arXiv subjects

Yishu Wang

Publications and source records attributed to Yishu Wang.

At least 19 recordsLinked to original sources

arXiv2026

CrystalMO-TuRBO: Multi-Objective Trust-Region Bayesian Optimization for High-precision Joint Crystal Structure Refinement

Crystal structure refinement is a fundamental inverse problem in materials characterization, where structural parameters are optimized to reproduce experimental diffraction data. Conventional approaches, such as least-squares and likelihood-based optimization, rely on local search and often struggle with non-convex, noisy, and highly correlated parameter landscapes, particularly when integrating multiple diffraction modalities. Joint refinement of X-ray and neutron data is especially challenging due to their complementary but competing sensitivities, which are typically combined through scalarized objectives requiring manual weighting and leading to suboptimal solutions. We propose CrystalMO-TuRBO, a multi-objective trust region Bayesian optimization architecture for joint crystal structure refinement. The method models X-ray and neutron discrepancies as separate objectives and transforms the problem into a normalized maximization setting. A two-phase optimization strategy is introduced: Phase 1 performs global exploration using parallel trust-region Bayesian optimization across multiple scalarizations to identify promising regions of the parameter space, while Phase 2 conducts localized refinement within a shrinking region to achieve high-precision solutions. This design explicitly separates global search from fine-grained optimization, addressing the unique accuracy requirements of refinement tasks. We evaluate the proposed method on experimentally collected X-ray and neutron diffraction data from single-crystal Ho2Ti2O7. Results demonstrate improved convergence, robustness, and parameter precision compared to classical refinement methods and Bayesian optimization baselines on refinement of a single-crystal pyrochlore material system.

cs.LG↗

arXiv2026

PeroMAS: A Multi-agent System of Perovskite Material Discovery

As a pioneer of the third-generation photovoltaic revolution, Perovskite Solar Cells (PSCs) are renowned for their superior optoelectronic performance and cost potential. The development process of PSCs is precise and complex, involving a series of closed-loop workflows such as literature retrieval, data integration, experimental design, and synthesis. However, existing AI perovskite approaches focus predominantly on discrete models, including material design, process optimization,and property prediction. These models fail to propagate physical constraints across the workflow, hindering end-to-end optimization. In this paper, we propose a multi-agent system for perovskite material discovery, named PeroMAS. We first encapsulated a series of perovskite-specific tools into Model Context Protocols (MCPs). By planning and invoking these tools, PeroMAS can design perovskite materials under multi-objective constraints, covering the entire process from literature retrieval and data extraction to property prediction and mechanism analysis. Furthermore, we construct an evaluation benchmark by perovskite human experts to assess this multi-agent system. Results demonstrate that, compared to single Large Language Model (LLM) or traditional search strategies, our system significantly enhances discovery efficiency. It successfully identified candidate materials satisfying multi-objective constraints. Notably, we verify PeroMAS's effectiveness in the physical world through real synthesis experiments.

cs.MA↗

arXiv2026

FlowEvo: Self-Evolving Agents through the Co-Evolution of Workflows and Executable Skills

Large language model agents can adapt to complex tasks by constructing workflows at inference time, but procedures discovered in one episode are usually discarded after execution. Existing skill libraries provide reusable executable routines, but are typically assembled offline and do not grow from the agent's own workflows. We introduce FlowEvo, a training-free framework in which workflows and skills co-evolve at inference time. FlowEvo compiles successful workflows into callable skills, stores them in a persistent bank, and uses retrieved skills either through direct execution or as context for constructing new workflows. It also tracks each skill's downstream utility and suppresses skills that cause negative transfer. Using a shared GPT-4o-mini backbone, FlowEvo achieves the highest accuracy among 8 baselines on the full standard splits of ALFWorld, HumanEval, MBPP, GSM8K, and MATH-500. On ALFWorld, it reaches 85.6%, 26.4 points above the strongest baseline, while using roughly one third as many tokens. Across 10 base models spanning 7B to 671B parameters, FlowEvo outperforms ExpeL in 49 of 50 model-dataset comparisons. Code is available at https://github.com/DEFENSE-SEU/FlowEvo.

