arXiv subjects

Xintong Zhang

Publications and source records attributed to Xintong Zhang.

25 records · Page 2Linked to original sources

arXiv2024

CLOVA: A Closed-Loop Visual Assistant with Tool Usage and Update

Utilizing large language models (LLMs) to compose off-the-shelf visual tools represents a promising avenue of research for developing robust visual assistants capable of addressing diverse visual tasks. However, these methods often overlook the potential for continual learning, typically by freezing the utilized tools, thus limiting their adaptation to environments requiring new knowledge. To tackle this challenge, we propose CLOVA, a Closed-Loop Visual Assistant, which operates within a framework encompassing inference, reflection, and learning phases. During the inference phase, LLMs generate programs and execute corresponding tools to complete assigned tasks. In the reflection phase, a multimodal global-local reflection scheme analyzes human feedback to determine which tools require updating. Lastly, the learning phase employs three flexible approaches to automatically gather training data and introduces a novel prompt tuning scheme to update the tools, allowing CLOVA to efficiently acquire new knowledge. Experimental findings demonstrate that CLOVA surpasses existing tool-usage methods by 5% in visual question answering and multiple-image reasoning, by 10% in knowledge tagging, and by 20% in image editing. These results underscore the significance of the continual learning capability in general visual assistants.

cs.CV↗

arXiv2024

Four-fold non-Hermitian phase transitions and non-reciprocal coupled resonator optical waveguides

Non-Hermitian systems can exhibit extraordinary sensitivity to boundary conditions. Given that topological boundary modes and non-Hermitian skin effects can either coexist or individually appear in non-Hermitian systems, it is of great value to present a comprehensive non-Hermitian phase diagram, for further flexible control in realistic non-Hermitian systems. Here, we reveal four-fold non-Hermitian phase transitions at a mathematically level, where phase I exhibits only topological boundary modes, phase II displays both topological boundary modes and skin modes, phase III exhibits only skin modes, and phase IV cannot manifest any boundary modes. By deriving non-Hermitian winding numbers, the existence or non-existence condition of topological boundary modes are analytically expressed, consistent with the numerical results obtained through the iterative Green's function method. Combining with the study on non-Hermitian skin effects, we rigorously establish the four-fold phase diagram. We also design an array of coupled resonator optical waveguides. The introduction of non-Hermiticity in the photonic structure induces a phenomenon similar to band inversion in topological insulators, indicating the presence of topological boundary modes in the photonic bands.

quant-ph↗

arXiv2023

Thermal Conductivity Measurement Using Modulated Photothermal Radiometry for Nitrate and Chloride Molten Salts

Molten salts are being used or explored for thermal energy storage and conversion systems in concentrating solar power and nuclear power plants. Thermal conductivity of molten salts is an important thermophysical property dictating the performance and cost of these systems, but its accurate measurement has been challenging, as evidenced by wide scattering of existing data in literature. The corrosive and conducting nature of these fluids also leads to time consuming sample preparation processes of many contact-based measurements. Here, we report the measurement of thermal conductivity of molten salts using a modulated photothermal radiometry (MPR) technique, which is a laser-based, non-contact, frequency-domain method adopted for molten salts for the first time. By unitizing the advantages of front side sensing of frequency-domain measurements and the vertical holder orientation, the technique can minimize the natural convection and salt creeping effects, thus yielding accurate molten salt thermal conductivity. The MPR technique is first calibrated using standard molten materials including paraffin wax and sulfur. It is then applied on measuring pure nitrate salts ($NaNO_3$ and $KNO_3$), solar salt ($NaNO_3-KNO_3$ mixture), and chloride salt ($NaCl-KCl-MgCl_2$). The measurement results are compared with data from literature, especially those obtained from laser flash analysis (LFA). Our results demonstrate that the MPR is a convenient and reliable technique of measuring thermal conductivity of molten salts. Accurate thermal conductivity data of molten salts will be valuable in developing the next-generation high-temperature thermal energy storage and conversion systems.

