arXiv subjects

Fengming Liu

Publications and source records attributed to Fengming Liu.

15 recordsLinked to original sources

arXiv2026

Asymmetric Discourse Homogenization and Shared Language Technology: Evidence from Reddit

I document an ideologically asymmetric break in the pre-existing diversification trend of political discourse, emerging around late 2022, using 6 million Reddit comments from two cross-partisan forums, 2019-2025. Conservative users experienced an interruption of their prior diversification trajectory; progressive users showed no comparable change. The asymmetry is consistent across estimation strategies (ITS, DiD, RDiT, propensity-score matching) and temporal aggregations. A daily-frequency permutation test over 2,377 candidate cutoff dates shows the ChatGPT threshold produces an unremarkable estimate (49.8th percentile): the shift builds gradually instead of breaking at a single date. A continuous cumulative LLM index, tracking AI exposure across seven model releases, remains significant under a quadratic trend specification that eliminates the binary estimate. A stayer analysis narrows the mechanism: the homogenization effect disappears when the sample is restricted to authors active throughout the study period, and the stayer confidence interval excludes within-author effects even a tenth the size of the full-sample estimate. The mechanism is most parsimoniously ecological (community-level discursive convergence) rather than individual-level AI adoption, though the data cannot cleanly separate this account from concurrent secular change.

cs.CY↗

arXiv2026

The Translation Tax Is Not a Scalar: A Counterfactual Audit of English-Source Cue Inheritance in Chinese Multilingual Benchmarks

The Translation Tax is often treated as a scalar: translated benchmarks are assumed to inflate scores by preserving English-source cues. We audit this claim in an English-to-Chinese setting. Three proxy estimators disagree: back-translation gaps are small and parser-fragile; cue-score calibration does not predict item-level gains; and a six-model native-control comparison shows model-family rather than uniform benchmark effects. We add a same-item LLM-naturalization stress test that holds answer, options, and content fixed while rewriting Chinese surface form. After correcting a prompt-construction bug, this contrast no longer supports a model-family interaction, but it preserves a residue dose-response: high-residue items benefit while low-residue items do not. The result is not a single Translation Tax, but a set of estimator- and item-dependent validity risks. We release per-cell evidence, the naturalization protocol, human QC, and a reporting checklist for translated multilingual benchmark papers.

cs.CL↗

arXiv2026

Position: Mechanistic Interpretability Must Disclose Identification Assumptions for Causal Claims

Mechanistic interpretability papers increasingly use causal vocabulary: circuits, mediators, causal abstraction, monosemanticity. Such claims require explicit identification assumptions. A purposive audit of 10 papers across four methodological strands finds no dedicated identification-assumptions section and a recurring pattern: validation metrics such as faithfulness, completeness, monosemanticity, alignment, or ablation effects are reported as causal support without stating the assumptions that make them identifying. A two-human-coder audit on $n=30$ reproduces the direction of the main finding: dedicated identification sections are absent, and validation-metric substitution is common, though exact Dim B/D counts are coding-rule sensitive. The paper proposes a disclosure norm: state whether the claim is causal, name the identification strategy, enumerate assumptions, stress at least one, and explain how conclusions shift if assumptions fail. Validation is not identification.

cs.LG↗

arXiv2026

SPATIALALIGN: Aligning Dynamic Spatial Relationships in Video Generation

Most text-to-video (T2V) generators prioritize aesthetic quality, but often ignoring the spatial constraints in the generated videos. In this work, we present SPATIALALIGN, a self-improvement framework that enhances T2V models capabilities to depict Dynamic Spatial Relationships (DSR) specified in text prompts. We present a zeroth-order regularized Direct Preference Optimization (DPO) to fine-tune T2V models towards better alignment with DSR. Specifically, we design DSR-SCORE, a geometry-based metric that quantitatively measures the alignment between generated videos and the specified DSRs in prompts, which is a step forward from prior works that rely on VLM for evaluation. We also conduct a dataset of text-video pairs with diverse DSRs to facilitate the study. Extensive experiments demonstrate that our fine-tuned model significantly out performs the baseline in spatial relationships. The code will be released in Link. Project page: https://fengming001ntu.github.io/SpatialAlign/

cs.CV↗

arXiv2025

Broadband Reflective Elastic Mode Conversion Enabled by a Single Row of Inclined Long-Slits

Broadband longitudinal-to-transverse mode conversion under normal incidence remains difficult to achieve, especially with structurally simple designs. Numerical simulations show that a single periodic row of inclined long-slits near a free surface enables high-efficiency broadband conversion, where the conversion rate exceeds 0.8 across a normalized-frequency range of 0.27 to 0.74, corresponding to a 93.1% relative bandwidth. The broadband response originates from two intrinsic deformation modes of the mass blocks between adjacent inclined slits: a rotational mode and a quadrupole mode. The spectral overlap of these two modes sustains a continuous high-efficiency band. The effect is robust against geometric variations, demonstrating that geometric asymmetry alone offers a minimal yet effective route to broadband elastic-wave mode conversion.

