arXiv subjects

Xingfu Li

Publications and source records attributed to Xingfu Li.

9 recordsLinked to original sources

arXiv2026

Complexity of Sequence-to-Graph Alignment with Co-Linear Chaining

Sequence alignment is a cornerstone technique in computational biology for assessing similarities and differences among biological sequences. A key variant, sequence-to-graph alignment, plays a crucial role in effectively capturing genetic variations. In this work, we introduce two novel formulations within this framework: the Gap-sensitive Co-Linear Chaining (Gap-CLC) problem and the Co-Linear Chaining with Errors based on Edit Distance (Edit-CLC) problem, and we investigate their computational complexity. We show that solving the Gap-CLC problem in sub-quadratic time is highly unlikely unless the Strong Exponential Time Hypothesis fails -- even when restricted to binary alphabets. Furthermore, we establish that the Edit-CLC problem is NP-hard in the presence of errors within the pan-genome graph. These findings emphasize that incorporating co-linear structures into sequence-to-graph alignment models fails to reduce computational complexity, highlighting that these models remain at least as computationally challenging to solve as those lacking such prior information.

cs.CC↗

arXiv2026

Polynomial-Time Solutions for Longest Common Subsequence Related Problems Between a Sequence and a Pangenome Graph

A pangenome captures the genetic diversity across multiple individuals simultaneously, providing a more comprehensive reference for genome analysis than a single linear genome, which may introduce allele bias. A widely adopted pangenome representation is a node-labeled directed graph, wherein the paths correspond to plausible genomic sequences within a species. Consequently, evaluating sequence-to-pangenome graph similarity constitutes a fundamental task in pangenome construction and analysis. This study explores the Longest Common Subsequence (LCS) problem and three of its variants involving a sequence and a pangenome graph. We present four polynomial-time reductions that transform these LCS-related problems into the longest path problem in a directed acyclic graph (DAG). These reductions demonstrate that all four problems can be solved in polynomial time, establishing their membership in the complexity class P.

cs.DS↗

arXiv2025

The Metacognitive Bottleneck: Japanese Riddles Reveal Fundamental Limits of Machine Insight and Self-Evaluation in Reasoning AI

Benchmark saturation and training-data contamination increasingly obscure whether reported gains in large language models (LLMs) reflect genuine advances in reasoning or familiarity with recurring patterns in benchmark problems. We introduce the NazoNazo Benchmark, a renewable and extensible evaluation dataset derived from Japanese children's riddles that isolates a specific class of reasoning processes: insight-like representational restructuring and metacognitive evaluation. Rather than modeling reasoning in general, these tasks provide a focused test of failure modes that are difficult to detect in standard benchmarks. We curate 201 riddles and establish a human reference on a 120-item subset (n = 126; mean accuracy 52.9%). The benchmark is fully open, low-cost to refresh, and designed for continual evaluation under reduced contamination risk. We evaluate 38 frontier LLMs (2023-2025) under a strict retrieval-free, zero-shot protocol. On the human-comparison subset, non-reasoning models achieve 7.6% accuracy and reasoning-oriented models reach 17.6%, compared with a human mean of 52.9%, although performance varies substantially across models. Beyond accuracy, qualitative analysis of model-generated thought-logs identifies a distinctive failure mode, which we call verification failure: models generate a correct intermediate candidate but fail to endorse it as their final answer. This dissociation between candidate generation and endorsement reveals a metacognitive bottleneck: across the models with usable thought-logs, verification failures account for between 5% and 39% of a model's incorrect answers. By isolating the gap between generation and verification, this work provides a practical framework for diagnosing reasoning reliability and suggests concrete directions for improvement, including better calibration, structured verification, and stopping mechanisms.

cs.AI↗

arXiv2024

Undulation-induced moir\'e superlattices with 1D polarization domains and 1D flat bands in 2D bilayer semiconductors

Two-dimensional (2D) materials have a high F\"oppl-von K\'arm\'an number and can be easily bent, much like a paper, making undulations a novel way to design distinct electronic phases. Through first-principles calculations, we reveal the formation of 1D polarization domains and 1D flat electronic bands by 1D bending modulation to a 2D bilayer semiconductor. Using 1D sinusoidal undulation of a hexagonal boron nitride (hBN) bilayer as an example, we demonstrate how undulation induces nonuniform shear patterns, creating regions with unique local stacking and vertical polarization akin to sliding-induced ferroelectrics observed in twisted moir\'e systems. This sliding-induced polarization is also observed in double-wall BN nanotubes due to curvature differences between inner and outer tubes. Furthermore, undulation generates a shear-induced 1D moir\'e pattern that perturbs electronic states, confining them into 1D quantum-well-like bands with kinetic energy quenched in modulation direction while dispersive in other directions (1D flat bands). This electronic confinement is attributed to modulated shear deformation potential resulting from tangential polarization due to the moir\'e pattern. Thus, bending modulation and interlayer shear offer an alternative avenue, termed "curvytronics", to induce exotic phenomena in 2D bilayer materials.

