arXiv subjects

Jianting Liu

Publications and source records attributed to Jianting Liu.

3 recordsLinked to original sources

arXiv2026

Temporal-Aware Heterogeneous Graph Reasoning with Multi-View Fusion for Temporal Question Answering

Question Answering over Temporal Knowledge Graphs (TKGQA) has attracted growing interest for handling time-sensitive queries. However, existing methods still struggle with: 1) weak incorporation of temporal constraints in question representation, causing biased reasoning; 2) limited ability to perform explicit multi-hop reasoning; and 3) suboptimal fusion of language and graph representations. We propose a novel framework with temporal-aware question encoding, multi-hop graph reasoning, and multi-view heterogeneous information fusion. Specifically, our approach introduces: 1) a constraint-aware question representation that combines semantic cues from language models with temporal entity dynamics; 2) a temporal-aware graph neural network for explicit multi-hop reasoning via time-aware message passing; and 3) a multi-view attention mechanism for more effective fusion of question context and temporal graph knowledge. Experiments on multiple TKGQA benchmarks demonstrate consistent improvements over multiple baselines.

cs.CL↗

arXiv2026

DiT-HC: Enabling Efficient Training of Visual Generation Model DiT on HPC-oriented CPU Cluster

Generative foundation models have become an important tool for data reconstruction and simulation in scientific computing, showing a tight integration with traditional numerical simulations. At the same time, with the development of new hardware features, such as matrix acceleration units and high-bandwidth memory, CPU-based clusters offer promising opportunities to accelerate and scale such models, facilitating the unification of artificial intelligence and scientific computing. We present DiT-HC, the first system to train and scale the generative model DiT on a next-generation HPC CPU cluster. DiT-HC introduces three key techniques: (1) communication-free tensor parallelism (CFTP) with AutoMem for automated memory-aware dataflow, (2) HCOps, a suite of optimized GEMM and operator kernels leveraging vector and matrix acceleration units, and (3) a custom MPI backend that overlaps computation, communication, and memory movement. Experiments show 8.2 to 87.7 times speedups over native or public CPU libraries and 90.6% weak scaling efficiency on 256 nodes. These results demonstrate the feasibility of large-scale generative model training on CPU clusters and provide new insights for future HPC-AI co-design.

cs.DC↗

arXiv2023

Topological materials or structures: Origin of higher-order topological states

Higher-order topological states (HOTS) have been extensively investigated in classical wave systems. They do not exist in the band gaps of infinite materials, while exhibit as the in-gap localized modes once the infinite materials are truncated to be the finite structures. Here, we will experimentally reveal the origin of HOTSs in acoustic systems. We design the hollow acoustic structures exclusively composed of hinge and corner resonators. We present the experimental proof that, despite the lack of surfaces and bulks, the hollow acoustic structures can still support the topologically protected hinge and corner states, indicating that the local configurations of boundaries are the sources for the generation of HOTSs. We then get the composite structures by assembling the 2D and 3D topological hollow acoustic structures, and experimentally observe the robust HOTSs in them. Our results provide a fundamental perspective on HOTSs in periodic structures, and we foresee that these findings will pave the way toward designing new topological devices for energy recovering, information processing, non-destructive testing and acoustic sensing.

cond-mat.mtrl-sci↗