arXiv subjects

Zhengyang Wei

Publications and source records attributed to Zhengyang Wei.

5 recordsLinked to original sources

arXiv2026

IBLTs Measure Before They Decode: Self-Sizing Set Reconciliation for Database Consistency Verification

Cross-system data replication pipelines cannot confirm end-to-end consistency from the local guarantees of each hop, so the two endpoints must be compared directly on a periodic basis. Once the rows of a fixed snapshot are normalized into fingerprints, the task reduces to finding the symmetric difference of the two sets. Set reconciliation encodes both sets with an Invertible Bloom Lookup Table (IBLT), whose communication grows only with the difference cardinality $d$ and is independent of table size, which suits large tables. The capacity of an IBLT, however, must be fixed while $d$ is still unknown. Across 41,603 production reconciliations over 90 days, nonzero $d$ spans about seven orders of magnitude, and no reliable empirical constant exists. We show that the count array of an IBLT has already measured $d$ before decoding. The measurement is in-band: it is carried by the recovery sketch itself and adds no bytes dedicated to estimation. A mapping-aware theorem carries the same construction to Irregular, Rateless, and MET IBLTs. The protocol reads the estimate only after a decoding failure; we prove that the failure-conditioned lower quantile bounds the risk of underestimation, which gives the second-round capacity a configurable success-probability guarantee. On this basis we build a self-sizing protocol: the first round attempts recovery with a small sketch and stops there on success; on failure it reads $d$ and sets the second-round capacity, completing reconciliation in at most two rounds. Against a controlled oracle, communication is 1.29--1.47 times that of a scheme given $d$ in advance; production workload characterization, relational-database replay, and a cross-city KV deployment confirm the end-to-end mechanism. Measurement and recovery share one IBLT and need no separate estimation structure.

cs.DB↗

arXiv2026

LineGraph2Road: Structural Graph Reasoning on Line Graphs for Road Network Extraction

Extracting routable road networks from satellite imagery requires accurate topology recovery beyond pixel-level segmentation. Recent methods decompose the task into keypoint detection and connectivity prediction, but reliably inferring road connectivity under structural ambiguity remains challenging. Overpasses introduce non-planar crossings that can create false shortcuts, and occlusions break visual continuity, requiring long-range structural reasoning. We formulate road extraction as fully unobserved connectivity inference and construct a global but sparse Euclidean graph from detected keypoints to better leverage the long-term context. To improve link prediction, we transform this graph into its line graph and perform reasoning using a Graph Transformer. We propose an end-to-end pipeline that integrates vision-based segmentation, sparse graph construction, and structured inference. Our method explicitly models overpasses and uses topology-preserving vertex extraction to mitigate routing-critical errors. This approach achieves state-of-the-art performance on City-scale, SpaceNet, and Global-scale benchmarks in topology metrics including TOPO-F1 and APLS. In addition to quantitative gains, our method improves reconstruction of multi-level road structures crucial for real-world routing reliability. The code is available at: https://github.com/wzzzzzzy/LineGraph2Road.

cs.CV↗

arXiv2026

UniLab: A Heterogeneous Architecture for Robot RL Beyond GPU-Dominant Paradigms

Simulation-based RL for contemporary robot control is increasingly organized around GPU-resident simulation: physics, rollout collection, and learning are placed on a single GPU-centric execution path. This paradigm has greatly improved training speed, but it has also encouraged a default assumption that efficient training requires physics to reside on the GPU. We revisit this assumption. Our view is that, in simulation-dominated robot control, the essential question is not which processor runs physics, but whether simulation throughput, policy learning, and runtime synchronization form an efficient end-to-end loop. We present UniLab, a heterogeneous CPU-simulation / GPU-learning architecture that decouples CPU-parallel simulation from GPU policy updates through a unified runtime for data movement, buffering, and synchronization. UniLab is implemented as a complete and extensible training system using MuJoCoUni and MotrixSim CPU-batched physics backends, supporting PPO, FastSAC, FlashSAC, and APPO. On representative simulation-based robot control tasks, UniLab improves end-to-end training efficiency by 3--10$\times$ under the same hardware configuration, while reducing dependence on the NVIDIA CUDA-based software stack and supporting cross-platform execution on the Apple macOS platform and the AMD ROCm and Intel XPU accelerator backends. These results show that GPU simulation is an effective path to efficient training, but not a necessary one, broadening the practical system choices available for robot RL training. Project page: https://unilabsim.github.io.

cs.RO↗

arXiv2026

Upper bound of transient growth in accelerating and decelerating wall-driven flows using the Lyapunov method

This work analyzes accelerating and decelerating wall-driven flows by quantifying the upper bound of transient energy growth using a Lyapunov-type approach. By formulating the linearized Navier-Stokes equations as a linear time-varying system and constructing a time-dependent Lyapunov function, we obtain an upper bound on transient energy growth by solving linear matrix inequalities. This Lyapunov method can obtain the upper bound of transient energy growth that closely matches transient growth computed via the singular value decomposition of the state-transition matrix of linear time-varying systems. Our analysis captures that decelerating base flows exhibit significantly larger transient growth compared with accelerating flows. Our Lyapunov method offers the advantages of providing a certificate of uniform stability and an invariant set to bound the solution trajectory.

physics.flu-dyn↗

arXiv2025

Nonlinear input-output analysis of transitional shear flows using small-signal finite-gain $\mathcal{L}_p$ stability

This SSFG Lp stability theorem can predict permissible forcing amplitudes below which a finite nonlinear input-output gain can be maintained. Our analysis employs Linear Matrix Inequalities (LMI) and Sum-of-Squares (SOS) as the primary tools to search for a quadratic Lyapunov function of an unforced nonlinear system. The resulting Lyapunov function can certify the SSFG Lp stability of a nonlinear input-output system. We demonstrate the applicability of the SSFG Lp stability theorem using a nine-mode shear flow model with a random body force. The predicted nonlinear input-output Lp gain is consistent with numerical simulations; the Lp norm of the output from numerical simulations remains bounded by the theoretical prediction from SSFG Lp stability theorem, with the gap between simulated and theoretical bounds narrowing as $p \rightarrow \infty$. The input-output gain obtained from the nonlinear SSFG Lp stability theorem is higher than the linear Lp gain. Both nonlinear Lp gain and linear Lp gain are valid for each $p\in [1,\infty]$, and such generalizability leads to much higher upper bounds on input-output gain than those predicted by linear L2 gain. The SSFG Lp stability theorem requires the input forcing to be smaller than a permissible forcing amplitude to maintain finite input-output gain, which is an inherently nonlinear behavior that cannot be predicted by linear input-output analysis. We also identify such permissible forcing amplitude using numerical simulations and bisection search, where below such forcing amplitude the output norm at any time will be lower than a given threshold value. The permissible forcing amplitude identified from the SSFG Lp stability theorem is conservative but also consistent with that obtained by numerical simulations and bisection search.

physics.flu-dyn↗