cs.AI↗

arXiv2026

DREvo: Distilling Recalibrated Historical Experience for Harness Self-Evolution

Harness plays a critical role in large language model agent performance, and building a high-performing harness requires substantial expert effort. Therefore, recent research has increasingly explored harness self-evolution, which iteratively proposes, evaluates, and improves harnesses using historical trial experience. However, accumulated historical experience does not always translate into stable search guidance, and performance often fluctuates substantially across evolution iterations, making it difficult to reliably discover high-performing harnesses under a limited evolution budget. We identify two limitations in how existing harness self-evolution methods leverage historical experience: (1) Lack of dynamic reassessment of whether historical experience remains valid for the current harness, and (2) Lack of explicit mechanisms for translating valid historical experience into actionable search directions. To address these limitations, we propose a new harness self-evolution method, named DREvo, which integrates function-level evidence anchoring, state-dependent evidence recalibration, and role-conditioned search intent distillation to determine which historical evidence remains valid and where the harness should evolve next. Under limited evolution budgets, DREvo exhibits smoother evolution trajectories, achieves the highest accuracy on all five benchmarks, and delivers average gains of 16.2% and 14.2% over the evaluated baselines on domain reasoning and agentic tasks, respectively.

cs.MA↗

arXiv2026

Beyond Forecasting: The Belief-to-Trade Layer in Prediction-Market Agents

Forecasting future events has attracted growing attention as a testbed for general-purpose AI. A natural way to ground this evaluation is let the models trade in the prediction markets. Trading, however, requires more than forecasting. Moreover, recent benchmarks report a substantial gap between calibrated probability scores and the trading results. We propose Raven-Agent, to the best of our knowledge, the first autonomous trading agent for prediction markets. On a controlled replay over an archived decision set, our architecture achieves the only positive return and the only positive risk-adjusted return among all tested policies. We have released our code in https://github.com/Alchemist-X/predict-raven .

cs.AI↗

arXiv2025

Physically-Constrained Autoencoder-Assisted Bayesian Optimization for Refinement of High-Dimensional Defect-Sensitive Single Crystalline Structure

Physical properties and functionalities of materials are dictated by global crystal structures as well as local defects. To establish a structure-property relationship, not only the crystallographic symmetry but also quantitative knowledge about defects are required. Here we present a hybrid Machine Learning framework that integrates a physically-constrained variational-autoencoder (pcVAE) with different Bayesian Optimization (BO) methods to systematically accelerate and improve crystal structure refinement with resolution of defects. We chose the pyrochlore structured Ho2Ti2O7 as a model system and employed the GSAS2 package for benchmarking crystallographic parameters from Rietveld refinement. However, the function space of these material systems is highly nonlinear, which limits optimizers like traditional Rietveld refinement, into trapping at local minima. Also, these naive methods don't provide an extensive learning about the overall function space, which is essential for large space, large time consuming explorations to identify various potential regions of interest. Thus, we present the approach of exploring the high Dimensional structure parameters of defect sensitive systems via pretrained pcVAE assisted BO and Sparse Axis Aligned BO. The pcVAE projects high-Dimensional diffraction data consisting of thousands of independently measured diffraction orders into a lowD latent space while enforcing scaling invariance and physical relevance. Then via BO methods, we aim to minimize the L2 norm based chisq errors in the real and latent spaces separately between experimental and simulated diffraction patterns, thereby steering the refinement towards potential optimum crystal structure parameters. We investigated and compared the results among different pcVAE assisted BO, non pcVAE assisted BO, and Rietveld refinement.

cond-mat.mtrl-sci↗

arXiv2025

L2-relaxation for Economic Prediction

We leverage an ensemble of many regressors, the number of which can exceed the sample size, for economic prediction. An underlying latent factor structure implies a dense regression model with highly correlated covariates. We propose the L2-relaxation method for estimating the regression coefficients and extrapolating the out-of-sample (OOS) outcomes. This framework can be applied to policy evaluation using the panel data approach (PDA), where we further establish inference for the average treatment effect. In addition, we extend the traditional single unit setting in PDA to allow for many treated units with a short post-treatment period. Monte Carlo simulations demonstrate that our approach exhibits excellent finite sample performance for both OOS prediction and policy evaluation. We illustrate our method with two empirical examples: (i) predicting China's producer price index growth rate and evaluating the effect of real estate regulations, and (ii) estimating the impact of Brexit on the stock returns of British and European companies.

