arXiv subjects

Xiaodong Guo

Publications and source records attributed to Xiaodong Guo.

16 recordsLinked to original sources

arXiv2026

Density-wave order enhances the phonon thermal Hall effect in a trilayer nickelate

Ruddlesden--Popper nickelates have emerged as a promising platform for high-temperature superconductivity, yet the role of lattice degrees of freedom in their correlated normal state remains largely unexplored. Here, we report the observation of a finite phonon thermal Hall effect in the trilayer nickelate La$_4$Ni$_3$O$_{10}$ at ambient pressure. Remarkably, the thermal Hall response is strongly enhanced below the density-wave transition at $T^*\approx140$ K, exhibiting two distinct plateaus in the thermal Hall resistivity. The characteristic energy scale extracted from the thermal Hall response ($\sim4.1$ meV) closely matches the magnon--phonon crossing span energy ($\sim3.2$ meV), pointing to magnon--phonon hybridization as the primary mechanism enhancing the thermal Hall effect. These results provide new insight into the interplay between lattice and spin excitations in nickelates, with implications for understanding both their superconductivity and the multiple possible origins of insulating thermal Hall signals.

cond-mat.str-el↗

arXiv2026

TUNI: Unifying Pre-training and Fine-tuning with Modality-Aware Mutual Learning and Rectification for RGB-T Semantic Segmentation

RGB-thermal (RGB-T) semantic segmentation improves the environmental perception of autonomous platforms in challenging conditions. Prevailing RGB-T segmentation frameworks suffer from suboptimal multi-modal feature extraction and fusion, unbalanced modality dependency, and inadequate utilization of thermal information. To address these challenges, we propose TUNI, a unified pre-training and fine-tuning framework for efficient and real-time RGB-T semantic segmentation. It pre-trains an RGB-T encoder that incorporates an RGB-T local module that selectively emphasizes salient consistent and distinct local features across modalities, thereby integrating cross-modal feature extraction and fusion in a unified manner. To alleviate the modality bias issue during RGB-T pre-training, modality-inverted contrastive mutual learning is introduced to enable knowledge exchange between two RGB-dominated and thermal-dominated encoders. In the fine-tuning phase, modality rectification learning fully exploits residual thermal information by focusing on correct yet divergent prediction regions between two modality-specific decoders. We further develop three TUNI variants, covering lightweight, balanced, and high-performance requirements. Extensive experiments on five RGB-T semantic segmentation datasets demonstrate that TUNI achieves superior accuracy, generalization, and compactness compared with 15 state-of-the-art models. The code is available at https://github.com/xiaodonguo/TUNI-v2.

cs.CV↗

arXiv2026

Multiband transport hierarchy and large Nernst effect in EuAuBi: Establishing a Nernst scaling for asymmetric multiband systems

In correlated materials, coexisting pockets of vastly different carrier densities raise two fundamental questions: which pocket governs the various transport coefficients, and does the conventional Nernst scaling $ν/T \propto μ/E_F$, originally derived for single-band systems, still hold? We address both questions in the polar semimetal EuAuBi, where a dilute electron pocket ($n_e \sim 10^{16}~\mathrm{cm}^{-3}$) coexists with a dense hole pocket ($n_h \sim 10^{21}~\mathrm{cm}^{-3}$). We find a clear hierarchy: the hole pocket dominates the longitudinal resistivity; the Hall effect crosses from electron- to hole-dominance with increasing field; the Seebeck coefficient is dominated by the electron pocket at low temperature and by both pockets at high temperature. Remarkably, the Nernst effect is governed entirely by the ultrahigh-mobility electron pocket, yielding a large low-field signal of $\sim 5~μ\mathrm{V/K}$ near 1~T at 202~K, comparable to anomalous Nernst signals in magnetic Weyl semimetals. By analyzing the two-band thermoelectric conductivity, we show that the Nernst coefficient follows a scaling $ν/T \propto μ_e/{E_{F, tot}}$. This scaling originates from a compensation between the electron-to-hole conductivity ratio and the Fermi-energy ratio, establishing that the large Nernst effect is a semiclassical multiband phenomenon rather than a topological Berry-curvature contribution. This understanding advances the thermoelectric transport physics of multiband electronic systems and offers a guiding principle for low-field transverse thermoelectric design.

cond-mat.str-el↗

arXiv2026

Pulsed thermal annealing enables switching of chiral antiferromagnetic order with a sub-millitesla field in Mn$_3$Sn