physics.chem-ph↗

arXiv2023

Enhance Reasoning Ability of Visual-Language Models via Large Language Models

Pre-trained visual language models (VLM) have shown excellent performance in image caption tasks. However, it sometimes shows insufficient reasoning ability. In contrast, large language models (LLMs) emerge with powerful reasoning capabilities. Therefore, we propose a method called TReE, which transfers the reasoning ability of a large language model to a visual language model in zero-shot scenarios. TReE contains three stages: observation, thinking, and re-thinking. Observation stage indicates that VLM obtains the overall information of the relative image. Thinking stage combines the image information and task description as the prompt of the LLM, inference with the rationals. Re-Thinking stage learns from rationale and then inference the final result through VLM.

cs.CL↗

arXiv2023

Solvable non-Hermitian skin effects and real-space exceptional points: Non-Hermitian generalized Bloch theorem

Non-Hermitian systems can exhibit extraordinary boundary behaviors, known as the non-Hermitian skin effects, where all the eigenstates are localized exponentially at one side of lattice model. To give a full understanding and control of non-Hermitian skin effects, we have developed the non-Hermitian generalized Bloch theorem to provide the analytical expression for all solvable eigenvalues and eigenstates, in which translation symmetry is broken due to the open boundary condition. By introducing the Vieta's theorem for any polynomial equation with arbitrary degree, our approach is widely applicable for one-dimensional non-Hermitian tight-binding models. With the non-Hermitian generalized Bloch theorem, we can analyze the condition of existence or non-existence of the non-Hermitian skin effects at a mathematically rigorous level. Additionally, the non-Hermitian generalized Bloch theorem allows us to explore the real-space exceptional points. We also establish the connection between our approach and the generalized Brillouin zone method. To illustrate our main results, we examine two concrete examples including the Su-Schrieffer-Heeger chain model with long-range couplings, and the ladder model with non-reciprocal interaction. Our non-Hermitian generalized Bloch theorem provides an efficient way to analytically study various non-Hermitian phenomena in more general cases.

quant-ph↗

arXiv2021

Thermal Conductivity Modeling of Monodispersed Microspheres using Discrete Element Method

Particle beds are widely used in various systems and processes, such as particle heat exchangers, granular flow reactors, and additive manufacturing. Accurate modeling of thermal conductivity of particle beds and understanding of their heat transfer mechanisms are important. However, previous models were based on a simple cubic packing of particles which could not accurately represent the actual heat transfer processes under certain conditions. Here, we examine the effect of the packing structure on thermal conductivity of particle beds. We use monodispersed silica microspheres with average article sizes ranging from 23 to 330 um as a model material. We employ a transient hot-wire technique to measure the thermal conductivity of the particle beds with packing density of 43 to 57% within a temperature range of room temperature to 500 deg. C and under N2 gaseous pressures of 20 to 760 Torr. We then use a discrete element method (DEM) to obtain the realistic packing structure of the particles, which is then fed into a finite-element model (FEM) to calculate the thermal conductivity, with the consideration of solid conduction, gas conduction, and radiation heat transfer. Our results show that the thermal conductivity model based on the more realistic random packing structure derived from the DEM shows better agreement with the experimental data compared to that based on the simple cubic packing structure. The combined DEM and FEM methodology can serve as a useful tool to predict effective thermal conductivity of particle beds and to quantify different heat transfer mechanisms under various conditions.

cond-mat.mtrl-sci↗

arXiv2016

Interlayer Conductance of Graphene with Multiple Transfer Process

Electrical properties of multi-layer graphene are subject to variations due to random interlayer alignments. In this work we reported graphene interlayer conductance without special layer aligning. Ohmic contacts between two graphene layers are observed with resistance variations of more than one order. With Raman spectroscopy we identify that the lattice angle between twisted graphene layers is the key variation source. The angular dependence and temperature dependence of the interlayer conductance suggest that a phonon assistant tunneling mechanism is valid for the interlayer transport of graphene prepared by multiple transfer process. We finally derive that the multi-layer graphene resistance shows an exponential-like distribution due to the random interlayer misalignments.

cond-mat.mes-hall↗