physics.app-ph↗

arXiv2025

MimiTalk: Revolutionizing Qualitative Research with Dual-Agent AI

We present MimiTalk, a dual-agent constitutional AI framework designed for scalable and ethical conversational data collection in social science research. The framework integrates a supervisor model for strategic oversight and a conversational model for question generation. We conducted three studies: Study 1 evaluated usability with 20 participants; Study 2 compared 121 AI interviews to 1,271 human interviews from the MediaSum dataset using NLP metrics and propensity score matching; Study 3 involved 10 interdisciplinary researchers conducting both human and AI interviews, followed by blind thematic analysis. Results across studies indicate that MimiTalk reduces interview anxiety, maintains conversational coherence, and outperforms human interviews in information richness, coherence, and stability. AI interviews elicit technical insights and candid views on sensitive topics, while human interviews better capture cultural and emotional nuances. These findings suggest that dual-agent constitutional AI supports effective human-AI collaboration, enabling replicable, scalable and quality-controlled qualitative research.

cs.HC↗

arXiv2024

Lines of Bound States in the Continuum in a Phononic Crystal Slab

We demonstrate that bound states in the continuum (BICs) form continuous lines along high-symmetry directions of momentum space in a simple phononic crystal slab. Contrary to common sense, these BICs are symmetry-protected (SP) BICs not only at the center of the Brillouin zone (gamma point) but also off the gamma point. We utilize numerical simulations, a group theory method, and a mode expansion method to comprehensively understand the formation of the BICs. It is revealed that these BICs correspond to phase singularity lines of far-field radiation, and the phase winding number can be defined as a topological index. This makes the BICs topologically protected and robust to any parameter variation that maintains periodicity and rotational symmetry. Finally, the generation of the BICs lines is experimentally demonstrated.

cond-mat.mtrl-sci↗

arXiv2024

Rumor Forwarding Prediction Model Based on Uncertain Time Series

The rapid spread of rumors in social media is mainly caused by individual retweets. This paper applies uncertainty time series analysis (UTSA) to analyze a rumor retweeting behavior on Weibo. First, the rumor forwarding is modeled using uncertain time series, including order selection, parameter estimation, residual analysis, uncertainty hypothesis testing and forecast, and the validity of using uncertain time series analysis is further supported by analyzing the characteristics of the residual plot. The experimental results show that the uncertain time series can better predict the next stage of rumor forwarding. The results of the study have important practical significance for rumor management and the management of social media information dissemination.

cs.SI↗

arXiv2023

Complete mode conversion for elastic waves reflected by elastic metamaterial slab with double hexapole resonances

In this study, we investigate the phenomenon of mode conversion in elastic bulk waves using coupled hexapole resonances. A metamaterial slab is proposed enabling the complete conversion between longitudinal and transverse modes. Each unit of the elastic metamaterial slab comprises a pair of scatterers, and their relative direction is oriented at an oblique angle. The interaction between the coupled hexapoles and the background results in oblique displacements, which are responsible for the mode conversion. Moreover, this conversion exhibits a broader frequency range compared to the quadrupole resonance. This innovative design significantly broadens the range of possibilities for developing mode-converting metamaterials.

physics.app-ph↗

arXiv2022

Dynamic Sparse R-CNN

Sparse R-CNN is a recent strong object detection baseline by set prediction on sparse, learnable proposal boxes and proposal features. In this work, we propose to improve Sparse R-CNN with two dynamic designs. First, Sparse R-CNN adopts a one-to-one label assignment scheme, where the Hungarian algorithm is applied to match only one positive sample for each ground truth. Such one-to-one assignment may not be optimal for the matching between the learned proposal boxes and ground truths. To address this problem, we propose dynamic label assignment (DLA) based on the optimal transport algorithm to assign increasing positive samples in the iterative training stages of Sparse R-CNN. We constrain the matching to be gradually looser in the sequential stages as the later stage produces the refined proposals with improved precision. Second, the learned proposal boxes and features remain fixed for different images in the inference process of Sparse R-CNN. Motivated by dynamic convolution, we propose dynamic proposal generation (DPG) to assemble multiple proposal experts dynamically for providing better initial proposal boxes and features for the consecutive training stages. DPG thereby can derive sample-dependent proposal boxes and features for inference. Experiments demonstrate that our method, named Dynamic Sparse R-CNN, can boost the strong Sparse R-CNN baseline with different backbones for object detection. Particularly, Dynamic Sparse R-CNN reaches the state-of-the-art 47.2% AP on the COCO 2017 validation set, surpassing Sparse R-CNN by 2.2% AP with the same ResNet-50 backbone.