econ.EM↗

arXiv2025

Temporal and spatial separations between spin glass and short-range order

Broken-symmetry-induced order parameters account for many phenomena in condensed matter physics. For spin glasses, such a framework dictates its theoretical construction, whereas experiments have only established dynamical behaviors such as frequency dependent magnetic susceptibility and aging but not the thermodynamic phase. Experimental techniques have limitations when the spin glass is probed as an isolated state. To resolve this conundrum, we create an evolution from long-range order using a well-controlled tuning of the disorder on a spinel's sublattice. Cross-referencing a series of specimens at both long (milliseconds to seconds) and short (picosecond) time scales illustrates the relationship between spin glass and long- and short-range orders. The dynamics of short- and long-range order formations are not affected by disorder, as revealed by neutron magnetic diffuse scattering, however the ranges of these orderings are changed by the introduced disorder. Across all specimens, the inflection point of the correlation length's temperature dependence fully matches with the peak in heat capacity, while spin glass can freeze either below or well above this characteristic temperature of spin order formation. Our results identify an uncorrelated coexistence of the two and attribute components of the spin glass to individual spins at domain walls between spin clusters.

cond-mat.dis-nn↗

arXiv2025

The Future of the Correlated Electron Problem

A central problem in modern condensed matter physics is the understanding of materials with strong electron correlations. Despite extensive work, the essential physics of many of these systems is not understood and there is very little ability to make predictions in this class of materials. In this manuscript we share our personal views on the major open problems in the field of correlated electron systems. We discuss some possible routes to make progress in this rich and fascinating field. This manuscript is the result of the vigorous discussions and deliberations that took place at Johns Hopkins University during a three-day workshop January 27, 28, and 29, 2020 that brought together six senior scientists and 46 more junior scientists. Our hope, is that the topics we have presented will provide inspiration for others working in this field and motivation for the idea that significant progress can be made on very hard problems if we focus our collective energies.

cond-mat.str-el↗

arXiv2025

Fermi surface origin of the low-temperature magnetoresistance anomaly

A magnetoresistance (MR) anomaly at low temperatures has been observed in a variety of systems, ranging from low-dimensional chalcogenides to spin and charge density wave (SDW/CDW) metals and, most recently, topological semimetals. In some systems parabolic magnetoresistance can rise to hundreds of thousands of times its low-temperature, zero-field value. While the origin of such a dramatic effect remains unresolved, these systems are often low-carrier-density compensated metals, and the physics is expected to be quasi-classical. Here we demonstrate that this MR anomaly in temperature also exists in high conductivity good metals with large Fermi surfaces, namely Cr, Mo, and W, for both linear and quadratic field-dependent regimes with their non-saturation attributed to open orbit and electron-hole compensation, respectively. We provide evidence that quantum transport across sharp Fermi surface arcs, but not necessarily the full cyclotron orbit, governs this low-temperature MR anomaly. In Cr, extremely sharp curvatures are induced by superposed lattice and SDW band structures. One observes an overlay of the temperature dependence of three phenomena: namely, MR at a constant high field, linear MR in the low-field limit, and Shubnikov-de Haas (SdH) oscillations of the lightest orbit. In Mo, the temperature dependence of low-T MR anomaly extends beyond those of its SdH oscillations but disappears at temperatures where Kohler's scaling reemerges. In the low-temperature and high-field limit, large magnetoresistance from carriers circling quantum orbits is the three-dimensional analogy to the zero-conductance state of carrier localization in the integer quantum Hall effect, especially with regard to the adverse effect of disorder.

cond-mat.mtrl-sci↗

arXiv2024

Continuum of magnetic excitations in the Kitaev honeycomb iridate D$_3$LiIr$_2$O$_6$

