complement to global information and thereby enhancing the local details clarity of the global features. Considering the potential redundancy and semantic differences between local and global features, we design an adaptive feature aggregation module that automatically filters and effectively aggregates both types of features, incorporating an auxiliary aggregation loss to optimize the aggregation process. Furthermore, to better constrain cross-modality features and intra-modal features, we design a modal consistency identity constraint loss to alleviate cross-modality differences and extract modality-shared information.Extensive experiments conducted on the SYSU-MM01, RegDB, and LLCM datasets demonstrate that our proposed ReMamba outperforms state-of-the-art VI-ReID methods.

Mask-Space Optimized Transformer for Semantic Segmentation of Lithium Battery Surface Defect Images

The segmentation of surface defects in lithium batteries is crucial for enhancing the overall quality of the production process. However, the severe foreground–background imbalance in surface images of lithium batteries, along with the irregular shapes and random distribution of foreground regions, poses significant challenges for defect segmentation. Based on these observations, this paper focuses on the separation of foreground and background in surface defect images of lithium batteries and proposes a novel Mask Space Optimization Transformer (MSOFormer) for semantic segmentation of these images. Specifically, the Mask Boundary Loss (MBL) module in our model provides more efficient supervision during training to enhance the accuracy of the mask computation within the mask attention mechanism, thereby improving the model’s performance in separating foreground and background. Additionally, the Dynamic Spatial Query (DSQ) module allocates spatial information of the image to each query, enhancing the model’s sensitivity to the positions of small foreground targets in various scenes. The Efficient Pixel Decoder (EPD) ensures deformable receptive fields for irregularly shaped foregrounds while further improving the model’s performance and efficiency. Experimental results demonstrate that our method outperforms other state-of-the-art methods in terms of mean Intersection over Union (mIoU). Specifically, our approach achieves an mIoU of 84.18% on the lithium battery surface defect test set and 85.53% and 87.05% mIoUs on two publicly available defect test sets with similar defect characteristics to lithium batteries.

Exceptional points induced by unidirectional coupling in electronic circuits

Exceptional points in non-Hermitian systems have attracted considerable attention due to their novel applications in several fields such as optics, electronics, and mechanics. Typically, exceptional points are constructed through gain and loss modulation or dissipative coupling within the framework of parity-time symmetry or anti-parity-time symmetry. Recent demonstration of unidirectional coupling in optical resonators to create exceptional points has offered an alternative approach. This study extends this concept to electronic circuits, examining exceptional points that emerge in unidirectionally coupled LC circuits. We show that this circuit undergoes resonance frequency splitting that exhibits either linear or square-root scaling with the strength of the applied perturbation. We further explore the circuit’s scattering properties when connected to an input-output channel and demonstrate both theoretically and experimentally the splitting of transmission dips or peaks when a perturbation is applied—highlighting the potential for building sensors with enhanced sensitivity. This work not only deepens the understanding of exceptional points in electronic circuits but also encourages the exploration and application of non-Hermiticity in electronics.

Self-mixing thinly sliced ruby laser for laser Doppler velocimetry with high optical sensitivity

In self-mixing laser Doppler velocimetry (LDV), the motion of a moving target is observed by using intensity-modulated laser light detected by a simple photodetector. Here, the self-mixing laser output modulation takes place, reflecting the pronounced effective loss modulation index, which is proportional to the fluorescence-to-photon lifetime ratio. The fluorescence lifetime of a ruby laser is extremely long, so if a ruby crystal can be used as a laser light source for a self-mixing LDV system, high-sensitivity LDV measurements can be performed with it. We describe a method for velocimetry of moving targets using self-mixing LDV in which a CW oscillating ruby laser is the light source. The oscillation mechanism of the thin-slice ruby laser with a large fluorescence-to-photon lifetime ratio, which is suitable for LDV measurements, is clarified and the results of highly sensitive LDV measurements are presented, featuring nonlinear dynamics observed associated with the self-mixing velocimetry experiment. The measurement accuracy is clarified by measuring the rotating disc with various conditions using self-mixing LDV.

Broadband high birefringence and single-polarization hollow-core anti-resonant fibers with an elliptical-like core

Due to the light-guiding mechanism of hollow-core anti-resonant fibers (HC-ARFs), the low overlap between core modes and anti-resonant layers poses a challenge in achieving high birefringence. To address this issue, we propose three high-birefringence HC-ARFs with an elliptical core and varying cladding compositions and conduct a detailed optimization and analysis for structure (c). Simulation results demonstrate that structure (a) and structure (b) provide high birefringence, broadband, and low-loss transmission properties. Specifically, structure (a) achieves a birefringence greater than 1.3×10−4 over a 680 nm wavelength range, and structure (b) achieves a birefringence greater than 1.6×10−4 over a 580 nm wavelength range, both maintaining fundamental mode (FM) loss below 1 dB/m. Structure (c) enhances birefringence by an order of magnitude and offers excellent single-polarization properties by introducing high-refractive-index material. At 1.55μm, structure (c) achieves a birefringence of 0.98×10−3, with a low FM loss for x-polarization of 0.01 dB/m and a polarization extinction ratio (PER) of 57450. Additionally, structure (c) exhibits low bend loss in the x-direction, with only 0.06 dB/m for a bend radius of 6 cm.

