arXiv subjects

Junhe Wang

Publications and source records attributed to Junhe Wang.

2 recordsLinked to original sources

arXiv2026

Lifting connectivity bottlenecks in superconducting quantum processors via enriched native two-qubit gates

Limited qubit connectivity is a central architectural constraint in superconducting quantum processors, whose planar layouts require additional gates to mediate interactions between distant qubits. Here, we use the AshN control scheme, where rich two-qubit control on every nearest-neighbour pair allows a logical interaction and the required qubit routing to be merged into a single native operation, effectively transforming a sparse hardware graph into a more connected computational architecture. For the benchmark instances studied, the resulting synthesis capability enables reliable execution on constrained one- and two-dimensional lattices, with compiled two-qubit gate counts approaching those of an all-to-all-connected reference. Across seven benchmark circuits on one- and two-dimensional topologies, the AshN-based implementation achieves geometric-mean reductions of $45.2\%$ and $43.7\%$ in two-qubit gate count compared with controlled-Z-based compilation, respectively. Using AshN gates, we prepare an eight-qubit two-excitation Dicke state with a fidelity of $0.736$ and certify its genuine multipartite entanglement using a fully positive-partial-transpose witness, whereas the same witness does not certify entanglement for the CZ-based implementation. The state fidelity and entanglement certification remain robust across the tested lattice configurations, including those with up to three connectivity defects. Our work establishes native-gate engineering as a practical approach to mitigating connectivity constraints.

quant-ph↗

arXiv2024

KiGRAS: Kinematic-Driven Generative Model for Realistic Agent Simulation

Trajectory generation is a pivotal task in autonomous driving. Recent studies have introduced the autoregressive paradigm, leveraging the state transition model to approximate future trajectory distributions. This paradigm closely mirrors the real-world trajectory generation process and has achieved notable success. However, its potential is limited by the ineffective representation of realistic trajectories within the redundant state space. To address this limitation, we propose the Kinematic-Driven Generative Model for Realistic Agent Simulation (KiGRAS). Instead of modeling in the state space, KiGRAS factorizes the driving scene into action probability distributions at each time step, providing a compact space to represent realistic driving patterns. By establishing physical causality from actions (cause) to trajectories (effect) through the kinematic model, KiGRAS eliminates massive redundant trajectories. All states derived from actions in the cause space are constrained to be physically feasible. Furthermore, redundant trajectories representing identical action sequences are mapped to the same representation, reflecting their underlying actions. This approach significantly reduces task complexity and ensures physical feasibility. KiGRAS achieves state-of-the-art performance in Waymo's SimAgents Challenge, ranking first on the WOMD leaderboard with significantly fewer parameters than other models. The video documentation is available at \url{https://kigras-mach.github.io/KiGRAS/}.

cs.RO↗