arXiv subjects

Tianshu Li

Publications and source records attributed to Tianshu Li.

At least 19 recordsLinked to original sources

arXiv2026

Anisotropic Short-Range Order Modulates Ferroelectric Switching in Wurtzite ScAlN Alloys

Ferroelectric switching in wurtzite alloys is typically understood in terms of composition, strain, defects, and interfaces, while local chemical order is often neglected or treated as a secondary perturbation. Here we show that short-range order (SRO) is a previously overlooked microscopic variable that substantially influences the intrinsic switching barrier. Using first-principles canonical sampling, we find that wurtzite ScAlN develops a robust, highly anisotropic SRO that challenges the conventional random-alloy picture. This ordering suppresses in-plane Sc--N--Sc motifs while enhancing columnar mixed-cation chains along the polar $c$ axis, reflecting the symmetry-distinct polar and basal directions of the wurtzite lattice and reorganizing its polar connectivity. Relative to random-alloy structures, SRO systematically increases the intrinsic switching barrier across a broad composition range. Motif-resolved analysis further identifies the population of columnar Sc--N--Al--N--Sc motifs as the primary structural descriptor underlying switching-barrier variations among configurations with different local order. These results establish anisotropic SRO as an independent degree of freedom for tuning ferroelectric switching. More broadly, they reveal how local chemical order can couple to the symmetry-distinct directions of a polar semiconductor lattice to modify functional behavior. Our findings lay a foundation for SRO engineering as a route to tailoring switching barriers without changing alloy composition.

cond-mat.mtrl-sci↗

arXiv2026

LoopMoE: Unifying Iterative Computation with Mixture-of-Experts for Language Modeling

Mixture-of-Experts (MoE) and looped architectures scale models along two orthogonal axes, namely parameter capacity and effective depth. However, mainstream looped architectures rely on dense backbones that couple parameter count with per-token FLOPs, which makes it impossible to isolate the effect of iterative computation under matched budgets. To this end, we present LoopMoE, a looped MoE language model that integrates sparse routing with iterative weight-shared computation through two designs. The first is IterAdaLN, which resolves weight-sharing symmetry via a modulation signal jointly conditioned on the iteration index and the per-token hidden state. The second is a capacity-balancing strategy that recovers the attention-to-FFN active parameter ratio of well-tuned non-looped references. Together, these designs enable the first strictly controlled, head-to-head evaluation of a looped MoE against a Vanilla MoE under identical total parameters, per-token FLOPs, and active sublayer ratios. Across nine downstream benchmarks, LoopMoE's average improvement over its matched vanilla MoE increases from over 1 point at the 3B scale to approximately 3 points at the 9B scale. These results provide initial evidence that the benefits of iterative sparse computation may strengthen with scale, positioning LoopMoE as a promising architecture for scalable looped language models.

cs.LG↗

arXiv2026

Shining light on short-range atomic ordering in semiconductors alloys

The functional properties of semiconductors are typically controlled by tailoring their chemical composition and their state of strain, and by controlling their long-range structural order, including the presence of extended defects such as dislocations. In addition to these approaches, theoretical predictions suggest that short-range order (SRO) of atoms in group-IV semiconductor alloys can modify the bandgap, a defining property of any semiconductor. Herein, a new machine learning enabled, computation-guided methodology for extended X-ray absorption fine structure (EXAFS) analysis of SRO is used to quantify the effects of local atomic order on the bandgap of germanium-tin (GeSn) alloy single crystal nanostructures with well-controlled strain and composition. Correlative analysis of EXAFS and photoluminescence (PL) establishes the relationship between bandgap and the Warren-Cowley short-range order (WC-SRO) parameter of the GeSn alloys. It is further demonstrated that SRO can be tuned over a broad range by post-deposition annealing of the alloy crystals. This work establishes control of SRO as an important design parameter for semiconducting properties and suggests the potential for quantitative measurement and tuning of SRO in other semiconductor alloy systems.

cond-mat.mtrl-sci↗

arXiv2025

Defining the urban "local" with low dimensional manifolds of human mobility networks

Urban science has largely relied on universal models, rendering the heterogeneous and locally specific nature of cities effectively invisible. Here we introduce a topological framework that defines and detects localities in human mobility networks. We empirically demonstrate that these human mobility network localities are rigorous geometric entities that map directly to geographic localities, revealing that human mobility networks lie on manifolds of dimension <=5. This representation provides a compact theoretical foundation for spatial embedding and enables efficient applications to facility location and propagation modeling. Our approach reconciles local heterogeneity with universal representation, offering a new pathway toward a more comprehensive urban science.

