Sasaki-Einstein metrics on connected sums of $S^3\times S^4$
We completely characterize the $n$-fold connected sums of $S^3\times S^4$ that admit Sasaki-Einstein structures by proving that such a structure exists if and only if $n$ is even.
math.DG↗
arXiv subjects
Jihun Park
Publications and source records attributed to Jihun Park.
CRAFT: Constrained Reward via Attention Fine-Tuning for Subject Personalization without Composed Targets
Subject-driven image personalization---generating new images that preserve the identity of one or several reference subjects in novel scenes---is a foundational capability for modern visual content creation. It is currently dominated by generalized methods that fine-tune a pretrained multimodal diffusion transformer (MMDiT) on hundreds of thousands to millions of paired \emph{(reference, composed-target)} examples, where each composed target is a synthesized image of the subject in a novel scene. Producing such targets demands a costly multi-stage curation pipeline---LLM-based prompt generation, T2I-based composed-target synthesis, reference-subject extraction, VLM-based quality filtering, and correspondence labeling---and tightly couples each method to a particular target synthesizer and curation choice. We introduce \emph{CRAFT} (Constrained Reward via Attention Fine-Tuning), a single-step ReFL framework that fine-tunes a pre-trained \emph{reference-aware} MMDiT via LoRA adapters using a compact reference-only data construction---$10$K reference images and subject masks, with no composed-target supervision. CRAFT realizes a \emph{Where to look} principle: attention-level rewards align noise- and phrase-token attention with the correct reference subject, and the resulting per-subject attention masks gate a pixel-level identity reward to keep image-space supervision consistent with the learned attention routing. Applied to FLUX.2-klein-9B, CRAFT achieves state-of-the-art performance on XVerseBench \rev{while using no composed-target supervision---only $10$K reference-only samples, whereas prior generalized methods require $150$K to over $2$M composed-target pairs}. The same recipe transfers to other reference-aware backbones, consistently improving performance. Project page: https://jihun999.github.io/projects/CRAFT/.
cs.CV↗
Soft Mixture-of-Recursions: Going Deeper with Recursive Vision Transformers
Recent recursive Transformer studies have primarily reused shared parameters across computation steps to construct compact, parameter-efficient models. In this work, we leverage recursion to build effectively deeper Transformers with stronger representational capacity. However, in Vision Transformers, simply increasing recursion depth does not reliably improve performance, as existing recursive approaches do not fully utilize the intermediate representations produced throughout recursive computation. We propose Soft Mixture-of-Recursions (SoftMoR) and its Vision Transformer instantiation, Soft Recursive Vision Transformer (SR-ViT). SoftMoR learns token-wise mixture weights to softly combine outputs from all recursion steps, allowing intermediate representations to be utilized in a learnable and flexible way. Across diverse vision tasks, SR-ViT consistently improves as recursion depth increases with minimal parameter overhead. On ImageNet-1K, increasing recursion depth from 1 to 4 improves SR-ViT-S top-1 accuracy from 79.83% to 82.48% with only 1.7M additional parameters, outperforming the substantially larger DeiT-B while using approximately 27% of its parameters. These results demonstrate that SoftMoR provides a parameter-efficient path to deeper and stronger Vision Transformers through recursion.
cs.CV↗
CascadeOcc: Rethinking 3D Occupancy World Models with Cascaded VQ Representations
This letter proposes CascadeOcc, a novel occupancy world model that prioritizes intrinsic structural hierarchy over extrinsic auxiliary modalities for autonomous driving. Occupancy world models -- forecasting the future driving environment and planning the driving trajectory -- effectively bridge perception and planning, but current approaches often heavily rely on external modalities or large language models, failing to fully exploit the inherent structural potential of occupancy representations themselves. To enhance representational capacity for complex 3D scenes, we integrate a cascaded Vector Quantized (VQ) mechanism into an autoregressive framework. Following a coarse-to-fine principle, CascadeOcc progressively refines fine-grained details from global structures through a multi-scale architecture. Additionally, we incorporate a TimeMixer to capture multi-scale temporal dependencies, establishing a dual-hierarchy mechanism in both space and time. Experimental results on 4D occupancy forecasting and motion planning benchmarks demonstrate that CascadeOcc achieves superior performance among vision-centric approaches, validating that optimizing inherent representations is a powerful alternative to relying on external foundation models.
cs.CV↗
On K-stability of Fano's last Fanos
We study K-stability of smooth Fano threefolds of Picard rank $2$ and degree $22$ which can be obtained by blowing up a smooth complete intersection of two quadrics in $\mathbb{P}^5$ along a conic. We also describe the automorphism groups of these threefolds.
