arXiv subjects

Lin Su

Publications and source records attributed to Lin Su.

At least 19 recordsLinked to original sources

arXiv2026

Electrostriction in a Bose-Einstein Condensate of Dipolar Molecules

The recent creation of a Bose-Einstein condensate (BEC) of dipolar molecules has opened a new frontier for many-body quantum systems in which dipolar interactions can drive novel self-organization phenomena. Here, we observe electrostriction in a molecular BEC, an elliptical deformation driven by anisotropic dipolar interactions. We use double microwave dressing, involving $\sigma$- and $\pi$-polarized fields, to control non-axially symmetric dipolar interactions. We compare the experimental observations of electrostriction to a model based on an extended Gross-Pitaevskii equation and find excellent agreement in the regime of weak to moderate interactions. Using electrostriction, we demonstrate that the molecular BEC can be torqued by dynamically changing the orientation of the elliptical $\sigma$ microwave field. This provides a route to setting molecular quantum gases into rotation, opening opportunities to probe vorticity, superfluidity, and supersolidity in strongly dipolar matter.

cond-mat.quant-gas↗

arXiv2026

A Quantum Coherence Microscope in the Hubbard Regime

Quantum coherence underlies collective quantum phenomena and emerging quantum technologies. Quantum gas microscopes have transformed quantum simulation by providing projective snapshots of many-body states with single-atom resolution, but spatially resolved measurements of off-diagonal correlations have remained elusive. Here, using the Talbot effect, we introduce a quantum coherence microscope that maps off-diagonal correlations onto site-resolved density signals with near-single-site resolution. We use this technique to locally probe the superfluid-Mott transition in a layer of a three-dimensional optical lattice and to measure coherence beyond nearest neighbors in an engineered potential landscape. By mapping off-diagonal correlations onto density signals through controlled Talbot evolution, this work opens new possibilities for accessing observables beyond the density basis through tailored matter-wave evolution and recapture.

quant-ph↗

arXiv2026

Many-Body Super- and Subradiance in Ordered Atomic Arrays

When quantum emitters couple indistinguishably to light, they can synchronize into a collective light matter system with radiative properties profoundly different from those of independent particles. To date, the resulting collective effects have largely been confined to point like or homogeneous ensembles. Here, we open access to a qualitatively new collective regime by realizing geometrically ordered, spatially extended atom arrays with subwavelength spacing. This establishes a fundamentally new platform in which collective emission is no longer confined to a single Dicke mode but instead emerges from an ordered network of photon mediated interactions. We find that 2D atom arrays undergo strong super and subradiant emission. Despite subwavelength spacing, we achieve site resolved imaging and directly observe the buildup of spatial correlations, demonstrating the transformation of cooperative decay into a strongly correlated many-body process. We observe extensive scaling of superradiance, uncover superradiant revivals, and reveal the ferromagnetic nature of superradiance and the antiferromagnetic nature of subradiance. Our results realize a novel programmable platform for exploring and utilizing dissipative many-body quantum physics, opening new possibilities for photon capture, storage, and atom photon entanglement.

quant-ph↗

arXiv2026

IAPO: Information-Aware Policy Optimization for Token-Efficient Reasoning

Large language models increasingly rely on long chains of thought to improve accuracy, yet such gains come with substantial inference-time costs. We revisit token-efficient post-training and argue that existing sequence-level reward-shaping methods offer limited control over how reasoning effort is allocated across tokens. To bridge the gap, we propose IAPO, an information-theoretic post-training framework that assigns token-wise advantages based on each token's conditional mutual information (MI) with the final answer. This yields an explicit, principled mechanism for identifying informative reasoning steps and suppressing low-utility exploration. We provide a theoretical analysis showing that our IAPO can induce monotonic reductions in reasoning verbosity without harming correctness. Empirically, IAPO consistently improves reasoning accuracy while reducing reasoning length by up to 36%, outperforming existing token-efficient RL methods across various reasoning datasets. Extensive empirical evaluations demonstrate that information-aware advantage shaping is a powerful and general direction for token-efficient post-training. The code is available at https://github.com/YinhanHe123/IAPO.