cond-mat.mes-hall↗

arXiv2023

Bottom-up Integration of TMDCs with Pre-Patterned Device Architectures via Transfer-free Chemical Vapor Deposition

Two-dimensional (2D) transition metal dichalcogenides (TMDCs) remain a topic of immense interest. Specifically, given their low operational switching costs, they find many niche applications in new computing architectures with the promise of continued miniaturization. However, challenges lie in Back End of Line (BEOL) integration temperature and time compliance regarding current requirements for crystal growth. Additionally, deleterious and time-consuming transfer processes and multiple steps involved in channel/contact engineering can cripple device performance. This work demonstrates kinetics-governed in-situ growth regimes (surface or edge growth from gold) of WSe2 and provides a mechanistic understanding of these regimes via energetics across various material interfaces. As a proof-of-concept, field effect transistors (FET) with an in-situ grown WSe2 channel across Au contacts are fabricated, demonstrating a 2D semiconductor transistor via a transfer-free method within the 450-600 C 2h-time window requirement BEOL integration. We leverage directional edge growth to fabricate contacts with robust thickness-dependent Schottky-to-Ohmic behavior. By transitioning between Au and SiO2 growth substrates in situ, this work achieves strain-induced subthreshold swing of 140 mV/decade, relatively high mobility of 107 +- 19 cm2V-1s-1, and robust ON/OFF ratios 10^6 in the fabricated FETs.

cond-mat.mtrl-sci↗

arXiv2021

On the computational complexity of the Steiner $k$-eccentricity

The Steiner $k$-eccentricity of a vertex $v$ of a graph $G$ is the maximum Steiner distance over all $k$-subsets of $V (G)$ which contain $v$. A linear time algorithm for calculating the Steiner $k$-eccentricity of a vertex on block graphs is presented. For general graphs, an $O(n^{ν(G)+1}(n(G) + m(G) + k))$ algorithm is designed, where $ν(G)$ is the cyclomatic number of $G$. A linear algorithm for computing the Steiner $3$-eccentricities of all vertices of a tree is also presented which improves the quadratic algorithm from [Discrete Appl.\ Math.\ 304 (2021) 181--195].

math.CO↗

arXiv2021

On the average Steiner 3-eccentricity of trees

The Steiner $k$-eccentricity of a vertex $v$ of a graph $G$ is the maximum Steiner distance over all $k$-subsets of $V(G)$ which contain $v$. In this paper Steiner $3$-eccentricity is studied on trees. Some general properties of the Steiner $3$-eccentricity of trees are given. A tree transformation which does not increase the average Steiner $3$-eccentricity is given. As its application, several lower and upper bounds for the average Steiner $3$-eccentricity of trees are derived.

math.CO↗

arXiv2020

The Steiner $k$-eccentricity on trees

We study the Steiner $k$-eccentricity on trees, which generalizes the previous one in the paper [X.~Li, G.~Yu, S.~Klavžar, On the average Steiner 3-eccentricity of trees, arXiv:2005.10319, 2020]. To support the algorithm, we achieve much stronger properties for the Steiner $k$-ecc tree than that in the previous paper. Based on this, a linear time algorithm is devised to calculate the Steiner $k$-eccentricity of a vertex in a tree. On the other hand, the lower and upper bounds of the average Steiner $k$-eccentricity index of a tree on order $n$ are established based on a novel technique which is quite different from that in the previous paper but much easier to follow.

math.CO↗

arXiv2014

A 4/3-approximation algorithm for finding a spanning tree to maximize its internal vertices

This paper focuses on finding a spanning tree of a graph to maximize the number of its internal vertices. We present an approximation algorithm for this problem which can achieve a performance ratio $\frac{4}{3}$ on undirected simple graphs. This improves upon the best known approximation algorithm with performance ratio $\frac{5}{3}$ before. Our algorithm benefits from a new observation for bounding the number of internal vertices of a spanning tree, which reveals that a spanning tree of an undirected simple graph has less internal vertices than the edges a maximum path-cycle cover of that graph has. We can also give an example to show that the performance ratio $\frac{4}{3}$ is actually tight for this algorithm. To decide how difficult it is for this problem to be approximated, we show that finding a spanning tree of an undirected simple graph to maximize its internal vertices is Max-SNP-Hard.

cs.DS↗