arXiv subjects

Yufei Ge

Publications and source records attributed to Yufei Ge.

14 recordsLinked to original sources

arXiv2026

RiskWorld: Object-Centric Latent World Modeling for Autonomous Driving Risk Identification

Autonomous driving risk identification aims to determine which observed object is likely to become safety-critical to the ego vehicle. Existing approaches typically predict scene-level accidents, infer risk objects indirectly from ego behavior, or apply geometric checks after trajectory forecasting, without directly using predicted ego--object relations for risk-source localization. We propose RiskWorld, an object-centric latent world model that identifies risk from the imagined evolution of each candidate relative to the ego vehicle. RiskWorld combines pretrained predictive video representations with structured ego--object histories, contextualizes observed interactions, and rolls relation-aware object states into the future using RSSM-style latent dynamics. It decodes the rollout into object-level risk scores, supported by auxiliary future-relation and temporal-risk predictions. Inference uses only observations up to the current time, while logged futures provide training supervision. On RiskBench, RiskWorld achieves the best overall F1 of 63.0\% and the lowest false-alarm rate of 2.1\%. Further analyses show that the learned rollout captures the evolution of object-level risk before critical events, while RiskWorld's selections preserve planning-critical information under filtered observation.

cs.RO↗

arXiv2026

Self-induced crystalline fluctuation spin-glass state in Mn7C3 binary compounds

Crystalline spin glasses are attractive compounds owing to their unique nature and applications. Here, we synthesised a bulk Pnma-type Mn7C3 spin glass by a high-temperature, high-pressure method. Experimental characterisation including X-ray diffraction and magnetic susceptibility measurements demonstrated that the compound has a triangular Ising-model-based structure, high freezing temperature of 37.4 K, and novel competition mechanism. Theoretical calculations and simulations revealed that the triangular Mn units are spontaneously frustrated and bridge neighbouring Mn units via polarised C atoms and messenger Mn atoms. Triangular C units each share one electron within a three-pronged electron cloud. This electron is the direct cause of frustration and competition in Mn7C3. The competition within the triangular Mn units suggests that the possible magnetic configurations are highly degenerate and that the Mn7C3 spin glass has high robustness. This work introduces a new family of spin glasses with ordered microgeometries that drive electronic structure disorder, and an application-friendly spin-glass material for use in fields like high-efficiency hardware and algorithm design in artificial intelligence.

cond-mat.mtrl-sci↗

arXiv2026

End-to-end Conditional Diffusion for Realistic and Controllable Visual Traffic Scenario Generation

Generating closed-loop traffic scenarios that are both realistic and controllable is crucial for evaluating autonomous driving systems, especially under rare safety-critical interactions. Existing learning-based methods often struggle to balance controllability and realism, offering either limited fine-grained control over traffic behavior or controllable scenarios at the expense of behavioral plausibility. This paper presents E2E-CDiff, an end-to-end conditional diffusion framework for controllable and realistic scenario generation. Conditioned on front-view visual observations, E2E-CDiff jointly denoises future motion states and executable low-level controls for route-interacting background vehicles. This unified state-action generation mitigates the planning-control mismatch in conventional two-stage trajectory-then-controller pipelines. Differentiable guidance further regulates speed, enforces drivable-area compliance, and supports collision-avoidance or collision-seeking behaviors, enabling both naturalistic and safety-critical scenario generation. Experiments on Bench2Drive show that E2E-CDiff achieves a favorable controllability-realism trade-off compared with representative reinforcement- and imitation-learning baselines, while its collision-guided variant induces challenging interactions across multiple autonomous driving systems. E2E-CDiff also performs competitively as a learning-based ego planner, demonstrating the generality of end-to-end state-action diffusion.

cs.RO↗

arXiv2024

Efficient and Robust Parameter Optimization of the Unitary Coupled-Cluster Ansatz