160 kW beam commissioning for the Rapid Cycling Synchrotron of the China Spallation Neutron Source

For the China Spallation Neutron Source (CSNS), the rapid cycling synchrotron (RCS)accumulates and accelerates the injection beam to the design energy of 1.6 GeV and then extractsthe high energy beam to the target. In this paper, the 160 kW beam power commissioning forthe CSNS RCS has been comprehensively studied, including: increase in the pulse width ofthe injection beam to optimize the transverse painting, commissioning of the two dualharmonic radio frequency (RF) cavities, optimization of the transverse and momentum collimators,longitudinal dynamic optimization, beam instability suppression,global closed orbit correction, local closed orbit correction, optimization of the RCS occasionallarge beam loss, optimization of the extraction mode and extraction occasional large beam loss,and so on.In order to meet the requirements of beam power increase and stable operation of the accelerator,the RCS beam losses from different sources are studied and optimized. With the aid of weeklyradiation dose measurement, the hot spots of the RCS are studied in depth to explore the causes.

The Simulation of Mode Control for a Photonic Lantern Adaptive Amplifier.

A photonic lantern is a low-loss device that connects a single multimode waveguide to multiple single-mode waveguides and can enhance the beam quality of a fiber laser by adaptively controlling the optical parameters (amplitude, phase, polarization) at the input. In this work, we combined the gains and losses of individual modes within the fiber amplifier and introduced a mode content parameter at the amplifier's output as an evaluation function to simulate mode control effects. Mode competition within the gain fiber can degrade the control effect of the fundamental mode and lead to it taking a longer time for the control to converge. Optimal parameters, such as the gain fiber length and pumping method, were identified to improve control effectiveness. Specifically, an optimal gain fiber length of 8 m was determined, and backward pumping was found to achieve higher pumping efficiency and better control results. The system demonstrated significant power amplification potential and could stabilize mode control under different pumping powers ranging from 50 W to 5 kW. In conclusion, our research demonstrates that an adaptive fiber amplifier based on a photonic lantern can achieve a stable, high-power, large-mode-field, near-fundamental-mode output from the gain fiber. Although mode competition within the gain fiber can degrade the control effect of the fundamental mode and cause the control to take a longer time to converge, these aspects should be further studied to improve the control's effectiveness. These findings contribute to the development of advanced simulation models that guide high-power mode control experiments and deepen our understanding of physical processes in science and technology.

A zero-shot industrial process fault diagnosis method based on domain-shift constraints

BackgroundFault diagnosis is crucial for industrial maintenance, but existing supervised methods rely on extensive data, which is often difficult to collect. The challenge of gathering comprehensive fault samples limits the performance of traditional fault diagnosis methods. MethodIn this paper, we propose a fault diagnosis method named ZSIDM-OC to address the zero-shot problem in industrial processes, specifically concerning the domain shift issue. This novel framework includes three key modules: the Hierarchical Global-Local Feature Integration Module for capturing both global and local features of the fault data; the Prototype-Based Discriminative Loss Module, which reduces feature redundancy and enhances the model's ability to recognize unknown fault classes; and the Bidirectional Consistency Enforcement Module ensuring consistent data distribution in both low-dimensional and high-dimensional spaces, thereby reducing domain shift. Significant FindingsOur analysis indicates that the domain shift problem is inevitable in a zero-shot setting and significantly affects the performance of existing methods. Experimental results demonstrate that under zero-shot conditions, ZSIDM-OC offers significant advantages on both the Energy Storage Plant dataset and the Tennessee Eastman dataset. This method effectively mitigates the challenges posed by domain shift and limited fault sample availability, showcasing its potential to improve fault diagnosis in industrial processes.

Weakly Supervised Underwater Object Real-time Detection Based on High-resolution Attention Class Activation Mapping and Category Hierarchy

Recently, deep learning-based underwater object detection technology has achieved remarkable success. However, the accuracy and completeness of dataset instance annotation are crucial for its success. The quality of underwater images is low, severe objects clustering, and occlusion, acquiring object's annotations demands substantial time and labor costs, while mis annotation and missed annotation can also degrade model performance and limit their application in practical scenarios. To address this issue, this paper presents a novel weakly supervised underwater object real-time detection method, which is divided into two subtasks: weakly supervised object localization and real-time object detection. In the weakly supervised object localization task, we design a novel category hierarchy structure network that integrates the high-resolution attention-class activation mapping algorithm to obtain high-quality object class activation maps, weaken background interference, and obtain more complete object regions. The parameterized spatial loss module is devised to enable the model to escape from local optimal solutions, thus accurately and efficiently obtaining object pseudo-detection annotation boxes. For the real-time object detection task, the single-stage detector YOLOv7 is selected as the basic detection model, and an object perception loss function is designed based on the class activation map to jointly supervise the training process. A method for filtering noisy pseudo-supervision information is proposed to enhance the pseudo-supervision information involved in training. Ablation experiments and multi-method comparison experiments were conducted on the URPC and RUOD datasets, and the results verify the effectiveness of the proposed strategy, and our model exhibits significant advantages in detection performance and detection efficiency compared to current mainstream and advanced models.

