arXiv subjects

Ru Chen

Publications and source records attributed to Ru Chen.

14 recordsLinked to original sources

arXiv2026

From What to How: Bridging User Requirements with Software Development Using Large Language Models

Recently, large language models (LLMs) are extensively utilized to enhance development efficiency, leading to numerous benchmarks for evaluating their performance. However, these benchmarks predominantly focus on implementation, overlooking the equally critical aspect of software design. This gap raises two pivotal questions: (1) Can LLMs handle software design? (2) Can LLMs write code following the specific designs? To investigate these questions, this paper proposes DesBench, a design-aware benchmark for evaluating LLMs on three software design-related tasks: design-aware code generation, object-oriented modeling, and the design of acceptance test cases. DesBench comprises 30 manually crafted Java projects that include requirement documents, design models, implementations, and acceptance tests, amounting to a total of 30 design models, 194 Java classes, and 737 test cases. We evaluated seven state-of-the-art LLMs, including three DeepSeek R1, two Qwen2.5, and two GPT models, using DesBench. The results reveal that LLMs remain significantly challenged by the intricacies of software design: (1) For code generation, LLMs struggle to produce correct implementations when provided with only high-level or no designs. (2) In object-oriented modeling, while LLMs can accurately identify objects and classes, they face challenges in defining operations and inter-class relationships. (3) Acceptance test cases generated by LLMs from functional requirements achieve code coverage quality comparable to those written by humans. Our research highlights the current limitations of LLMs in managing software design and calls for further investigation into new design methodologies and languages suitable for LLM-based development.

cs.SE↗

arXiv2025

A Hybrid Approach for EMF Code Generation:Code Templates Meet Large Language Models

Template-based and LLM-based code generation are both key enablers of automated software development. The former provides correctness guarantees but are rigid for complex requirements, whereas LLMs offer high flexibility at the risk of producing faulty code.This paper proposes iEcoreGen, a hybrid approach that integrates Eclipse Modeling Framework (EMF) and LLMs. In EMF, an Ecore model defines a system structure and acts as a blueprint for code-generation.iEcoreGen decomposes requirements to derive operation specifications, uses EMF's template-based generator to produce initial Java code, and serializes specifications into docstrings. LLMs are then invoked to complete and fix unimplemented methods. We assessed iEcoreGen on twenty code-generation tasks across five LLMs. It surpasses LLM-only baselines on pass@k and performs on par with them on compilation@k. An ablation study clarified the contribution of each component of iEcoreGen. Overall, the findings indicate that LLM-enhanced model-driven development is a promising path toward more efficient software automation.

cs.SE↗

arXiv2025

Climate-Adaptive and Cascade-Constrained Machine Learning Prediction for Sea Surface Height under Greenhouse Warming

Machine learning (ML) has achieved remarkable success in climate and marine science. Given that greenhouse warming fundamentally reshapes ocean conditions such as stratification, circulation patterns and eddy activity, evaluating the climate adaptability of the ML models is crucial. While physical constraints have been shown to enhance the performance of ML models, kinetic energy (KE) cascade has not been used as a constraint despite its importance in regulating multi-scale ocean motions. Here we develop two sea surface height (SSH) prediction models (with and without KE cascade constraint) and quantify their climate adaptability at the Kuroshio Extension. Both models exhibit only slight performance degradation under greenhouse warming conditions. Incorporating the KE cascade as a physical constraint significantly improves the model performance, reducing eddy kinetic energy errors by 14.7% in the present climate and 15.9% under greenhouse warming. Additional validations using satellite observations and in the Gulf Stream region further confirm the robustness of the proposed models. Compared with the KE spectrum constraint, both constraints improve the cross-scale transfer and spectrum of KE, but the KE cascade constraint yields larger improvements in the cross-scale transfer. This work presents the first application of the KE cascade as a physical constraint for ML-based ocean state prediction and demonstrates its robust adaptability across climates, offering guidance for the further development of global ML models for both present and future conditions.