physics.soc-ph↗

arXiv2025

Mo Atom Rearrangement Drives Layer-Dependent Reactivity in Two-Dimensional MoS2

Two-dimensional (2D) materials offer a valuable platform for manipulating and studying chemical reactions at atomic level, owing to the ease of controlling their microscopic structure at the nanometer scale. While extensive research has been conducted on the structure-dependent chemical activity of 2D materials, the influence of structural transformation during the reaction remains largely unexplored. In this work, we report the layer-dependent chemical reactivity of MoS2 during a nitridation atomic substitution reaction and attribute it to the rearrangement of Mo atoms. Our results show that the chemical reactivity of MoS2 decreases as the number of layers is reduced in the few-layer regime. In particular, monolayer MoS2 exhibits significantly lower reactivity compared to its few-layer and multilayer counterparts. Atomic-resolution transmission electron microscope (TEM) reveals that MoN nanonetworks form as reaction products from monolayer and bilayer MoS2, with the continuity of the MoN crystals increasing with layer number, consistent with the local conductivity mapping data. The layer-dependent reactivity is attributed to the relative stability of the hypothetically formed MoN phase which retain the number of Mo atomic layers present in the precursor. Specifically, the low chemical reactivity of monolayer MoS2 is attributed to the high energy cost associated with Mo atom diffusion and migration necessary to form multi-layer Mo lattices in the thermodynamically stable MoN phase. This study underscores the critical role of lattice rearrangement in governing chemical reactivity and highlights the potential of 2D materials as versatile platforms for advancing the understanding of materials chemistry at atomic scale.

cond-mat.mtrl-sci↗

arXiv2025

Phase-based stimulated emission depletion (pSTED) magnetic particle imaging

Magnetic particle imaging (MPI) is an in vivo method to detect magnetic nanoparticles for cell tracking, vascular imaging, and molecular target imaging without ionizing radiation. Current magnetic particle imaging is accomplished by forming an field-free line (FFL) through a gradient selection field. By translating and rotating FFL under excitation and drive fields, the harmonic complex signal of a point source forms a Lorentzian-shape point spread function on the plane perpendicular to FFL. The Lorentzian PSF has a finite size and limited resolution due to the non-sharp Langevin function and weak selection field. This study proposes a donut-shaped focal spot by borrowing the stimulated emission depletion (STED) fluorescence microscopy principle. The influence of the gradient selection field on the relaxation time of magnetic particles determines the nonlinear phase shift of the harmonic complex signals, resulting in the formation of a donut-shaped focal spot. By subtracting the donut-shaped focal spot from the Lorentzian focal spot, the STED focal spot size was reduced by up to 4 times beyond the Langevin magnetization resolution barrier. In human brain FFL-based MPI scanner, the donut-shaped focal spot can be used to reconstruct images with super-resolution and super-sensitivity through the deconvoution of the STED focal spot and filtered backprojection algorithm.

physics.med-ph↗

arXiv2025

Pangu Ultra MoE: How to Train Your Big MoE on Ascend NPUs

Sparse large language models (LLMs) with Mixture of Experts (MoE) and close to a trillion parameters are dominating the realm of most capable language models. However, the massive model scale poses significant challenges for the underlying software and hardware systems. In this paper, we aim to uncover a recipe to harness such scale on Ascend NPUs. The key goals are better usage of the computing resources under the dynamic sparse model structures and materializing the expected performance gain on the actual hardware. To select model configurations suitable for Ascend NPUs without repeatedly running the expensive experiments, we leverage simulation to compare the trade-off of various model hyperparameters. This study led to Pangu Ultra MoE, a sparse LLM with 718 billion parameters, and we conducted experiments on the model to verify the simulation results. On the system side, we dig into Expert Parallelism to optimize the communication between NPU devices to reduce the synchronization overhead. We also optimize the memory efficiency within the devices to further reduce the parameter and activation management overhead. In the end, we achieve an MFU of 30.0% when training Pangu Ultra MoE, with performance comparable to that of DeepSeek R1, on 6K Ascend NPUs, and demonstrate that the Ascend system is capable of harnessing all the training stages of the state-of-the-art language models. Extensive experiments indicate that our recipe can lead to efficient training of large-scale sparse language models with MoE. We also study the behaviors of such models for future reference.

cs.CL↗

arXiv2025

Accurate, transferable, and verifiable machine-learned interatomic potentials for layered materials