The manipulation of antiferromagnetic (AFM) order is a central theme in modern spintronics. In this work, we achieve reliable switching of the chiral AFM state in the Weyl antiferromagnet Mn$_3$Sn using a heat pulse combined with a very small magnetic field as small as 0.1 mT. By systematically measuring the anomalous Hall effect (AHE) in high-quality single crystals, we show that the field needed for switching decreases as the temperature approaches the Néel temperature $T_N$, and vanishes at $T_N$. Pulsed thermal annealing above $T_N$ followed by cooling in a tiny external field enables full and reproducible switching of the magnetic octupole order. Our results show that thermal softening (heating above $T_N$ to temporarily remove the magnetic anisotropy) is a key step that lowers the energy barrier to nearly zero. This allows an extremely weak directional field (like the effective field from spin-orbit torque in thin-film devices) to set the final magnetic state during cooling. We also provide a simple model to estimate the temperature rise in nanoscale devices under current pulses, giving practical guidance for device design. This work highlights that thermal effects are not a side issue but an important partner to spin torques, and suggests that future work should take both into account.

cond-mat.mtrl-sci↗

arXiv2026

Hydrodynamics of the viscous electron fluid in cadmium

Thanks to electron-electron ($e$-$e$) collisions conserving momentum, metallic electron fluids are viscous. Yet, this viscosity is rarely detectable in bulk transport. Here, we report on the canonical realization of the Gurzhi effect in an elemental three-dimensional metal: cadmium. Using focused ion beam microstructuring to tune the effective thickness, we detected a low-temperature size-dependent resistivity upturn in a finite window sandwiched between ballistic and diffusive regimes. Within this window, the electrical conductivity displays a simultaneous quadratic dependence on both sample size and temperature -- fingerprint of a hydrodynamic flow. This leads us to quantify the amplitude and the temperature dependence of kinematic and dynamic viscosity of the electron fluid. In cadmium, in contrast with graphene and $^3$He, the rate of momentum-conserving $e$-$e$ collisions is not set by the main Fermi energy, but by Lilliputian energy scales and inter-valley bottlenecks.

cond-mat.str-el↗

arXiv2026

Interaction driven transverse thermal resistivity in a phonon gas

The amplitude of the Hall response of electrons can be understood without invoking interactions. Most theories of the phonon thermal Hall effect have likewise opted for a non-interacting picture. Here, we challenge this approach. Our study of WS$_2$, a transition metal dichalcogenide (TMD) insulator, finds that longitudinal, $κ_{xx}$, and transverse, $κ_{xy}$, thermal conductivities peak at almost the same temperature. Their ratio obeys an upper bound, as in other insulators. We then compare transverse thermal transport in a phonon gas and in a molecular gas. In the latter, the Senftleben-Beenakker effect is driven by the competition between molecular collisions and applied magnetic field in setting the distribution of molecular angular momenta. An off-diagonal transport response arises thanks to interactions between non-spherical particles, which do not need to be chiral. By analogy, we argue that in a phonon gas, magnetic field will influence phonon-phonon interactions, and generates a transverse thermal \emph{resistivity}, whose order of magnitude can be accounted for by invoking a Berry force on the drift velocity of the nuclei in the presence of a finite heat. This simple picture gives a reasonable account of the experimentally measured transverse thermal resistivity of seven different crystalline insulators.

cond-mat.mtrl-sci↗

arXiv2026

Thermal Hall conductivity of semimetallic graphite dominated by ambipolar phonon drag

It is now known that in addition to electrons, other quasi-particles such as phonons and magnons can also generate a thermal Hall signal. Graphite is a semimetal with extremely mobile charge carriers of both signs and a large lattice thermal conductivity. We present a study of the thermal Hall effect in highly oriented pyrolytic graphite (HOPG) samples with electronic, phononic and phonon drag contributions to the thermal Hall signal. The measured thermal Hall conductivity ($κ_{xy}$) is two orders of magnitude higher than what is expected by electronic carriers according to the electrical Hall conductivity and the Wiedemann-Franz law, yielding a record Hall Lorenz number of $164.9\times10^{-8}V^2 K^{-2}$ ($\sim$67$L_0$) - the largest ever observed in a metal. The temperature dependence of the thermal Hall conductivity significantly differs from its longitudinal counterpart, ruling out a purely phononic origin of the non-electronic component. Based on the temperature dependence and the amplitudes of the Seebeck and Nernst responses, we demonstrate that ambipolar phonon drag dominates the thermal Hall response of graphite.

cond-mat.mes-hall↗

arXiv2025

Phonon Thermal Hall Effect: The Roles of Disorder, Annealing, and Metallic Contacts