physics.ao-ph↗

arXiv2025

Intensification of Oceanic Inverse Energy Cascade Under Global Warming

Kinetic energy (KE) cascade in the turbulent ocean is pivotal in connecting diverse scales of oceanic motions, redistributing energy, and influencing ocean circulation and climate variability. However, its response to global warming remains poorly understood. Using a 24-year satellite altimetry dataset, we identify a pronounced intensification of inverse geostrophic kinetic energy cascade at the sea surface across most ocean regions during 1994-2017, with cascade amplitude increasing by 1% to 2% per decade. This intensification occurs not only in energetic regions but also in expansive quiescent areas. Contributing factors to this intensification of geostrophic KE cascade include enhanced vertical shear of horizontal velocity, deepened mixed layer, strengthened stratification, weakened eddy killing as well as changes in the KE budget. The dominant factors vary across regions. Our findings offer new insights into the ocean's response to global warming and improve understanding of feedback mechanisms for ocean circulation changes.

physics.geo-ph↗

arXiv2024

A Model Is Not Built By A Single Prompt: LLM-Based Domain Modeling With Question Decomposition

Domain modeling, a crucial part of model-driven engineering, demands extensive domain knowledge and experience from engineers. When the system description is highly complicated, the modeling task can become particularly challenging and time-consuming. Large language Models(LLMs) can assist by automatically generating an initial object model from the system description. Although LLMs have demonstrated remarkable code-generation ability, they still struggle with model-generation using a single prompt. In real-world domain modeling, engineers usually decompose complex tasks into easily solvable sub-tasks, significantly controlling complexity and enhancing model quality. Inspired by this, we propose an LLM-based domain modeling approach via question decomposition, similar to developer's modeling process. Following conventional modeling guidelines, we divide the model generation task into several sub-tasks, i.e., class generation, association and aggregation generation, and inheritance generation. For each sub-task, we carefully design the prompt by choosing more efficient query words and providing essential modeling knowledge to unlock the modeling potential of LLMs. To sum up all the sub-tasks solutions, we implemente a proof-of-object tool integrated into the standard Ecore editor that asks LLMs to generate an object model from the system description. We evaluate our approach with 20 systems from different application domains. The preliminary results show that our approach outperforms the single-prompt-based prompt by improving recall values and F1 scores in most systems for modeling the classes, attributes, and relationships.

cs.SE↗

arXiv2018

Idealized Wind-driven Ocean Circulations On Exoplanets

Motivated by the important role of the ocean in the Earth climate system, here we investigate possible scenarios of ocean circulations on exoplanets using a one-layer shallow water ocean model. Specifically, we investigate how planetary rotation rate, wind stress, fluid eddy viscosity and land structure (a closed basin vs. a reentrant channel) influence the pattern and strength of wind-driven ocean circulations. The meridional variation of the Coriolis force, arising from planetary rotation and the spheric shape of the planets, induces the western intensification of ocean circulations. Our simulations confirm that in a closed basin, changes of other factors contribute to only enhancing or weakening the ocean circulations (e.g., as wind stress decreases or fluid eddy viscosity increases, the ocean circulations weaken, and vice versa). In a reentrant channel, just as the Southern Ocean region on the Earth, the ocean pattern is characterized by zonal flows. In the quasi-linear case, the sensitivity of ocean circulations characteristics to these parameters is also interpreted using simple analytical models. This study is the preliminary step for exploring the possible ocean circulations on exoplanets, future work with multi-layer ocean models and fully coupled ocean-atmosphere models are required for studying exoplanetary climates.

astro-ph.EP↗

arXiv2017

Prediction of $\mathrm{TiRhAs}$ as a Dirac Nodal Line Semimetal via First-Principles Calculations

Using first-principles calculations we predict that $\mathrm{TiRhAs}$, a previously synthesized compound, is a Dirac nodal line (DNL) semimetal. The DNL in this compound is found to be protected both by the combination of inversion and time-reversal symmetry, and by a reflection symmetry, in the absence of spin-orbit coupling (SOC). Our calculations show that band velocities associated with the nodal line have a high degree of directional anisotropy, with in-plane velocities $v_\perp$ perpendicular to the nodal line between $1.2-2.8\times10^5$ m/s. The crossings along the DNL are further found to exhibit a prominent and position-dependent tilt along directions perpendicular to the nodal line. We calculate $\mathbb{Z}_2$ indices based on parity eigenvalues at time-reversal invariant momenta and show that $\mathrm{TiRhAs}$ is topological. A tight-binding model fit from our first-principles calculations demonstrates the existence of two-dimensional drumhead surface states on the surface Brillouin zone. Based on the small gapping of the DNL upon inclusion of SOC and the clean Fermi surface free from trivial bands, $\mathrm{TiRhAs}$ is a promising candidate for further studies of the properties of topological semimetals.