Twisted layered van-der-Waals materials often exhibit unique electronic and optical properties absent in their non-twisted counterparts. Unfortunately, predicting such properties is hindered by the difficulty in determining the atomic structure in materials displaying large moir\'e domains. Here, we introduce a split machine-learned interatomic potential and dataset curation approach that separates intralayer and interlayer interactions and significantly improves model accuracy -- with a tenfold increase in energy and force prediction accuracy relative to conventional models. We further demonstrate that traditional MLIP validation metrics -- force and energy errors -- are inadequate for moir\'e structures and develop a more holistic, physically-motivated metric based on the distribution of stacking configurations. This metric effectively compares the entirety of large-scale moir\'e domains between two structures instead of relying on conventional measures evaluated on smaller commensurate cells. Finally, we establish that one-dimensional instead of two-dimensional moir\'e structures can serve as efficient surrogate systems for validating MLIPs, allowing for a practical model validation protocol against explicit DFT calculations. Applying our framework to HfS2/GaS bilayers reveals that accurate structural predictions directly translate into reliable electronic properties. Our model-agnostic approach integrates seamlessly with various intralayer and interlayer interaction models, enabling computationally tractable relaxation of moir\'e materials, from bilayer to complex multilayers, with rigorously validated accuracy.

cond-mat.mtrl-sci↗

arXiv2025

Metastability and Ostwald Step Rule in the Crystallisation of Diamond and Graphite from Molten Carbon

The crystallisation of carbon from the melt under extreme conditions is highly relevant to earth and planetary science, materials manufacturing, and nuclear fusion research. The thermodynamic conditions near the graphite-diamond-liquid (GDL) triple point are especially of interest for geological and technological applications, but high-pressure flash heating experiments aiming to resolve this region of the phase diagram of carbon exhibit large discrepancies. Experimental challenges are often related to the persistence of metastable crystalline or glassy phases, superheated crystals, or supercooled liquids. A deeper understanding of the crystallisation kinetics of diamond and graphite is crucial for effectively interpreting the outcomes of these experiments. Here, we reveal the microscopic mechanisms of diamond and graphite nucleation from liquid carbon through molecular simulations with first-principles machine learning potentials. Our simulations accurately reproduce the experimental phase diagram of carbon in the region around the GDL triple point and show that liquid carbon crystallises spontaneously upon cooling at constant pressure. Surprisingly, metastable graphite crystallises in the domain of diamond thermodynamic stability at pressures above the triple point. Furthermore, whereas diamond crystallises through a classical nucleation pathway, graphite follows a two-step process in which low-density fluctuations forego ordering. Calculations of the nucleation rates of the two competing phases confirm this result and reveal a manifestation of Ostwald's step rule where the strong metastability of graphite hinders the transformation to the stable diamond phase. Our results provide a new key to interpreting melting and recrystallisation experiments and shed light on nucleation kinetics in polymorphic materials with deep metastable states.

cond-mat.mtrl-sci↗

arXiv2025

Semiconductor-compatible topological digital alloys

Recently, GeSn alloys have attracted much interest for direct-gap infrared photonics and as potential topological materials which are compatible with the semiconductor industry. However, for photonics, the high-Sn content required leads to low detectivity, associated with poor material quality, and the (>35%) Sn required for topological properties have been out of reach experimentally. Here, we demonstrate that by patterning the Sn distribution within Ge, the electronic properties have a far greater tunability than is possible with the random alloy. For the GeSn \delta-digital alloy (DA) formed by confining Sn atoms in atomic layer(s) along the [111] direction of Ge, we show that ~10% Sn can lead to a triple-point semimetal. These findings are understood in terms of Sn ordering causing spatial separation of Sn and Ge band edges, leading to band inversion. This mechanism can also lead to a weak topological insulator, Weyl semimetal, and enables tunable direct bandgaps down to 2 meV, covering the entire infrared range. This DA induced topological properties are also identified in compound semiconductors, such as InAs1-xSbx, showing the general applicability of the DA design for realizing topological properties on conventional semiconductor platforms. Our findings not only point to a new class of currently unexplored topological systems accessible by epitaxy, but also establish the promise of low-Sn GeSn DAs for application as infrared laser diodes and photodetectors in Si photonic integrated circuits and infrared image sensors.

cond-mat.mtrl-sci↗

arXiv2024

Combining Hyperbolic Quadrature Method of Moments and Discrete-Velocity-Direction Models for Solving BGK-type Equations