The phonon thermal Hall effect (THE) is a ubiquitous yet poorly understood phenomenon in insulators. Its microscopic origin remains debated, partly due to significant sample-dependent variations that hint at uncontrolled experimental parameters. Using SrTiO$_3$ as a model system, we identify disorder and uncontrolled strain as suppressors of a thermal Hall signal. Crystals with high thermal conductivity exhibit a substantial thermal Hall angle $\nabla T_y / \nabla T_x$ (up to 0.3\% at 9 T), whereas the effect is virtually absent in disordered samples. Crucially, annealing (in air atmosphere) these disordered samples partially restores the THE (approximately 0.1\% at 9 T) with little effect on the longitudinal thermal conductivity. This decoupling reveals that the amplitude of THE is not simply set by the phonon mean free path. Furthermore, measurements performed with metallic and insulating contacts yield identical results on the same sample. This definitively rules out parasitic signals as the effect's origin. Our work, by establishing the phonon THE as an intrinsic property of the crystal lattice and extremely sensitive to disorder, sharply constrains theoretical scenarios.

cond-mat.str-el↗

arXiv2025

Scalable Sondheimer oscillations driven by commensurability between two quantizations

The electrical conductivity of metallic crystals exhibits size effects when the electron mean free path exceeds the sample thickness. One such phenomenon, known as Sondheimer oscillations, was discovered decades ago. These oscillations, periodic in magnetic field, have been hitherto treated with no reference to Landau quantization. Here, we present a study of longitudinal and transverse conductivity in cadmium single crystals with thicknesses ranging from 12.6 to 475 $μ$m, and demonstrate that the amplitude of the first ten oscillations is determined by the quantum of conductance and a length scale that depends on the sample thickness, the magnetic length and the Fermi surface geometry. We argue that this scaling is unexpected in semiclassical scenarios and it arises from the degeneracy of the momentum derivative of the cross-sectional area $A$ along the orientation of the magnetic field $\frac{\partial A}{\partial k_z}$ in cadmium, which couples Landau quantization to the discretization of $k_z$ imposed by the finite sample thickness. We show that the oscillating component of the conductivity is uniquely governed by fundamental constants and the ratio of two degeneracies, which acts as an inverted filling factor. Our conjecture is supported by the absence of such scaling in thin copper crystals.

cond-mat.mes-hall↗

arXiv2025

Layer Decomposition and Morphological Reconstruction for Task-Oriented Infrared Image Enhancement

Infrared image helps improve the perception capabilities of autonomous driving in complex weather conditions such as fog, rain, and low light. However, infrared image often suffers from low contrast, especially in non-heat-emitting targets like bicycles, which significantly affects the performance of downstream high-level vision tasks. Furthermore, achieving contrast enhancement without amplifying noise and losing important information remains a challenge. To address these challenges, we propose a task-oriented infrared image enhancement method. Our approach consists of two key components: layer decomposition and saliency information extraction. First, we design an layer decomposition method for infrared images, which enhances scene details while preserving dark region features, providing more features for subsequent saliency information extraction. Then, we propose a morphological reconstruction-based saliency extraction method that effectively extracts and enhances target information without amplifying noise. Our method improves the image quality for object detection and semantic segmentation tasks. Extensive experiments demonstrate that our approach outperforms state-of-the-art methods.

cs.CV↗

arXiv2025

Cross-modal State Space Modeling for Real-time RGB-thermal Wild Scene Semantic Segmentation

The integration of RGB and thermal data can significantly improve semantic segmentation performance in wild environments for field robots. Nevertheless, multi-source data processing (e.g. Transformer-based approaches) imposes significant computational overhead, presenting challenges for resource-constrained systems. To resolve this critical limitation, we introduced CM-SSM, an efficient RGB-thermal semantic segmentation architecture leveraging a cross-modal state space modeling (SSM) approach. Our framework comprises two key components. First, we introduced a cross-modal 2D-selective-scan (CM-SS2D) module to establish SSM between RGB and thermal modalities, which constructs cross-modal visual sequences and derives hidden state representations of one modality from the other. Second, we developed a cross-modal state space association (CM-SSA) module that effectively integrates global associations from CM-SS2D with local spatial features extracted through convolutional operations. In contrast with Transformer-based approaches, CM-SSM achieves linear computational complexity with respect to image resolution. Experimental results show that CM-SSM achieves state-of-the-art performance on the CART dataset with fewer parameters and lower computational cost. Further experiments on the PST900 dataset demonstrate its generalizability. Codes are available at https://github.com/xiaodonguo/CMSSM.

cs.CV↗

arXiv2025

Abnormal planar Hall effect and disentanglement of incoherent and coherent transport in a Kondo lattice

The nature of localized-itinerant transition in Kondo lattice systems remains a mystery despite intensive investigations in past decades. While it is often identified from the coherent peak in magnetic resistivity, recent angle-resolved photoemission spectroscopy and ultrafast optical spectroscopy revealed a precursor incoherent region with band bending and hybridization fluctuations. This raises the question of how the coherent heavy-electron state is developed from an incoherent background of fluctuating localized moments and then established at sufficiently low temperatures. Here, on the example of the quasi-one-dimensional Kondo lattice compound CeCo$_2$Ga$_8$, we show that planar Hall effect and planar anisotropic magnetoresistance measurements provide an effective way to disentangle the incoherent Kondo scattering contribution and the coherent heavy-electron contribution, and a multi-stage process is directly visualized with lowering temperature by their distinct angle-dependent patterns in magneto-transport. Our idea may be extended to other measurements and thereby opens up a pathway for systematically investigating the fundamental physics of Kondo lattice coherence.

