Searcharxiv⌕ Search

arXiv subjects

Hongtao Xue

Publications and source records attributed to Hongtao Xue.

3 recordsLinked to original sources

Circuit compilation in the quspect quantum SDK: selecting routings after coupling-constrained resynthesis

quspect is an open-source quantum computing SDK that provides circuit construction, C++ state-vector simulation with adjoint gradients, support for variational and combinatorial-optimization algorithms, Hamiltonian simulation, control-pulse envelopes and backends for several cloud quantum processors. Here we describe and evaluate its circuit compiler. Routing and synthesis interact: inserting a SWAP can change the circuit structure that later optimization exploits. At its highest optimization level, the compiler evaluates a screened set of routing candidates by their cost after coupling-constrained resynthesis. The pipeline combines parity-state A* search for small CNOT-phase regions, heuristic synthesis for larger regions, and numerical three-qubit instantiation with a per-block 10^-9 check. For a fixed phase polynomial, A* certifies the minimum CX count when it reaches a goal within budget. On an evaluation set of 55 circuits from 41 arXiv papers on five topologies, the compiler uses 0.616 times as many two-qubit gates as Qiskit O3 (95% circuit-cluster bootstrap interval 0.555-0.679) at 0.592 times its depth. BQSKit level 2 reaches 0.853 and tket 0.930. On a 17-qubit superconducting processor, the compiled circuits had the fewest CZ gates in all six scenarios of the main run. Ablations assess the contributions within the integrated pipeline.

quant-ph↗

A Pre-trained Data Deduplication Model based on Active Learning

In the era of big data, the issue of data quality has become increasingly prominent. One of the main challenges is the problem of duplicate data, which can arise from repeated entry or the merging of multiple data sources. These "dirty data" problems can significantly limit the effective application of big data. To address the issue of data deduplication, we propose a pre-trained deduplication model based on active learning, which is the first work that utilizes active learning to address the problem of deduplication at the semantic level. The model is built on a pre-trained Transformer and fine-tuned to solve the deduplication problem as a sequence to classification task, which firstly integrate the transformer with active learning into an end-to-end architecture to select the most valuable data for deduplication model training, and also firstly employ the R-Drop method to perform data augmentation on each round of labeled data, which can reduce the cost of manual labeling and improve the model's performance. Experimental results demonstrate that our proposed model outperforms previous state-of-the-art (SOTA) for deduplicated data identification, achieving up to a 28% improvement in Recall score on benchmark datasets.

cs.LG↗

Chalcogenide Perovskites- an Emerging Class of Ionic Semiconductors

We report the synthesis and characterization of a novel class of ionic semiconductor materials- inorganic chalcogenide perovskites. Several different compounds including BaZrS3, CaZrS3, SrTiS3 and SrZrS3 were synthesized by high temperature sulfurization of their oxide counterparts. Their crystal structures were identified by XRD and composition by EDX. UV-vis and photoluminescence measurements confirmed that they are direct gap semiconductors with band gap values consistent with theoretical predictions. By adopting an anion alloying approach, we demonstrate widely tunable band gap from 1.73 eV to 2.87 eV. These strongly ionic semiconductors provide a new avenue for engineering the semiconducting properties for applications such as energy harvesting, solid state lighting and sensing.

cond-mat.mtrl-sci↗