This paper introduces the discrete-velocity-direction model (DVDM) in conjunction with the hyperbolic quadrature method of moments (HyQMOM) to develop a multidimensional spatial-temporal approximation of the BGK equation, termed DVD-HyQMOM. Serving as a multidimensional extension of HyQMOM, DVD-HyQMOM model achieves higher accuracy than other DVDM submodels, especially with an increased number of abscissas. The efficiency and effectiveness of this model are demonstrated through various numerical tests.

physics.comp-ph↗

arXiv2024

Geographic Space as Manifolds

The communications and interrelations between different locations on the Earth's surface have far-reaching implications for both social and natural systems. Effective spatial analytics ideally require a spatial representation, where geographic principles are succinctly expressed within a defined metric space. However, common spatial representations, including map-based or network-based approaches, fall short by incompletely or inaccurately defining this metric space. Here we show, by introducing an inverse friction factor that captures the spatial constraints in spatial networks, that a homogeneous, low-dimensional spatial representation - termed the Geographic Manifold - can be achieved. We illustrate the effectiveness of the Geographic Manifold in two classic scenarios of spatial analytics - location choice and propagation, where the otherwise complicated analyses are reduced to straightforward regular partitioning and concentric diffusing, respectively on the manifold with a high degree of accuracy. We further empirically explain and formally prove the general existence of the Geographic Manifold, which is grounded in the intrinsic Euclidean low-dimensional statistical physics properties of geographic phenomena. This work represents a step towards formalizing Tobler's famous First Law of Geography from a geometric approach, where a regularized geospace thereby yielded is expected to contribute in learning abstract spatial structure representations for understanding and optimization purposes.

physics.soc-ph↗

arXiv2024

Atomic short-range order: a new degree of freedom for band engineering of GeSn semiconductor alloys

Chemical short-range order (SRO) in alloys denotes the statistical preference or avoidance between atomic species on neighboring lattice sites. Here, we highlight SRO as a powerful new mechanism for semiconductor alloy band engineering. Atom probe tomography reveals a significantly higher probability of Sn-Sn first nearest neighbors (1NNs) in thin-film GeSn alloys grown by molecular beam epitaxy (MBE) vs. chemical vapor deposition (CVD). Remarkably, although lower Sn concentration typically widens the bandgap, we find that the stronger presence of Sn-Sn 1NN pairs in MBE samples overrides this trend, resulting in a narrower bandgap despite having 2 at.\% lower Sn content than CVD samples. First-principles modeling corroborates this effect, attributing these SRO variations to distinctive surface terminations and growth temperatures between MBE and CVD. These findings establish SRO as a new degree of freedom for semiconductor band engineering beyond composition, strain, and quantum confinement, unlocking novel device mechanisms for the post-Moore era.

cond-mat.mtrl-sci↗

arXiv2024

Stimulated Emission Depletion (STED) Magnetic Particle Imaging

Magnetic particle imaging (MPI) is an in-vivo imaging method to detect magnetic nanoparticles for blood vessel imaging and molecular target imaging. Compared with conventional molecular imaging devices (such as nuclear medicine imaging PET and SPECT), magnetic nanoparticles have longer storage periods than radionuclides without ionizing radiation. MPI has higher detection sensitivity compared with MRI. To accurately locate molecular probes in living organisms, high-resolution images are needed to meet the requirements of precision medicine. The spatial resolution of the latest domestic and international MPI equipment is 1-6 mm and has not yet met the requirements of medical imaging detection. We previously studied the spatial encoding technology based on pulsed square wave stimulation, which significantly improved the image resolution along the field free line (FFL) direction. This study proposes an innovative idea of high-resolution MPI based on stimulated emission depletion (STED) of magnetic nanoparticle signals. The stimulated emission was implemented by using cosine stimulation on FFL-based MPI scanner systems. The STED signal was generated by adding an offset magnetic field parallel to the FFL, which may form a donut-shaped focal spot or a regular Gaussian focal spot depending on the offset field strength. Focal spot modulation techniques and deconvolution algorithms were developed to improve image resolution.

physics.med-ph↗

arXiv2023

Local Alloy Order in a Ge1-xSnx/Ge Epitaxial Layer

The local ordering of atoms in alloys directly has a strong impact on their electronic and optical properties. This is particularly relevant in nonrandom alloys, especially if they are deposited using far from the equilibrium processes, as is the case of epitaxial Ge1-xSnx layers. In this work, we investigate the arrangement of Ge and Sn atoms in optoelectronic grade Ge1-xSnx epitaxial layers featuring a Sn content in the 5-14% range by using polarization-dependent Raman spectroscopy and density-functional-theory calculations. The thorough analysis of the polarization-dependent spectra in parallel and perpendicular configuration allowed us to properly tag all the observed vibrational modes, and to shed light on that associated to disorder-assisted Raman transitions. Indeed, with the help of large-scale atomistic simulations, we were able to highlight how the presence of Sn atoms, that modify the local environments of Ge atoms, gives rise to two spectral features at different Raman shifts, corresponding to distortions of the atomic bonds. This analysis provides a valuable framework for advancing the understanding of the vibrational properties in Ge1-xSnx alloys, particularly with regard to the impact of local ordering of the different atomic species.