cond-mat.mes-hall↗

arXiv2016

Topological Materials Discovery using Electron Filling Constraints

Nodal semimetals, materials systems with nodal-point or -line Fermi surfaces, are much sought after for their novel transport and topological properties. Identification of experimental materials candidates, however, has mainly relied on exhaustive database searches. Here we show how recent studies on the interplay between electron filling and nonsymmorphic space-group symmetries can guide the search for nodal semimetals which are `filling-enforced'. We recast the previously-derived constraints on the allowed electron fillings for band insulators in the 230 space groups into a new form, which enables effective screening of materials candidates based solely on their (1) space group, (2) electron count in the formula unit, and (3) multiplicity of the formula unit. This criterion greatly reduces the computation load for discovering topological materials in a database of previously-synthesized compounds. Of the more than 30,000 entires listed in a few selected nonsymmorphic space groups, our filling criterion alone eliminates 96% of the entries before they are passed on for further analysis. From this guided search, we discover a handful of candidates, including the monoclinic crystals Ca$_2$Pt$_2$Ga, AgF$_2$, and Ca$_2$InOsO$_6$, and the orthorhombic crystal CsHg$_2$. Based on ab initio calculations, we show that these materials have filling-enforced Dirac nodes near the Fermi energy. In addition, we also identify CaPtGa as a promising filling-enforced Dirac-ring semimetal candidate.

cond-mat.mtrl-sci↗

arXiv2014

Z$_2$ topology and superconductivity from symmetry lowering of a 3D Dirac Metal Au$_2$Pb

3D Dirac semi-metals (DSMs) are materials that have massless Dirac electrons and exhibit exotic physical properties It has been suggested that structurally distorting a DSM can create a Topological Insulator (TI), but this has not yet been experimentally verified. Furthermore, quasiparticle excitations known as Majorana Fermions have been theoretically proposed to exist in materials that exhibit superconductivity and topological surface states. Here we show that the cubic Laves phase Au$_2$Pb has a bulk Dirac cone above 100 K that gaps out upon cooling at a structural phase transition to create a topologically non trivial phase that superconducts below 1.2 K. The nontrivial Z$_2$ = -1 invariant in the low temperature phase indicates that Au$_2$Pb in its superconducting state must have topological surface states. These characteristics make Au$_2$Pb a unique platform for studying the transition between bulk Dirac electrons and topological surface states as well as studying the interaction of superconductivity with topological surface states.

cond-mat.supr-con↗

arXiv2014

Correlation between metal-insulator transitions and structural distortions in high-electron-density SrTiO3 quantum wells

The electrical and structural characteristics of SmTiO3/SrTiO3/SmTiO3 and GdTiO3/SrTiO3/GdTiO3 heterostructures are compared. Both types of structures contain narrow SrTiO3 quantum wells, which accommodate a confined, high-density electron gas. As shown previously [Phys. Rev. B 86, 201102(R) (2012)] SrTiO3 quantum wells embedded in GdTiO3 show a metal-to-insulator transition when their thickness is reduced so that they contain only two SrO layers. In contrast, quantum wells embedded in SmTiO3 remain metallic down to a single SrO layer thickness. Symmetry-lowering structural distortions, measured by quantifying the Sr-column displacements, are present in the insulating quantum wells, but are either absent or very weak in all metallic quantum wells, independent of whether they are embedded in SmTiO3 or in GdTiO3. We discuss the role of orthorhombic distortions, orbital ordering, and strong electron correlations in the transition to the insulating state.