Multi-plane light conversion (MPLC) LP mode multiplexer based on grayscale maskless lithography.

Multi-plane light conversion (MPLC) is a technique that uses multiple phase plates to modulate a light beam step-by-step. This technique has attracted widespread attention in the field of mode-division multiplexing (MDM) communications due to its high flexibility. MPLC device requires precisely controlled fabrication accuracy in experiments, but conventional multi-etching processes will accumulate alignment errors. Here, the fabrication of the MPLC device using maskless grayscale lithography was proposed, which requires only a single-exposure process. Through single-exposure lithography, the continuous phase of the digital mask ranging from 0 to 2π on MPLC is discretized into 128 steps. The digital masks of the MPLC with more steps of phase can reduce the insertion loss and mode crosstalk of LP modes. By using the fabricated MPLC, we experimentally demonstrate the MDM of LP01, LP11a, LP11b, and LP21 modes with mode crosstalk less than -22 dB, and the insertion loss less than 4 dB. In high-speed optical communications, each LP mode carries a 10 Gbit/s on-off keying (OOK) signals, and the experimentally measured bit error rates (BER) curves power penalty is less than -7 dB. The experiment demonstrated that maskless grayscale lithography can efficiently and accurately fabricate MPLC mode multiplexers.

Integration of Electrical Properties and Polarization Loss Modulation on Atomic Fe–N-RGO for Boosting Electromagnetic Wave Absorption

HighlightsSingle-atom Fe–N4 sites embedded into graphene were successfully synthesized to exert the dielectric properties of graphene.The absorption mechanisms of metal-nitrogen doping reduced graphene oxide mainly include enhanced dipole polarization, interface polarization, conduction loss and defect-induced polarization.Excellent reflection loss of − 74.05 dB (2.0 mm) and broad effective absorption bandwidth of 7.05 GHz (1.89 mm, with filler loading only 1 wt%) were obtained.

Dual-mode 1 × 2 optical switch with simultaneous modulation based on inverse design devices

Abstract Mode division multiplexing (MDM) technology, based on the parallelism inherent in mode dimensions, provides an advancement in enhancing on-chip optical communication channel capacity. In MDM communication systems, the routing and switching of optical signals are of essential importance. However, conventional multimode optical switches typically follow the demultiplexing-processing-multiplexing technological route, leading to an unavoidable increase in device size. In scenarios where multiple modes need to be routed synchronously, the implementation of simultaneous modulation optical switches offers a more efficient and feasible solution. Here, we propose two 1×2 dual-mode optical switches with simultaneous modulation on a silicon-on-insulator (SOI) platform in the 1525-1565 nm wavelength range, utilizing two optical phase modulation techniques: mode transformation and waveguide widening. Simultaneously, we employ inverse design methodologies based on the adjoint variable method (AVM) and level-set method to create the compact single-connected devices, which are compatible with CMOS fabrication processes. The experimental results show that the insertion losses for both TE0 and TE1 modes are less than 1.6 dB (2.5 dB), with the worst modal crosstalk at most -13.5 dB (-12.7 dB) for the switch based on mode transformation (waveguide widening) strategy at the wavelength of 1550nm. The extinction ratio (ER) of the two proposed optical switches exceeds 25 dB at the same wavelength. Furthermore, the switches exhibit a 10%-90% rise time of 15.2 μs and a 90%-10% fall time of 19.5 μs at 1550 nm, indicating the switching speed can be up to kilohertz (kHz). Our proposed 1×2 optical switches hold potential as a fundamental unit for optical signal switching in high-integration multimode optical communication systems.

Design of a switchable triple-waveguide-based TE-converted polarization beam splitter based on lithium niobate

The lithium-niobate-on-insulator (LNOI) platform has been the new force to drive the developments of integrated photonics, where efficient polarization management and mode conversion are the fundamental issues to be solved. In this work, we proposed a switchable TE-converted polarization beam splitter (PBS) based on optical phase change material (PCM) and a triple-waveguide in simulation. To optimize and analyze the proposed design, the 3D finite difference time domain (3D-FDTD) method is applied in this work. The simulated results indicate that when PCM is in the amorphous state, the device can effectively separate the TM and TE modes, achieving a low insertion loss (<0.5dB) for both modes at 1550 nm. While the PCM is in the crystalline state, the proposed device can realize the efficient conversion from the input TE0 mode to output TE1 mode, which obtains a mode conversion efficiency of 99.5%, crosstalk of −26dB, and insertion loss of 0.5 dB at 1550 nm. We hope such a device can make compact integration and capacity improvement for the integrated photonics in LNOI.