Multi-vector retrieval models (e.g., ColBERTv2) offer high retrieval accuracy but suffer from efficiency problems at scale. Recently, several methods have been developed to enhance the efficiency of multi-vector retrieval. On one hand, the state-of-the-art GPU-based method PLAID-GPU exploits the massive parallelism of the GPU to accelerate computation, but it needs to process a considerable amount (e.g., ten thousand) of document candidates. On the other hand, the state-of-the-art (SOTA) CPU-based method IGP employs a more effective strategy to reduce the number of candidates, but fails to utilize the massive parallelism of the GPU. To get the best of both worlds, we propose GIGP+, a GPU-based method designed to achieve high parallelism and low computational overhead. Our contributions are: (1) an efficient candidate generation kernel that enjoys parallelism while retaining the effectiveness of IGP, (2) a score reordering mechanism that reduces the synchronization overhead and (3) a scheduling strategy for efficient batch processing. Our experiments demonstrate that GIGP+ achieves a 11.0× improvement in query per second (QPS) and reduces latency by 7.6× compared to PLAID-GPU, while maintaining equivalent retrieval accuracy. As for cloud pricing, GIGP+ delivers a 2.3× improvement in queries per dollar over SOTA CPU-based solutions.

Neural embedding models are extensively employed in retrieval applications, including passage retrieval, question answering, and web search. In particular, multi-vector models (e.g., ColBERTv2), which represent a document as multiple embedding vectors, have been demonstrated to achieve superior retrieval quality. Nevertheless, these models incur significant overhead at the retrieval time due to the massive amount of embedding vectors. Several promising proposals (e.g., PLAID, DESSERT, EMVB, and MUVERA) have been made to optimize the query latency. To yield high recall, these methods need to generate a considerable amount (e.g., ten thousands) of document candidates, rendering both the candidate generation phase and the refinement phase inefficient. In this paper, we propose a high-quality candidate generation technique that produces only hundreds of candidates yet achieves high recall. Specifically, we develop an incremental next-similar retrieval technique for a proximity graph index in order to facilitate high-quality candidate generation. Our experiments on real datasets show that our proposed method IGP achieves 2x-3x query throughput compared to existing methods at the same accuracy level.

Embedding models represent users and products as high-dimensional embedding vectors and are widely used for recommendation. In this paper, we study the reverse$k-\mathbf{ranks}$query, which finds the users that are the most interested in a product and has many applications including product promotion, targeted advertising, and market analysis. As reverse$k-\mathbf{ranks}$solutions for low dimensionality (e.g., trees) fail for the high-dimensional embeddings generated by embedding models, we propose the QSRP framework. QSRP precomputes the score table between all user and product embeddings to facilitate pruning and refinement at query time. As the score table is usually large, QSRP samples some of its columns as the index to fit in memory. To tackle the problem that naive uniform sampling results in poor pruning effect, we propose query-aware sampling, which conducts sampling by explicitly maximizing the pruning effect for a set of sample queries. Moreover, we introduce regression-based pruning, which fits cheap linear functions to predict the bounds used for pruning. We also design techniques to build the index with limited memory, reduce index building time, and handle updates. We evaluate QSRP under various configurations and compare with state-of-the-art baselines. The results show that QSRP achieves shorter query time than the baselines in all cases, and the speedup is usually over 100x.

Message-passing interface (MPI) programs are non-deterministic and challenging to ensure correctness. The introduction of one-sided communications makes the problem of non-determinism more severe for MPI programs. This paper reports our in-progress work of symbolic execution for the MPI programs with one-sided communications. Our approach can cover the non-determinism caused by the inputs, one-sided communication, and message- passing operations of MPI programs. The preliminary evaluation's results indicate the promising of our approach.

In this article, a hypersonic target electromagnetic (EM) scattering echo model combined with the inhomogeneous zonal medium model (IZMM) and the classical scattering center model (SCM) is proposed with a distributed satelliteborne array radar as the detection platform. A parallel physical optics (PO) method is used for multiview inverse synthetic aperture radar (ISAR) imaging of a moving hypersonic target covered with plasma sheath based on the analysis of high-resolution range profile in the S-X ultrawideband range, reconstructing 2-D EM scattering echo data (the target) and motion compensation. The results show that the surface of the inhomogeneous plasma sheath flow field is an excitation layer with random and irregular fluctuation characteristics, which increases the false scattering centroid of the 1-D range profile of the hypersonic target and can interfere with and disrupt the radar localization of the target along the radial direction. In addition, shallow scattering of EM waves occurs in the plasma sheath, and the average signal intensity of the target can gradually reduce from 0.5$\times \,\,10^{-5}$at 60 km and 20 Ma to 0.1$\times \,\,10^{-5}$at 30 km and 20 Ma, with a fivefold weakening of the overall scattered echo signal. In particular, the faster the hypersonic target travels at 30-km altitude, the weaker the imaged scattered echo signal becomes, with the average intensity of the imaged signal weakening by approximately threefold from 15 to 25 Ma. This study provides considerable technical support and data assurance to establish a synthetic aperture radar (SAR) automatic target recognition (ATR) database, and the findings of this study can be used as a reference for fine-structure feature analysis of hypersonic targets for feature extraction and the classification and identification of targets.