The variational quantum eigensolver (VQE) framework has been instrumental in advancing near-term quantum algorithms. However, parameter optimization remains a significant bottleneck for VQE, requiring a large number of measurements for successful algorithm execution. In this paper, we propose sequential optimization with approximate parabola (SOAP) as an efficient and robust optimizer specifically designed for parameter optimization of the unitary coupled-cluster ansatz on quantum computers. SOAP leverages sequential optimization and approximates the energy landscape as quadratic functions, minimizing the number of energy evaluations required to optimize each parameter. To capture parameter correlations, SOAP incorporates the average direction from the previous iteration into the optimization direction set. Numerical benchmark studies on molecular systems demonstrate that SOAP achieves significantly faster convergence and greater robustness to noise compared to traditional optimization methods. Furthermore, numerical simulations up to 20 qubits reveal that SOAP scales well with the number of parameters in the ansatz. The exceptional performance of SOAP is further validated through experiments on a superconducting quantum computer using a 2-qubit model system.

physics.chem-ph↗

arXiv2023

Roles of non-local electron-phonon coupling on the electrical conductivity and Seebeck coefficient: a TD-DMRG study

Organic molecular materials are potential high-performance thermoelectric materials. Theoretical understanding of thermoelectric conversion in organic materials is essential for rational molecular design for efficient energy conversion materials. In organic materials, non-local electron-phonon coupling plays a vital role in charge transport and leads to complex transport mechanisms, including hopping, phonon-assist current, band, and transient localization. In this work, based on time-dependent density matrix renormalization group (TD-DMRG) method, we look at the role of non-local electron-phonon coupling on the thermoelectric conversion in organic systems described by Holstein-Peierls model. We calculate the current-current correlation and the heat current-current correlation functions. We find that (i) non-local electron-phonon coupling has very weak influence on Seebeck coefficient because of the cancellation between heat current-current correlation function and the current-current correlation function, but it has strong influence on conductivity through dynamic disorders; (ii) doping concentration has a strong influence on both conductivity and Seebeck coefficient significantly, and the optimal doping ratio to reach the highest power factor is 3%-10% fillings.

cond-mat.mtrl-sci↗

arXiv2021

Learning from Multiple Annotators by Incorporating Instance Features

Learning from multiple annotators aims to induce a high-quality classifier from training instances, where each of them is associated with a set of possibly noisy labels provided by multiple annotators under the influence of their varying abilities and own biases. In modeling the probability transition process from latent true labels to observed labels, most existing methods adopt class-level confusion matrices of annotators that observed labels do not depend on the instance features, just determined by the true labels. It may limit the performance that the classifier can achieve. In this work, we propose the noise transition matrix, which incorporates the influence of instance features on annotators' performance based on confusion matrices. Furthermore, we propose a simple yet effective learning framework, which consists of a classifier module and a noise transition matrix module in a unified neural network architecture. Experimental results demonstrate the superiority of our method in comparison with state-of-the-art methods.

cs.LG↗

arXiv2013

One-dimensional array of ion chains coupled to an optical cavity

We present a novel hybrid system where an optical cavity is integrated with a microfabricated planar-electrode ion trap. The trap electrodes produce a tunable periodic potential allowing the trapping of up to 50 separate ion chains spaced by 160 $\mu$m along the cavity axis. Each chain can contain up to 20 individually addressable Yb\textsuperscript{+} ions coupled to the cavity mode. We demonstrate deterministic distribution of ions between the sites of the electrostatic periodic potential and control of the ion-cavity coupling. The measured strength of this coupling should allow access to the strong collective coupling regime with $\lesssim$10 ions. The optical cavity could serve as a quantum information bus between ions or be used to generate a strong wavelength-scale periodic optical potential.

quant-ph↗

arXiv2012

Laser-induced charging of microfabricated ion traps

Electrical charging of metal surfaces due to photoelectric generation of carriers is of concern in trapped ion quantum computation systems, due to the high sensitivity of the ions' motional quantum states to deformation of the trapping potential. The charging induced by typical laser frequencies involved in doppler cooling and quantum control is studied here, with microfabricated surface electrode traps made of aluminum, copper, and gold, operated at 6 K with a single Sr$^+$ ion trapped 100 $μ$m above the trap surface. The lasers used are at 370, 405, 460, and 674 nm, and the typical photon flux at the trap is 10$^{14}$ photons/cm$^2$/sec. Charging is detected by monitoring the ion's micromotion signal, which is related to the number of charges created on the trap. A wavelength and material dependence of the charging behavior is observed: lasers at lower wavelengths cause more charging, and aluminum exhibits more charging than copper or gold. We describe the charging dynamic based on a rate equation approach.