cond-mat.mtrl-sci↗

arXiv2023

Accelerated Screening of Ternary Chalcogenides for High-Performance Optoelectronic Materials

Chalcogenides, which refer to chalcogen anions, have attracted considerable attention in multiple fields of applications, such as optoelectronics, thermoelectrics, transparent contacts, and thin film transistors. In comparison to oxide counterparts, chalcogenides have demonstrated higher mobility and \textit{p}-type dopability, owing to larger orbital overlaps between metal-X covalent chemical bondings and higher-energy valence bands derived by p-orbitals. Despite the potential of chalcogenides, the number of successfully synthesized compounds remains relatively low compared to oxides, suggesting the presence of numerous unexplored chalcogenides with fascinating physical characteristics. In this study, we implemented a systematic high-throughput screening process combined with first-principles calculations on ternary chalcogenides using 34 crystal structure prototypes. We generated a computational material database containing over 400,000 compounds by exploiting the ion-substitution approach at different atomic sites with elements in the periodic table. The thermodynamic stabilities of the candidates were validated using the chalcogenides included in the Open Quantum Materials Database. Moreover, we trained a model based on Crystal Graph Convolutional Neural Networks to predict the thermodynamic stability of novel materials. Furthermore, we theoretically evaluated the electronic structures of the stable candidates using accurate hybrid functionals. A series of in-depth characteristics, including the carrier effective masses, electronic configuration, and photovoltaic conversion efficiency, was also investigated. Our work provides useful guidance for further experimental research in the synthesis and characterization of such chalcogenides as promising candidates, as well as charting the stability and optoelectronic performance of ternary chalcogenides.

cond-mat.mtrl-sci↗

arXiv2022

Spin-state Directed Synthesis of >20 micrometers 2D Layered Transition Metal Hydroxides via Edge-on Condensation

Layered transition metal hydroxides (LTMHs) with transition metal centers sandwiched between layers of coordinating hydroxide anions have attracted considerable interest for their potential in developing clean energy sources and storage technologies. However, two dimensional (2D) LTMHs remain largely unstudied in terms of their physical properties and the applications in electronic devices. Here, directed by the relationship of the spin state of 3d transition metal (TM) ions such as Ni, Co, Cu, and the corresponding geometry of the crystal field, we discover that Ni2+ with perfect Oh symmetry is ideal for intraplanar growth, leading to the achievement of >20 {\mu}m {\alpha}-Ni(OH)2 2D crystals with high yield, which are the largest 2D domains reported so far. We also report the successful synthesis of 2D Co(OH)2 crystals (>40 {\mu}m) with less yield due to the slight geometry distortion resulted from uneven number of electrons. Moreover, the detailed structural characterization of synthesized {\alpha}-Ni(OH)2 are performed; the optical band gap energy is extrapolated as 2.54 eV from optical absorption measurements and is measured as 2.50 eV from reflected electrons energy loss spectroscopy (REELS), suggesting the potential as insulating 2D dielectric material for electronic devices. Furthermore, key parameters of the hydrothermal reaction including soaking temperature, starting pH and cooling rate, are systematically tuned to understand their effects on morphological and crystallographic perspectives, allowing the establishment of a 2D growth mechanism. This work demonstrates a scalable pathway to synthesize large 2D LTMHs from simple methods, paving the way for the study of fundamental physical properties and device applications of 2D LTMHs.

cond-mat.mtrl-sci↗

arXiv2022

Two-Timescale Transmission Design for Wireless Communication Systems Aided by Active RIS

This paper considers an active reconfigurable intelligent surface (RIS)-aided communication system, where an M-antenna base station (BS) transmits data symbols to a single-antenna user via an N-element active RIS. We use two-timescale channel state information (CSI) in our system, so that the channel estimation overhead and feedback overhead can be decreased dramatically. A closed-form approximate expression of the achievable rate (AR) is derived and the phase shift at the active RIS is optimized. In addition, we compare the performance of the active RIS system with that of the passive RIS system. The conclusion shows that the active RIS system achieves a lager AR than the passive RIS system.

eess.SP↗