arXiv subjects

Haoyu Cheng

Publications and source records attributed to Haoyu Cheng.

11 recordsLinked to original sources

arXiv2026

PhysElite: How Far Are LLMs from Solving Olympiad-Level Physics Problems?

Understanding how (multimodal) large language models perform on physics problems requires benchmarks that reflect the difficulty and breadth of expert-level physical reasoning. Existing physics benchmarks remain limited in the following two important ways: (1) short of high-difficulty datasets, and (2) lack of comprehensive coverage of visual forms, knowledge points, and step-by-step solution processes. As a result, model performance on current datasets may not be fully representative of their ability to solve complex physics problems. To address these issues, we present PhysElite, a large-scale bilingual multimodal benchmark for Olympiad-level physics reasoning. PhysElite contains 11,586 Olympiad-tier problems. For each problem, we provide corresponding visual diagrams, step-by-step bilingual Chinese-English solution derivations, and the final answer. We benchmark 18 open-source and closed-source MLLMs, and find that even the strongest model reaches only 33.7% answer accuracy. We additionally conduct step-level process evaluation to diagnose where models fail in the reasoning chain. Our datasets are released at https://huggingface.co/datasets/physelite/PhysElite.

cs.AI↗

arXiv2026

Exact $L^p$ growth rates of Laplace eigenfunctions on the unit disk

We determine the logarithmic growth exponents of the $L^p$ norms, $1\le p\le\infty$, of $L^2$-normalized Laplace eigenfunctions on the unit disk, for both Dirichlet and Neumann boundary conditions. We also prove sharp uniform $L^p$ upper and lower bounds for every $L^2$-normalized Dirichlet eigenfunction and every non-constant Neumann eigenfunction $u_{\lambda}$ on the disk. The proof uses stationary phase estimates and integral estimates for Bessel functions.

math.SP↗

arXiv2026

Hilbert-Geo: Solving Solid Geometric Problems by Neural-Symbolic Reasoning

Geometric problem solving, as a typical multimodal reasoning problem, has attracted much attention and made great progress recently, however most of works focus on plane geometry while usually fail in solid geometry due to 3D spatial diagrams and complex reasoning. To bridge this gap, we introduce Hilbert-Geo, the first unified formal language framework for solid geometry, including an extensive predicate library and a dedicated theorem bank. Based on this framework, we propose a Parse2Reason method containing two steps of first parsing then reasoning. In the parsing step, we utilize conditional description language (CDL), a formalized language composed of predicates specifically designed to construct geometric conditions, to represent both problem description (natural text) and solid diagrams (visual image). In the reasoning step, we leverage those formal CDL and the theorem bank to perform relational inference and algebraic computation, generating strictly correct, verifiable, and human-readable reasoning processes. Notably, our proposed Hilbert-Geo is also applicable to plane geometry. To advance geometric reasoning, we curate two expert-annotated dataset SolidFGeo2k and PlaneFGeo3k, which are furnished with geometric formal language annotations, solutions and answers. Extensive experiments show that our proposed method achieves the state-of-the-art (SOTA) performance 77.3% in SolidFGeo2k and 84.1% in MathVerse-Solid (one small subset in MathVerse dedicated to solid geometry), substantially outperforming leading MLLMs, such as Gemini-2.5-pro (54.2% on SolidFGeo2k) and GPT-5 (62.9% on MathVerse-Solid). In addition, our method achieves the SOTA accuracy 80.2% in PlaneFGeo3k, demonstrating the generality of the Hilbert-Geo in geometric reasoning. Our code and datasets are released at https://github.com/PremiLab-Math/Hilbert-Geo.

cs.CV↗

arXiv2024

Locate n' Rotate: Two-stage Openable Part Detection with Foundation Model Priors

Detecting the openable parts of articulated objects is crucial for downstream applications in intelligent robotics, such as pulling a drawer. This task poses a multitasking challenge due to the necessity of understanding object categories and motion. Most existing methods are either category-specific or trained on specific datasets, lacking generalization to unseen environments and objects. In this paper, we propose a Transformer-based Openable Part Detection (OPD) framework named Multi-feature Openable Part Detection (MOPD) that incorporates perceptual grouping and geometric priors, outperforming previous methods in performance. In the first stage of the framework, we introduce a perceptual grouping feature model that provides perceptual grouping feature priors for openable part detection, enhancing detection results through a cross-attention mechanism. In the second stage, a geometric understanding feature model offers geometric feature priors for predicting motion parameters. Compared to existing methods, our proposed approach shows better performance in both detection and motion parameter prediction. Codes and models are publicly available at https://github.com/lisiqi-zju/MOPD

cs.CV↗

arXiv2023

Scalable telomere-to-telomere assembly for diploid and polyploid genomes with double graph

Despite recent advances in the length and the accuracy of long-read data, building haplotype-resolved genome assemblies from telomere to telomere still requires considerable computational resources. In this study, we present an efficient de novo assembly algorithm that combines multiple sequencing technologies to scale up population-wide telomere-to-telomere assemblies. By utilizing twenty-two human and two plant genomes, we demonstrate that our algorithm is around an order of magnitude cheaper than existing methods, while producing better diploid and haploid assemblies. Notably, our algorithm is the only feasible solution to the haplotype-resolved assembly of polyploid genomes.