cs.CL↗

arXiv2026

Semantic Search At LinkedIn

Semantic search with large language models (LLMs) enables retrieval by meaning rather than keyword overlap, but scaling it requires major inference efficiency advances. We present LinkedIn's LLM-based semantic search framework for AI Job Search and AI People Search, combining an LLM relevance judge, embedding-based retrieval, and a compact Small Language Model trained via multi-teacher distillation to jointly optimize relevance and engagement. A prefill-oriented inference architecture co-designed with model pruning, context compression, and text-embedding hybrid interactions boosts ranking throughput by over 75x under a fixed latency constraint while preserving near-teacher-level NDCG, enabling one of the first production LLM-based ranking systems with efficiency comparable to traditional approaches and delivering significant gains in quality and user engagement.

cs.IR↗

arXiv2025

MixLM: High-Throughput and Effective LLM Ranking via Text-Embedding Mix-Interaction

Large language models (LLMs) excel at capturing semantic nuances and therefore show impressive relevance ranking performance in modern recommendation and search systems. However, they suffer from high computational overhead under industrial latency and throughput requirements. In particular, cross-encoder ranking systems often create long context prefill-heavy workloads, as the model has to be presented with the user, query and item information. To this end, we propose MixLM, a novel LLM-based ranking framework, which significantly improves the system throughput via reducing the input context length, while preserving the semantic strength of cross-encoder rankers. In contrast to a standard ranking system where the context is presented to the model as pure text, we propose to use mix-interaction, a mixture of text and embedding tokens to represent the input. Specifically, MixLM encodes all items in the catalog into a few embedding tokens and stores in a nearline cache. The encoded item descriptions are used during online inference, effectively reducing the item length from a few thousand text tokens to a few embedding tokens. We share insights from deploying our MixLM framework to a real-world search application at LinkedIn, including a detailed discussion of our training pipelines, as well as a thorough analysis of our online serving infrastructure optimization. With the same latency budget and on-par relevance metrics, MixLM increased throughput by 10.0x comparing with strong baselines, 75.9x over full-text LLM rerankers. The efficiency gains delivered by MixLM enabled full-traffic deployment of LLM-powered search, which resulted in a significant 0.47\% increase in Daily Active Users (DAU) in online A/B tests.

cs.IR↗

arXiv2025

SemCoT: Accelerating Chain-of-Thought Reasoning through Semantically-Aligned Implicit Tokens

The verbosity of Chain-of-Thought (CoT) reasoning hinders its mass deployment in efficiency-critical applications. Recently, implicit CoT approaches have emerged, which encode reasoning steps within LLM's hidden embeddings (termed ``implicit reasoning'') rather than explicit tokens. This approach accelerates CoT by reducing the reasoning length and bypassing some LLM components. However, existing implicit CoT methods face two significant challenges: (1) they fail to preserve the semantic alignment between the implicit reasoning (when transformed to natural language) and the ground-truth reasoning, resulting in a significant CoT performance degradation, and (2) they focus on reducing the length of the implicit reasoning; however, they neglect the considerable time cost for an LLM to generate one individual implicit reasoning token. To tackle these challenges, we propose a novel semantically-aligned implicit CoT framework termed SemCoT. In particular, for the first challenge, we design a contrastively trained sentence transformer that evaluates semantic alignment between implicit and explicit reasoning, which is used to enforce semantic preservation during implicit reasoning optimization. To address the second challenge, we introduce an efficient implicit reasoning generator by finetuning a lightweight language model using knowledge distillation. This generator is guided by our sentence transformer to distill ground-truth reasoning into semantically aligned implicit reasoning, while also optimizing for accuracy. SemCoT is the first approach that enhances CoT efficiency by jointly optimizing token-level generation speed and preserving semantic alignment with ground-truth reasoning. Extensive experiments demonstrate the superior performance of SemCoT compared to state-of-the-art methods in both efficiency and effectiveness. Our code can be found at https://github.com/YinhanHe123/SemCoT/.