Inelastic neutron scattering (INS) measurements of powder D$_3(^{7}$Li)($^{193}$Ir)$_2$O$_6$ reveal low energy magnetic excitations with a scattering cross section that is broad in $|Q|$ and consistent with a Kitaev spin-liquid (KSL) state. The magnetic nature of the excitation spectrum is demonstrated by longitudinally polarized neutron studies. The total magnetic moment of 1.7(2)$μ_B$/Ir inferred from the total magnetic scattering cross section is consistent with the effective moment inferred from magnetic susceptibility data and expectations for the $J_{\rm eff}=1/2$ single ion state. The rise in the dynamic correlation function ${\cal S}(Q,ω)$ for $\hbarω<5~$meV can be described by a nearest-neighbor Kitaev model with interaction strength $K\approx-13(5)$~meV. Exchange disorder associated with the mixed D-Li site could play an important role in stabilizing the low $T$ quantum fluctuating state.

cond-mat.str-el↗

arXiv2023

Quantum interference in superposed lattices

Charge transport in solids at low temperature reveals a material's mesoscopic properties and structure. Under a magnetic field, Shubnikov-de Haas (SdH) oscillations inform complex quantum transport phenomena that are not limited by the ground state characteristics. Here, in elemental metal Cr with two incommensurately superposed lattices of ions and a spin-density-wave ground state, we reveal that the phases of several low-frequency SdH oscillations in sigma_xx (rho_xx) and sigma_yy (rho_yy) are opposite, contrast with oscillations from normal cyclotron orbits that maintain identical phases. We trace the origin of the low frequency SdH oscillations to quantum interference effects arising from the incommensurate orbits of Cr's superposed reciprocal lattices, and explain the observed pi-phase shift by the reconnection of anisotropic joint open and closed orbits.

cond-mat.str-el↗

arXiv2023

A general framework for enumerating equivalence classes of solutions

When a problem has more than one solution, it is often important, depending on the underlying context, to enumerate (i.e., to list) them all. Even when the enumeration can be done in polynomial delay, that is, spending no more than polynomial time to go from one solution to the next, this can be costly as the number of solutions themselves may be huge, including sometimes exponential. Furthermore, depending on the application, many of these solutions can be considered equivalent. The problem of an efficient enumeration of the equivalence classes or of one representative per class (without generating all the solutions), although identified as a need in many areas, has been addressed only for very few specific cases. In this paper, we provide a general framework that solves this problem in polynomial delay for a wide variety of contexts, including optimization ones that can be addressed by dynamic programming algorithms, and for certain types of equivalence relations between solutions.

cs.DS↗

arXiv2022

Magnetic Excitations and Interactions in the Kitaev Hyperhoneycomb Iridate $β$-Li$_2$IrO$_3$

We present a thorough experimental study of the three-dimensional hyperhoneycomb Kitaev magnet $β$-Li$_2$IrO$_3$, using a combination of inelastic neutron scattering (INS), time-domain THz spectroscopy, and heat capacity measurements. The main results include a massive low-temperature reorganization of the INS spectral weight that evolves into a broad peak centered around 12 meV, and a distinctive peak in the THz data at 2.8(1) meV. A detailed comparison to powder-averaged spin-wave theory calculations reveals that the positions of these two features are controlled by the anisotropic $Γ$ coupling and the Heisenberg exchange $J$, respectively. The refined microscopic spin model places $β$-Li$_2$IrO$_3$ in close proximity to the Kitaev spin liquid phase.

cond-mat.str-el↗

arXiv2020

Monopolar and dipolar relaxation in spin ice Ho$_2$Ti$_2$O$_7$

When degenerate states are separated by large energy barriers, the approach to thermal equilibrium can be slow enough that physical properties are defined by the thermalization process rather than the equilibrium. The exploration of thermalization pushes experimental boundaries and provides refreshing insights into atomic scale correlations and processes that impact steady state dynamics and prospects for realizing solid state quantum entanglement. We present a comprehensive study of magnetic relaxation in Ho$_2$Ti$_2$O$_7$ based on frequency-dependent susceptibility measurements and neutron diffraction studies of the real-time atomic-scale response to field quenches. Covering nearly ten decades in time scales, these experiments uncover two distinct relaxation processes that dominate in different temperature regimes. At low temperatures (0.6K<T<1K) magnetic relaxation is associated with monopole motion along the applied field direction through the spin-ice vacuum. The increase of the relaxation time upon cooling indicates reduced monopole conductivity driven by decreasing monopole concentration and mobility as in a semiconductor. At higher temperatures (1K<T<2K) magnetic relaxation is associated with the reorientation of monopolar bound states as the system approaches the single-spin tunneling regime. Spin fractionalization is thus directly exposed in the relaxation dynamics.