cond-mat.mtrl-sci↗

arXiv2011

A microfabricated surface ion trap on a high-finesse optical mirror

A novel approach to optics integration in ion traps is demonstrated based on a surface electrode ion trap that is microfabricated on top of a dielectric mirror. Additional optical losses due to fabrication are found to be as low as 80 ppm for light at 422 nm. The integrated mirror is used to demonstrate light collection from, and imaging of, a single 88 Sr+ ion trapped $169\pm4 μ$m above the mirror.

quant-ph↗

arXiv2010

Superconducting microfabricated ion traps

We fabricate superconducting ion traps with niobium and niobium nitride and trap single 88Sr ions at cryogenic temperatures. The superconducting transition is verified and characterized by measuring the resistance and critical current using a 4-wire measurement on the trap structure, and observing change in the rf reflection. The lowest observed heating rate is 2.1(3) quanta/sec at 800 kHz at 6 K and shows no significant change across the superconducting transition, suggesting that anomalous heating is primarily caused by noise sources on the surface. This demonstration of superconducting ion traps opens up possibilities for integrating trapped ions and molecular ions with superconducting devices.

quant-ph↗

arXiv2010

Demonstration of a quantum logic gate in a cryogenic surface-electrode ion trap

We demonstrate quantum control techniques for a single trapped ion in a cryogenic, surface-electrode trap. A narrow optical transition of Sr+ along with the ground and first excited motional states of the harmonic trapping potential form a two-qubit system. The optical qubit transition is susceptible to magnetic field fluctuations, which we stabilize with a simple and compact method using superconducting rings. Decoherence of the motional qubit is suppressed by the cryogenic environment. AC Stark shift correction is accomplished by controlling the laser phase in the pulse sequencer, eliminating the need for an additional laser. Quantum process tomography is implemented on atomic and motional states using conditional pulse sequences. With these techniques we demonstrate a Cirac-Zoller Controlled-NOT gate in a single ion with a mean fidelity of 91(1)%.

quant-ph↗

arXiv2008

Individual addressing of ions using magnetic field gradients in a surface-electrode ion trap

Dense array of ions in microfabricated traps represent one possible way to scale up ion trap quantum computing. The ability to address individual ions is an important component of such a scheme. We demonstrate individual addressing of trapped ions in a microfabricated surface-electrode trap using a magnetic field gradient generated on-chip. A frequency splitting of 310(2) kHz for two ions separated by 5 um is achieved. Selective single qubit operations are performed on one of two trapped ions with an average of 2.2+/-1.0% crosstalk. Coherence time as measured by the spin-echo technique is unaffected by the field gradient.

quant-ph↗

arXiv2008

Temperature Dependence of Electric Field Noise Above Gold Surfaces

Electric field noise from fluctuating patch potentials is a significant problem for a broad range of precision experiments, including trapped ion quantum computation and single spin detection. Recent results demonstrated strong suppression of this noise by cryogenic cooling, suggesting an underlying thermal process. We present measurements characterizing the temperature and frequency dependence of the noise from 7 to 100 K, using a single Sr+ ion trapped 75 um above the surface of a gold plated surface electrode ion trap. The noise amplitude is observed to have an approximate 1/f spectrum around 1 MHz, and grows rapidly with temperature as T^beta for beta from 2 to 4. The data are consistent with microfabricated cantilever measurements of non-contact friction but do not extrapolate to the DC measurements with neutral atoms or contact potential probes.

quant-ph↗

arXiv2007

Suppression of Heating Rates in Cryogenic Surface-Electrode Ion Traps

Dense arrays of trapped ions provide one way of scaling up ion trap quantum information processing. However, miniaturization of ion traps is currently limited by sharply increasing motional state decoherence at sub-100 um ion-electrode distances. We characterize heating rates in cryogenically cooled surface-electrode traps, with characteristic sizes in 75 um to 150 um range. Upon cooling to 6 K, the measured rates are suppressed by 7 orders of magnitude, two orders of magnitude below previously published data of similarly sized traps operated at room temperature. The observed noise depends strongly on fabrication process, which suggests further improvements are possible.

quant-ph↗