cs.CL↗

arXiv2025

Exploring confinement transitions in $\mathbb{Z}_2$ lattice gauge theories with dipolar atoms beyond one dimension

Confinement of particles into bound states is a phenomenon spanning from high-energy to condensed matter physics, which can be studied in the framework of lattice gauge theories (LGTs). Achieving a comprehensive understanding of confinement continues to pose a major challenge, in particular at finite matter density and in the presence of strong quantum fluctuations. State-of-the-art quantum simulators constitute a promising platform to address this problem. Here we study confinement in coupled chains of $\mathbb{Z}_2$ LGTs coupled to matter fields, that can be mapped to a mixed-dimensional (mixD) XXZ model. We perform large-scale numerical matrix-product state calculations to obtain the phase diagram of this model, in which we uncover striped phases formed by the $\mathbb{Z}_2$ charges that can be melted at finite temperature or by increasing the tunneling rate. To explore this setting experimentally, we use our quantum simulator constituted by erbium atoms with dipolar interactions in a quantum gas microscope, and observe the predicted melting of a stripe phase by increasing the particle tunneling rate. Our explorative experimental studies of thermal deconfinement of $\mathbb{Z}_2$ charges motivate our further theoretical study of the mixD $\mathbb{Z}_2$ LGT, in which we predict a confined meson gas at finite temperature and low magnetization where thermal fluctuations destroy stripes but enable spontaneous commensurate spin order. Overall, we demonstrate that our platform can be used to study confinement in $\mathbb{Z}_2$ LGTs coupled to matter fields, including long-range interactions and beyond one dimension, paving the way for future research of confinement in the quantum many-body regime.

cond-mat.quant-gas↗

arXiv2025

Topological Phase Transitions and Mixed State Order in a Hubbard Quantum Simulator

Topological phase transitions challenge conventional paradigms in many-body physics by separating phases that are locally indistinguishable yet globally distinct. Using a quantum simulator of interacting erbium atoms in an optical lattice, we observe such a transition between one-dimensional crystalline symmetry-protected topological phases (CSPTs). We detect the critical point through non-local string order parameters and reveal its connection to the transition predicted between the Mott and Haldane insulators. Moreover, we demonstrate a striking property: stacking two identical systems eliminates the transition, confirming the predicted group structure and invertibility of SPTs. Finally, while introducing symmetry-breaking disorder also removes the transition, disorder averaging restores it. Consequently, the adjacent phases realize a form of mixed-state quantum order wherein the criticality between them depends on the observer's information. Our results demonstrate how topology and information influence quantum phase transitions, opening the doors to probing novel critical phenomena in programmable quantum matter.

cond-mat.quant-gas↗

arXiv2024

Spin Squeezing with Magnetic Dipoles

Entanglement can improve the measurement precision of quantum sensors beyond the shot noise limit. Neutral atoms, the basis of some of the most precise and accurate optical clocks and interferometers, do not naturally exhibit all-to-all interactions that are traditionally used to generate such entangled states. Instead, we take advantage of the magnetic dipole-dipole interaction native to most neutral atoms to realize spin-squeezed states. We achieve 7.1 dB of metrologically useful squeezing using the finite-range spin exchange interactions in an erbium quantum gas microscope. We further propose and demonstrate that introducing atomic motion protects the spin sector coherence at low fillings, significantly improving the achievable spin squeezing in a 2D dipolar system. This work's protocol can be implemented with most neutral atoms, opening the door to quantum-enhanced metrology in other itinerant dipolar systems, such as molecules or optical lattice clocks, and serves as a novel method for studying itinerant quantum magnetism with long-range interactions.