q-bio.GN↗

arXiv2023

De novo reconstruction of satellite repeat units from sequence data

Satellite DNA are long tandemly repeating sequences in a genome and may be organized as high-order repeats (HORs). They are enriched in centromeres and are challenging to assemble. Existing algorithms for identifying satellite repeats either require the complete assembly of satellites or only work for simple repeat structures without HORs. Here we describe Satellite Repeat Finder (SRF), a new algorithm for reconstructing satellite repeat units and HORs from accurate reads or assemblies without prior knowledge on repeat structures. Applying SRF to real sequence data, we showed that SRF could reconstruct known satellites in human and well-studied model organisms. We also found satellite repeats are pervasive in various other species, accounting for up to 12% of their genome contents but are often underrepresented in assemblies. With the rapid progress on genome sequencing, SRF will help the annotation of new genomes and the study of satellite DNA evolution even if such repeats are not fully assembled.

q-bio.GN↗

arXiv2021

Metagenome assembly of high-fidelity long reads with hifiasm-meta

Current metagenome assemblers developed for short sequence reads or noisy long readswere not optimized for accurate long reads. Here we describe hifiasm-meta, a new metagenome assembler that exploits the high accuracy of recent data. Evaluated on seven empirical datasets, hifiasm-meta reconstructed tens to hundreds of complete circular bacterial genomes per dataset, consistently outperforming other metagenome assemblers.

q-bio.GN↗

arXiv2021

Robust haplotype-resolved assembly of diploid individuals without parental data

Routine single-sample haplotype-resolved assembly remains an unresolved problem. Here we describe a new algorithm that combines PacBio HiFi reads and Hi-C chromatin interaction data to produce a haplotype-resolved assembly without the sequencing of parents. Applied to human and other vertebrate samples, our algorithm consistently outperforms existing single-sample assembly pipelines and generates assemblies of comparable quality to the best pedigree-based assemblies.

q-bio.GN↗

arXiv2020

Haplotype-resolved de novo assembly with phased assembly graphs

Haplotype-resolved de novo assembly is the ultimate solution to the study of sequence variations in a genome. However, existing algorithms either collapse heterozygous alleles into one consensus copy or fail to cleanly separate the haplotypes to produce high-quality phased assemblies. Here we describe hifiasm, a new de novo assembler that takes advantage of long high-fidelity sequence reads to faithfully represent the haplotype information in a phased assembly graph. Unlike other graph-based assemblers that only aim to maintain the contiguity of one haplotype, hifiasm strives to preserve the contiguity of all haplotypes. This feature enables the development of a graph trio binning algorithm that greatly advances over standard trio binning. On three human and five non-human datasets, including California redwood with a $\sim$30-gigabase hexaploid genome, we show that hifiasm frequently delivers better assemblies than existing tools and consistently outperforms others on haplotype-resolved assembly.

q-bio.GN↗

arXiv2019

LVMapper: A Large-variance Clone Detector Using Sequencing Alignment Approach

To detect large-variance code clones (i.e. clones with relatively more differences) in large-scale code repositories is difficult because most current tools can only detect almost identical or very similar clones. It will make promotion and changes to some software applications such as bug detection, code completion, software analysis, etc. Recently, CCAligner made an attempt to detect clones with relatively concentrated modifications called large-gap clones. Our contribution is to develop a novel and effective detection approach of large-variance clones to more general cases for not only the concentrated code modifications but also the scattered code modifications. A detector named LVMapper is proposed, borrowing and changing the approach of sequencing alignment in bioinformatics which can find two similar sequences with more differences. The ability of LVMapper was tested on both self-synthetic datasets and real cases, and the results show substantial improvement in detecting large-variance clones compared with other state-of-the-art tools including CCAligner. Furthermore, our new tool also presents good recall and precision for general Type-1, Type-2 and Type-3 clones on the widely used benchmarking dataset, BigCloneBench.

cs.SE↗

arXiv2017

FMtree: A fast locating algorithm of FM-indexes for genomic data

Motivation: As a fundamental task in bioinformatics, searching for massive short patterns over a long text is widely accelerated by various compressed full-text indexes. These indexes are able to provide similar searching functionalities to classical indexes, e.g., suffix trees and suffix arrays, while requiring less space. For genomic data, a well-known family of compressed full-text index, called FM-indexes, presents unmatched performance in practice. One major drawback of FM-indexes is that their locating operations, which report all occurrence positions of patterns in a given text, are particularly slow, especially for the patterns with many occurrences. Results: In this paper, we introduce a novel locating algorithm, FMtree, to fast retrieve all occurrence positions of any pattern via FM-indexes. When searching for a pattern over a given text, FMtree organizes the search space of the locating operation into a conceptual quadtree. As a result, multiple occurrence positions of this pattern can be retrieved simultaneously by traversing the quadtree. Compared with the existing locating algorithms, our tree-based algorithm reduces large numbers of redundant operations and presents better data locality. Experimental results show that FMtree is usually one order of magnitude faster than the state-of-the-art algorithms, and still memory-efficient.

cs.DS↗