cs.CV↗

arXiv2018

High Thermoelectric Performance of Au@Sb2Te3 Heterostructure Derived from the Potential Barriers

The correlated couple of electrical and thermal property is the challenge to realize a substantial leap in thermoelectric materials.Synthesis of semiconductor and metal composites is a significant and versatile design strategy to optimize the thermoelectric performance driven by tailored interface between nanoinclusions and matrix.In this study, we present the simultaneous increase of electrical conductivity and Seebeck coefficient, and reduction of thermal conductivity in Sb2Te3-Au system.The enhanced electrical conductivity lies in the incorporated Au nanostructures contributing to injecting carriers to Sb2Te3 matrix.The appropriate barriers originated from the Au-Sb2Te3 interface, which filter low energy carriers, results in enhancement of Seebeck coefficient.The increased boundaries and nanodomains block the transport of phonons, subsequently reducing the thermal conductivity.As a consequence, combination of these effects promote double of ZT value in 1% Au@ Sb2Te3 composites with respect to the pristine Sb2Te3.

physics.app-ph↗

arXiv2015

Elastic Waves Scattering without Conversion in Metamaterials with Simultaneous Zero Indices for Longitudinal and Transverse waves

We theoretically investigate elastic waves propagating in metamaterials with simultaneous zero indices for both the longitudinal and transverse waves. With scattering objects (here cylinders) present in the metamaterials slabs, while the elastic waves can mostly transmit through the metamaterials slabs perfectly, exhibiting the well-known cloaking effect of zero index metamaterials, they nevertheless become totally blocked at resonances, indicating strong elastic waves scattering by the objects in the cases. However, despite the occurrence of the elastic waves scattering, there is, counter-intuitively, no mode conversion between the longitudinal and transverse waves in the process, completely in contrast with that in conventional elastic media. A design of two-dimensional phononic crystal with these peculiar properties is presented.

cond-mat.mtrl-sci↗

arXiv2013

Three dimensional Dirac point at k=0 in photonic and phononic systems

While "Dirac cone" dispersions can only be meaningfully defined in two dimensional (2D) systems, the notion of a Dirac point can be extended to three dimensional (3D) classical wave systems. We show that a simple cubic photonic crystal composing of core-shell spheres exhibits a 3D Dirac point at the center of the Brillouin zone at a finite frequency. Using effective medium theory, we can map our structure to a zero refractive index material in which the effective permittivity and permeability are simultaneously zero at the Dirac point frequency. The Dirac point is six-fold degenerate and is formed by the accidental degeneracy of electric dipole and magnetic dipole excitations, each with three degrees of freedom. We found that 3D Dirac points at k=0 can also be found in simple cubic acoustic wave crystals, but different from the case in the photonic system, the 3D Dirac point in acoustic wave system is fourfold degenerate, and is formed by the accidental degeneracy of dipole and monopole excitations. Using effective medium theory, we can also describe this acoustic wave system as a material which has both effective mass density and reciprocal of bulk modulus equal to zero at the Dirac point frequency. For both the photonic and phononic systems, a subset of the bands has linear dispersions near the zone center, and they give rise to equi-frequency surfaces that are spheres with radii proportional to (omega-omega_dirac).

cond-mat.mtrl-sci↗

arXiv2013

Total transmission and total reflection of acoustic wave by zero index metamaterials loaded with general solid defects

This work investigates acoustic wave transmission property through a zero index metamaterials (ZIM) waveguide embedded with a general solid defect. Total transmission and total reflection can be achieved by adjusting the parameters of the solid defect. We comprehensively study how longitudinal/transverse wave speeds of the solid defect affect the acoustic wave propagation through the waveguide. A two-dimensional (2D) acoustic crystals (ACs) system with zero index is used to realize these intriguing transmission properties. Thus, our work provides more possibilities to manipulate acoustic wave propagation through ZIM.

cond-mat.mtrl-sci↗

arXiv2011

Dirac Cones at k=0 in Phononic Crystals

We show that two-dimensional phononic crystals exhibit Dirac cone dispersion at k=0 by exploiting dipole and quadrupole accidental degeneracy. While the equi-frequency surface of Dirac cone modes is isotropic, such systems exhibit "super-anisotropy", meaning that only transverse waves are allowed along certain directions while only longitudinal waves are allowed along some other directions. Only one mode, not two, is allowed near the Dirac point, and only two effective parameters, not four, are needed to describe the dispersion. Effective medium theory finds that the phononic crystals have effectively zero mass density and zero 1/C44 at the Dirac point. Numerical simulations are used to demonstrate the unusual elastic wave properties near the Dirac point frequency.

cond-mat.mtrl-sci↗