quant-ph↗

arXiv2024

Fast single atom imaging for optical lattice arrays

High-resolution fluorescence imaging of ultracold atoms and molecules is paramount to performing quantum simulation and computation in optical lattices and tweezers. Imaging durations in these experiments typically range from a millisecond to a second, significantly limiting the cycle time. In this work, we present fast, 2.4 microseconds single-atom imaging in lattices, with 99.4% fidelity - pushing the readout duration of neutral atom quantum platforms to be close to that of superconducting qubit platforms. Additionally, we thoroughly study the performance of accordion lattices. We also demonstrate number-resolved imaging without parity projection, which will facilitate experiments such as the exploration of high-filling phases in the extended Bose-Hubbard models, multi-band or SU(N) Fermi-Hubbard models, and quantum link models.

cond-mat.quant-gas↗

arXiv2023

Dipolar quantum solids emerging in a Hubbard quantum simulator

In quantum mechanical many-body systems, long-range and anisotropic interactions promote rich spatial structure and can lead to quantum frustration, giving rise to a wealth of complex, strongly correlated quantum phases. Long-range interactions play an important role in nature; however, quantum simulations of lattice systems have largely not been able to realize such interactions. A wide range of efforts are underway to explore long-range interacting lattice systems using polar molecules, Rydberg atoms, optical cavities, and magnetic atoms. Here, we realize novel quantum phases in a strongly correlated lattice system with long-range dipolar interactions using ultracold magnetic erbium atoms. As we tune the dipolar interaction to be the dominant energy scale in our system, we observe quantum phase transitions from a superfluid into dipolar quantum solids, which we directly detect using quantum gas microscopy with accordion lattices. Controlling the interaction anisotropy by orienting the dipoles enables us to realize a variety of stripe ordered states. Furthermore, by transitioning non-adiabatically through the strongly correlated regime, we observe the emergence of a range of metastable stripe-ordered states. This work demonstrates that novel strongly correlated quantum phases can be realized using long-range dipolar interaction in optical lattices, opening the door to quantum simulations of a wide range of lattice models with long-range and anisotropic interactions.

cond-mat.quant-gas↗

arXiv2022

Spatial Equity of Micromobility Systems: A Comparison of Shared E-scooters and Station-based Bikeshare in Washington DC

Many cities around the world have introduced dockless micromobility services in recent years and witnessed their rapid growth. Shared dockless e-scooters have the potential to benefit neighborhoods that lack access to station-based bikeshare services, but they may also exacerbate the existing spatial disparities. While some studies have examined the equity of station-based bikeshare systems, limited knowledge is available regarding dockless e-scooter services. This study uses Washington DC as a case study, a city with both dockless e-scooter and station-based bikeshare systems, to conduct equity analysis of the two types of micromobility options. We develop an analytical framework to examine how dockless e-scooter and station-based bikeshare differ on a set of equity-related outcomes (i.e., availability, accessibility, usage, and idle time) across neighborhoods of different socioeconomic categories. Results reveal that dockless e-scooter services increase accessibility to shared micromobility options for disadvantaged neighborhoods but also widen the access gap across neighborhoods. Compared to bikeshare, shared e-scooters have a higher level of spatial accessibility overall due to greater supply; however, the greater supply largely leads to longer average idle time of shared e-scooters rather than a greater number of trips. Finally, it appears that the bikeshare system's equity program effectively promotes low-income use but e-scooters' equity programs do not. Our findings suggest that increasing vehicle supply alone would probably not lead to higher micromobility use in disadvantaged neighborhoods. Instead, policymakers should combine a variety of strategies such as promoting the enrollment of equity programs and reducing access barriers (e.g., smartphone and banking requirements) to micromobility services.

stat.AP↗

arXiv2021

GEM: A General Evaluation Benchmark for Multimodal Tasks

In this paper, we present GEM as a General Evaluation benchmark for Multimodal tasks. Different from existing datasets such as GLUE, SuperGLUE, XGLUE and XTREME that mainly focus on natural language tasks, GEM is a large-scale vision-language benchmark, which consists of GEM-I for image-language tasks and GEM-V for video-language tasks. Comparing with existing multimodal datasets such as MSCOCO and Flicker30K for image-language tasks, YouCook2 and MSR-VTT for video-language tasks, GEM is not only the largest vision-language dataset covering image-language tasks and video-language tasks at the same time, but also labeled in multiple languages. We also provide two baseline models for this benchmark. We will release the dataset, code and baseline models, aiming to advance the development of multilingual multimodal research.