cond-mat.str-el↗

arXiv2020

A continuous metal-insulator transition driven by spin correlations

Metal-insulator transitions involve a mix of charge, spin, and structural degrees of freedom, and when strongly-correlated, can underlay the emergence of exotic quantum states. Mott insulators induced by the opening of a Coulomb gap are an important and well-recognized class of transitions, but insulators purely driven by spin correlations are much less common, as the reduced energy scale often invites competition from other degrees of freedom. Here we demonstrate a clean example of a spin-correlation-driven metal-insulator transition in the all-in-all-out pyrochlore antiferromagnet Cd2Os2O7, where the lattice symmetry is fully preserved by the antiferromagnetism. After the antisymmetric linear magnetoresistance from conductive, ferromagnetic domain walls is carefully removed experimentally, the Hall coefficient of the bulk reveals four Fermi surfaces, two of electron type and two of hole type, sequentially departing the Fermi level with decreasing temperature below the Néel temperature, T_N. Contrary to the common belief of concurrent magnetic and metal-insulator transitions in Cd2Os2O7, the charge gap of a continuous metal-insulator transition opens only at T~10K, well below T_N=227K. The insulating mechanism resolved by the Hall coefficient parallels the Slater picture, but without a folded Brillouin zone, and contrasts sharply with the behavior of Mott insulators and spin density waves, where the electronic gap opens above and at T_N, respectively.

cond-mat.str-el↗

arXiv2019

Approaching the quantum critical point in a highly-correlated all-in-all-out antiferromagnet

Continuous quantum phase transitions involving all-in-all-out (AIAO) antiferromagnetic order in strongly spin-orbit-coupled 5d compounds could give rise to various exotic electronic phases and strongly-coupled quantum critical phenomena. Here we experimentally trace the AIAO spin order in Sm2Ir2O7 using direct resonant x-ray magnetic diffraction techniques under high pressure. The magnetic order is suppressed at a critical pressure Pc=6.30 GPa, while the lattice symmetry remains in the cubic Fd-3m space group across the quantum critical point. Comparing pressure tuning and the chemical series R2Ir2O7 reveals that the suppression of the AIAO order and the approach to the spin-disordered state is characterized by contrasting evolutions of both the pyrochlore lattice constant a and the trigonal distortion x. The former affects the 5d bandwidth, the latter the Ising anisotropy, and as such we posit that the opposite effects of pressure and chemical tuning lead to spin fluctuations with different Ising and Heisenberg character in the quantum critical region. Finally, the observed low-pressure scale of the AIAO quantum phase transition in Sm2Ir2O7 identifies a circumscribed region of P-T space for investigating the putative magnetic Weyl-semimetal state.

cond-mat.str-el↗

arXiv2019

Odd-parity linear magnetoresistance and the planar Hall effect

The phenomena of odd-parity magnetoresistance and the planar Hall effect are deeply entwined with ferromagnetism. The intrinsic magnetization of the ordered state permits these unusual and rarely observed manifestations of Onsager's theorem when time reversal symmetry is broken at zero applied field. Here we study two classes of ferromagnetic materials, rare-earth magnets with high intrinsic coercivity and antiferromagnetic pyrochlores with strongly-pinned ferromagnetic layers at domain walls, which both exhibit odd-parity magnetoresistive behavior. The peculiar angular variation of the response with respect to the relative alignments of the magnetization, magnetic field, and current reveal the two underlying microscopic mechanisms: spin-polarization-dependent scattering of a Zeeman-shifted Fermi surface and magnetoresistance driven by the anomalous velocity physics usually associated with the anomalous Hall effect.

cond-mat.str-el↗