cond-mat.str-el↗

arXiv2013

Dimer Mott Insulator in an Oxide Heterostructure

We study the problem of designing an artificial Mott insulator in a correlated oxide heterostructure. We consider the extreme limit of quantum confinement based on ionic discontinuity doping, and argue that a unique dimer Mott insulator can be achieved for the case of a single SrO layer in a GdTiO$_3$ matrix. In the dimer Mott insulator, electrons are localized not to individual atoms but to bonding orbitals on molecular dimers formed across a bilayer of two TiO$_2$ planes, and is analogous to the Mott insulating state of Hubbard ladders, studied in the 1990s. We verify the existence of the dimer Mott insulator through both ab initio and model Hamiltonian studies, and find for reasonable values of Hubbard $U$ that it is stable and ferromagnetic with a clear bonding/anti-bonding splitting of order 0.65eV, and a significant smaller Mott gap whose size depends upon $U$. The combined effects of polar discontinuity, strong structural relaxation and electron correlations all contribute to the realization of this unique ground state.

cond-mat.str-el↗

arXiv2012

Ground States of Spin-1/2 Triangular Antiferromagnets in a Magnetic Field

We use a combination of density matrix renormalization group calculations and analytical approaches to study a simplified model for a spatially anisotropic spin-1/2 triangular lattice Heisenberg antiferromagnet: the three-leg triangular spin tube (TST). The model is described by three Heisenberg chains, with exchange constant J, coupled antiferromagnetically with exchange constant J' along the diagonals of the ladder system, with periodic boundary conditions in the shorter direction. We determine the full phase diagram of this model as a function of spatial anisotropy, J'/J, and magnetic field. We find a rich phase diagram, which is dominated by quantum states - phases corresponding to the classical ground state appears only in a small region. Among the dominant phases generated by quantum effects are commensurate and incommensurate coplanar quasi-ordered states, which appear in the vicinity of the isotropic region for most fields, and in the high field region for most anisotropies. The coplanar states, while not classical ground states, can be understood semiclassically. Even more strikingly, the largest region of phase space is occupied by a spin density wave phase, which has incommensurate collinear correlations along the field. This phase has no semiclassical analog, and may be ascribed to enhanced one-dimensional (1d) fluctuations due to frustration. Cutting across the phase diagram is a magnetization plateau, with a gap to all excitations and up up down spin order, with a quantized magnetization equal to 1/3 of the saturation value. In the TST, this plateau extends almost but not quite to the decoupled chains limit. Most of the above features are expected to carry over to the two dimensional system, which we also discuss. At low field, a dimerized phase appears, which is particular to the 1d nature of the TST, and which can be understood from quantum Berry phase arguments.

cond-mat.str-el↗

arXiv2011

Metal-insulator transition in a two-band model for the perovskite nickelates

Motivated by recent Fermi surface and transport measurements on LaNiO3, we study the Mott Metal-Insulator transitions of perovskite nickelates, with the chemical formula RNiO3, where R is a rare-earth ion. We introduce and study a minimal two-band model, which takes into account only the eg bands. In the weak to intermediate correlation limit, a Hartree-Fock analysis predicts charge and spin order consistent with experiments on R=Pr, Nd, driven by Fermi surface nesting. It also produces an interesting semi-metallic electronic state in the model when an ideal cubic structure is assumed. We also study the model in the strong interaction limit, and find that the charge and magnetic order observed in experiment exist only in the presence of very large Hund's coupling, suggesting that additional physics is required to explain the properties of the more insulating nickelates, R=Eu,Lu,Y. Next, we extend our analysis to slabs of finite thickness. In ultra-thin slabs, quantum confinement effects substantially change the nesting properties and the magnetic ordering of the bulk, driving the material to exhibit highly anisotropic transport properties. However, pure confinement alone does not significantly enhance insulating behavior. Based on these results, we discuss the importance of various physical effects, and propose some experiments.

cond-mat.str-el↗

arXiv2010

Landau theory of charge and spin ordering in the nickelates

Guided by experiment and band structure, we introduce and study a phenomenological Landau theory for the unusual charge and spin ordering associated with the Mott transition in the perovskite nickelates, with chemical formula RNiO$_3$, where R=Pr, Nd, Sm, Eu, Ho, Y, and Lu. While the Landau theory has general applicability, we show that for the most conducting materials, R=Pr, Nd, both types of order can be understood in terms of a nearly-nested spin density wave. Furthermore, we argue that in this regime, the charge ordering is reliant upon the orthorhombic symmetry of the sample, and therefore proportional to the magnitude of the orthorhombic distortion. The first order nature of the phase transitions is also explained. We briefly show by example how the theory is readily adapted to modified geometries such as nickelate films.

cond-mat.str-el↗