cs.CL↗

arXiv2020

Sub-second production of a quantum degenerate gas

Realizing faster experimental cycle times is important for the future of quantum simulation. The cycle time determines how often the many-body wave-function can be sampled, defining the rate at which information is extracted from the quantum simulation. We demonstrate a system which can produce a Bose-Einstein condensate of $8 \times 10^4$ $^{168}\text{Er}$ atoms with approximately 85% condensate fraction in 800 ms and a degenerate Fermi gas of $^{167}\text{Er}$ in 4 seconds, which are unprecedented times compared to many existing quantum gas experiments. This is accomplished by several novel cooling techniques and a tunable dipole trap. The methods used here for accelerating the production of quantum degenerate gases should be applicable to a variety of atomic species and are promising for expanding the capabilities of quantum simulation.

cond-mat.quant-gas↗

arXiv2020

M3P: Learning Universal Representations via Multitask Multilingual Multimodal Pre-training

We present M3P, a Multitask Multilingual Multimodal Pre-trained model that combines multilingual pre-training and multimodal pre-training into a unified framework via multitask pre-training. Our goal is to learn universal representations that can map objects occurred in different modalities or texts expressed in different languages into a common semantic space. In addition, to explicitly encourage fine-grained alignment between images and non-English languages, we also propose Multimodal Code-switched Training (MCT) to combine monolingual pre-training and multimodal pre-training via a code-switch strategy. Experiments are performed on the multilingual image retrieval task across two benchmark datasets, including MSCOCO and Multi30K. M3P can achieve comparable results for English and new state-of-the-art results for non-English languages.

cs.CL↗

arXiv2020

ImageBERT: Cross-modal Pre-training with Large-scale Weak-supervised Image-Text Data

In this paper, we introduce a new vision-language pre-trained model -- ImageBERT -- for image-text joint embedding. Our model is a Transformer-based model, which takes different modalities as input and models the relationship between them. The model is pre-trained on four tasks simultaneously: Masked Language Modeling (MLM), Masked Object Classification (MOC), Masked Region Feature Regression (MRFR), and Image Text Matching (ITM). To further enhance the pre-training quality, we have collected a Large-scale weAk-supervised Image-Text (LAIT) dataset from Web. We first pre-train the model on this dataset, then conduct a second stage pre-training on Conceptual Captions and SBU Captions. Our experiments show that multi-stage pre-training strategy outperforms single-stage pre-training. We also fine-tune and evaluate our pre-trained ImageBERT model on image retrieval and text retrieval tasks, and achieve new state-of-the-art results on both MSCOCO and Flickr30k datasets.

cs.CV↗

arXiv2018

Probing the Berry Curvature and Fermi Arcs of a Weyl Circuit

The Weyl particle is the massless fermionic cousin of the photon. While no fundamental Weyl particles have been identified, they arise in condensed matter and meta-material systems, where their spinor nature imposes topological constraints on low-energy dispersion and surface properties. Here we demonstrate a topological circuit with Weyl dispersion at low-momentum, realizing a 3D lattice that behaves as a half-flux Hofstadter model in all principal planes. The circuit platform provides access to the complete complex-valued spin-texture of all bulk- and surface- states, thereby revealing not only the presence of Weyl points and the Fermi arcs that connect their surface-projections, but also, for the first time, the Berry curvature distribution through the Brillouin zone and the associated quantized Chiral charge of the Weyl points. This work opens a path to exploration of interacting Weyl physics in superconducting circuits, as well as studies of how manifold topology impacts band topology in three dimensions.

cond-mat.quant-gas↗