cond-mat.str-el↗

arXiv2025

Angle-dependent planar thermal Hall effect by quasi-ballistic phonons in black phosphorus

The origin of the phonon thermal Hall effect in insulators is a matter of ongoing debate. The large amplitude of the signal in an elemental non-magnetic solid, such as black phosphorus (BP), calls for a minimal mechanism not invoking the spin degree of freedom. Here, we show that a longitudinal heat flow generates a transverse temperature gradient in BP even when the magnetic field, the heat current and the thermal gradient lie in the same plane. The phonon mean-free-path is close to the sample thickness. Therefore, it is unlikely that scattering by point-like symmetry-breaking defects play a major role. We show that the angular dependence of the signal can be mapped to the sum of two sinusoidal components each peaking when the magnetic field is parallel to a high symmetry. We propose that anharmonicity may play a major role and argue that the magnetic field can exert a torque on electric dipolar waves traveling with phonons.

cond-mat.str-el↗

arXiv2023

Onsager reciprocal relation between anomalous transverse coefficients of an anisotropic antiferromagnet

Whenever two irreversible processes occur simultaneously, time-reversal symmetry of microscopic dynamics gives rise, on a macroscopic level, to Onsager's reciprocal relations, which impose constraints on the number of independent components of any transport coefficient tensor. Here, we show that in the antiferromagnetic YbMnBi$_2$, which displays a strong temperature-dependent anisotropy, the Onsager's reciprocal relations are strictly satisfied for anomalous electric ($σ^A_{ij}$) and anomalous thermoelectric ($α^A_{ij}$) conductivity tensors. In contradiction with what was recently reported by Pan $et~al.$ [Nat.Mater. 21, 203 (2022)], we find that $σ^A_{ij} (H)= σ^A_{ji} (-H)$, and $α^A_{ij} (H)= α^A_{ji} (-H)$. This equality holds in the whole temperature window irrespective of the relative weights of the intrinsic or extrinsic mechanisms. The $α^A_{ij}/σ^A_{ij}$ ratio is close to $k_B/e$ at room temperature, but peaks to an unprecedented magnitude of 2.9 $k_B/e$ at $\sim$ 150 K, which may involve nondegenerate carriers of small Fermi surface pockets.

cond-mat.str-el↗

arXiv2023

Spectral2Spectral: Image-spectral Similarity Assisted Spectral CT Deep Reconstruction without Reference

Spectral computed tomography based on a photon-counting detector (PCD) attracts more and more attentions since it has the capability to provide more accurate identification and quantitative analysis for biomedical materials. The limited number of photons within narrow energy bins leads to imaging results of low signal-noise ratio. The existing supervised deep reconstruction networks for CT reconstruction are difficult to address these challenges because it is usually impossible to acquire noise-free clinical images with clear structures as references. In this paper, we propose an iterative deep reconstruction network to synergize unsupervised method and data priors into a unified framework, named as Spectral2Spectral. Our Spectral2Spectral employs an unsupervised deep training strategy to obtain high-quality images from noisy data in an end-to-end fashion. The structural similarity prior within image-spectral domain is refined as a regularization term to further constrain the network training. The weights of neural network are automatically updated to capture image features and structures within the iterative process. Three large-scale preclinical datasets experiments demonstrate that the Spectral2spectral reconstructs better image quality than other the state-of-the-art methods.

eess.IV↗

arXiv2022

Suppression of Correlated Noise with Similarity-based Unsupervised Deep Learning

Image denoising is a prerequisite for downstream tasks in many fields. Low-dose and photon-counting computed tomography (CT) denoising can optimize diagnostic performance at minimized radiation dose. Supervised deep denoising methods are popular but require paired clean or noisy samples that are often unavailable in practice. Limited by the independent noise assumption, current unsupervised denoising methods cannot process correlated noises as in CT images. Here we propose the first-of-its-kind similarity-based unsupervised deep denoising approach, referred to as Noise2Sim, that works in a nonlocal and nonlinear fashion to suppress not only independent but also correlated noises. Theoretically, Noise2Sim is asymptotically equivalent to supervised learning methods under mild conditions. Experimentally, Nosie2Sim recovers intrinsic features from noisy low-dose CT and photon-counting CT images as effectively as or even better than supervised learning methods on practical datasets visually, quantitatively and statistically. Noise2Sim is a general unsupervised denoising approach and has great potential in diverse applications.

cs.LG↗