arXiv subjects

Mu Tian

Publications and source records attributed to Mu Tian.

6 recordsLinked to original sources

arXiv2026

Ridge-Spin-Layer Coupling and Emergent Ridgetronics in 2D Altermagnets

Extending valleytronics from discrete points to continuous lines in momentum space transforms dispersionless bands into a controllable degree of freedom. Here we introduce ridge--spin--layer coupling (RSLC) in two-dimensional (2D) altermagnets, where a one-dimensional continuous line of dispersionless electronic states (a ridge) in momentum space locks to both spin polarization and atomic sublayer. This ridge-induced quenching of kinetic energy mimics flat-band physics, yet crucially, RSLC grants external control, allowing for layer-selective switching of ridge orientation in reciprocal space, spin-filtered transport in real space, and a distinct electric Hall response. Guided by collinear spin layer group symmetry, we identify three 2D candidate materials, namely Mg$_2$Mo$_2$(PO$_5$)$_2$, Ca(FeP)$_2$, and Mg$_2$V$_2$(SO$_5$)$_2$, each featuring a crossed-ridge structure with two ridges, one per spin channel and sublayer. Our work establishes ridgetronics as a controllable platform for direction-discriminating currents, bridging dispersionless bands with multifunctional device operation.

cond-mat.mtrl-sci↗

arXiv2026

Spin layer groups and their corepresentations

Spin layer groups are the crystallographic symmetry groups with a periodic plane, and their symmetry operations are inherited from three-dimensional (3D) spin space groups. However, the direct application of 3D symmetry groups to two-dimensional systems is often inadequate due to anisotropic axes and dimensional reduction. In this work, we systematically classify inequivalent spin layer groups and analytically derive their irreducible corepresentations. This classification establishes a foundational framework for investigating symmetry-protected properties and novel quantum states in low-dimensional magnetic materials.

cond-mat.mtrl-sci↗

arXiv2025

Nonvolatile Electrical Control of Spin via Sliding Fractional Quantum Multiferroics

We propose a fractionally quantized polarization induced by interlayer sliding in bilayer altermagnets, unveiling a previously unrecognized multiferroic phase termed sliding fractional quantum multiferroicity (SFQM). This unconventional magnetic phase uniquely integrates sliding ferroelectricity with fractional quantum ferroelectricity, enabling highly efficient switching and nonvolatile electrical control of spin.~Unlike conventional multiferroics, SFQM simultaneously exhibits lattice-scale atomic displacements, ultralow switching barriers, and spin splitting, giving rise to a large fractionally quantized polarization and strong magnetoelectric coupling. Through symmetry analysis and first-principles calculations, we identify bilayer altermagnet Ca(CoN)$_2$ and its family materials as promising candidates hosting SFQM. In contrast to gate-controlled schemes, the spin-layer coupling in SFQM is intrinsically induced by spontaneous electrical and layer polarization, requiring no sustained gate field and exhibiting nonvolatile character. This mechanism enables nonvolatile electrical control of spin through biaxial sliding, where displacements along the \textit{x}- and \textit{y}-axes generate opposite polarization directions in the layer-dependent electrical polarization. Furthermore, SFQM exhibits a fully switchable anomalous Hall effect and a pronounced magneto-optical response, which can be utilized for its detection and distinction. These findings highlight the promising role of sliding-mediated couplings among unconventional magnetism, fractional quantum ferroelectricity, and stacking order in realizing electrically controllable two-dimensional multiferroics.

cond-mat.mtrl-sci↗

arXiv2025

Symmetry Classification of Altermagnetism and Emergence of Type-IV Magnetism in Two Dimensions

Two-dimensional (2D) magnetism, particularly 2D altermagnetism (AM), has attracted considerable interest due to its exceptional physical properties and broad application potential. However, the classification of AM undergoes a fundamental paradigm shift when transitioning from three-dimensional (3D) to 2D symmetry-enforced fully compensated collinear magnetism$-$a shift that has remained largely overlooked. Here, by extending unconventional magnetism to 2D collinear systems, we identify the symmetry conditions and electronic band characteristics of a distinct magnetic phase: type-IV magnetism. This new class lies beyond the established descriptions of ferromagnetism, conventional antiferromagnetism, and AM. Type-IV magnetism supports the successive emergence of both nonrelativistic spin-degenerate and relativistic spin-splitting phenomena, belonging strictly to neither conventional antiferromagnetism nor standard AM. We further establish a universal symmetry classification framework for 2D type-IV magnets via a mapping from the collinear spin layer group to the magnetic layer group. Monolayer MgCr$_2$O$_3$ and monolayer BaMn$_2$Ch$_3$ (Ch=Se, Te) are showcased as representative materials, exhibiting gate-tunable reversible spin textures and the quantum electric Hall effect, respectively. Our work underscores the rich functional prospects of type-IV magnets, offering a new route toward spin manipulation and anomalous transport that promises innovative designs for high-performance spintronic devices.

cond-mat.mtrl-sci↗

arXiv2023

A dynamic interactive learning framework for automated 3D medical image segmentation

Many deep learning based automated medical image segmentation systems, in reality, face difficulties in deployment due to the cost of massive data annotation and high latency in model iteration. We propose a dynamic interactive learning framework that addresses these challenges by integrating interactive segmentation into end-to-end weak supervised learning with streaming tasks. We develop novel replay and label smoothing schemes that overcome catastrophic forgetting and improve online learning robustness. For each image, our multi-round interactive segmentation module simultaneously optimizes both front-end predictions and deep learning segmenter. In each round, a 3D "proxy mask" is propagated from sparse user inputs based on image registration, serving as weak supervision that enable knowledge distillation from the unknown ground truth. In return, the trained segmenter explicitly guides next step's user interventions according to a spatial residual map from consecutive front or back-end predictions. Evaluation on 3D segmentation tasks (NCI-ISBI2013 and BraTS2015) shows that our framework generates online learning performances that match offline training benchmark. In addition, with a 62% reduction in total annotation efforts, our framework produces competitive dice scores comparing to online and offline learning which equipped with full ground truth. Furthermore, such a framework, with its flexibility and responsiveness, could be deployed behind hospital firewall that guarantees data security and easy maintenance.

cs.CV↗

arXiv2023

AttenScribble: Attentive Similarity Learning for Scribble-Supervised Medical Image Segmentation

The success of deep networks in medical image segmentation relies heavily on massive labeled training data. However, acquiring dense annotations is a time-consuming process. Weakly-supervised methods normally employ less expensive forms of supervision, among which scribbles started to gain popularity lately thanks to its flexibility. However, due to lack of shape and boundary information, it is extremely challenging to train a deep network on scribbles that generalizes on unlabeled pixels. In this paper, we present a straightforward yet effective scribble supervised learning framework. Inspired by recent advances of transformer based segmentation, we create a pluggable spatial self-attention module which could be attached on top of any internal feature layers of arbitrary fully convolutional network (FCN) backbone. The module infuses global interaction while keeping the efficiency of convolutions. Descended from this module, we construct a similarity metric based on normalized and symmetrized attention. This attentive similarity leads to a novel regularization loss that imposes consistency between segmentation prediction and visual affinity. This attentive similarity loss optimizes the alignment of FCN encoders, attention mapping and model prediction. Ultimately, the proposed FCN+Attention architecture can be trained end-to-end guided by a combination of three learning objectives: partial segmentation loss, a customized masked conditional random fields and the proposed attentive similarity loss. Extensive experiments on public datasets (ACDC and CHAOS) showed that our framework not just out-performs existing state-of-the-art, but also delivers close performance to fully-supervised benchmark. Code will be available upon publication.

cs.CV↗