math.AG↗
Robust Nonlinear Transform Coding: A Framework for Generalizable Joint Source-Channel Coding
This paper proposes robust nonlinear transform coding (Robust-NTC), a generalizable digital joint source-channel coding (JSCC) framework that couples variational latent modeling with channel-adaptive transmission. Unlike learning-based JSCC methods that implicitly absorb channel variations, Robust-NTC explicitly models element-wise latent distributions via a variational objective with a Gaussian proxy for quantization and channel noise, allowing encoder-decoder to capture latent uncertainty without channel-specific training. Using the learned statistics, Robust-NTC also facilitates rate-distortion optimization to adaptively select element-wise quantizers and bit depths according to online channel conditions. To support practical deployment, Robust-NTC is integrated into an orthogonal frequency-division multiplexing (OFDM) system, where a unified resource allocation framework jointly optimizes latent quantization, bit allocation, modulation order, and power allocation to minimize transmission latency while guaranteeing learned distortion targets. Simulation results demonstrate that for practical OFDM systems, Robust-NTC achieves superior rate-distortion efficiency and stable reconstruction fidelity compared to both a conventional separated coding scheme and digital JSCC baselines across various channel conditions.
eess.SP↗
Bridging Geometric and Semantic Foundation Models for Generalized Monocular Depth Estimation
We present Bridging Geometric and Semantic (BriGeS), an effective method that fuses geometric and semantic information within foundation models to enhance Monocular Depth Estimation (MDE). Central to BriGeS is the Bridging Gate, which integrates the complementary strengths of depth and segmentation foundation models. This integration is further refined by our Attention Temperature Scaling technique. It finely adjusts the focus of the attention mechanisms to prevent over-concentration on specific features, thus ensuring balanced performance across diverse inputs. BriGeS capitalizes on pre-trained foundation models and adopts a strategy that focuses on training only the Bridging Gate. This method significantly reduces resource demands and training time while maintaining the model's ability to generalize effectively. Extensive experiments across multiple challenging datasets demonstrate that BriGeS outperforms state-of-the-art methods in MDE for complex scenes, effectively handling intricate structures and overlapping objects.
cs.CV↗
On K-stability of singular hyperelliptic Fano 3-folds
We study the K-stability of singular Fano 3-folds with canonical Gorenstein singularities whose anticanonical linear system is base-point-free but not very ample.
math.AG↗
Beam-Squint-Aided Hierarchical Sensing for Integrated Sensing and Communications with Uniform Planar Arrays
In this paper, we propose a novel hierarchical sensing framework for wideband integrated sensing and communications with uniform planar arrays (UPAs). Leveraging the beam-squint effect inherent in wideband orthogonal frequency-division multiplexing (OFDM) systems, the proposed framework enables efficient two-dimensional angle estimation through a structured multi-stage sensing process. Specifically, the sensing procedure first searches over the elevation angle domain, followed by a dedicated search over the azimuth angle domain given the estimated elevation angles. In each stage, true-time-delay lines and phase shifters of the UPA are jointly configured to cover multiple grid points simultaneously across OFDM subcarriers. To enable accurate and efficient target localization, we formulate the angle estimation problem as a sparse signal recovery problem and develop a modified matching pursuit algorithm tailored to the hierarchical sensing architecture. Additionally, we design power allocation strategies that minimize total transmit power while meeting performance requirements for both sensing and communication. Numerical results demonstrate that the proposed framework achieves superior performance over conventional sensing methods with reduced sensing power.
eess.SP↗
A Training-Free Style-aligned Image Generation with Scale-wise Autoregressive Model
We present a training-free style-aligned image generation method that leverages a scale-wise autoregressive model. While large-scale text-to-image (T2I) models, particularly diffusion-based methods, have demonstrated impressive generation quality, they often suffer from style misalignment across generated image sets and slow inference speeds, limiting their practical usability. To address these issues, we propose three key components: initial feature replacement to ensure consistent background appearance, pivotal feature interpolation to align object placement, and dynamic style injection, which reinforces style consistency using a schedule function. Unlike previous methods requiring fine-tuning or additional training, our approach maintains fast inference while preserving individual content details. Extensive experiments show that our method achieves generation quality comparable to competing approaches, significantly improves style alignment, and delivers inference speeds over six times faster than the fastest model.
cs.CV↗
A Training-Free Style-Personalization via SVD-Based Feature Decomposition
We present a training-free framework for style-personalized image generation that operates during inference using a scale-wise autoregressive model. Our method generates a stylized image guided by a single reference style while preserving semantic consistency and mitigating content leakage. Through a detailed step-wise analysis of the generation process, we identify a pivotal step where the dominant singular values of the internal feature encode style-related components. Building upon this insight, we introduce two lightweight control modules: Principal Feature Blending, which enables precise modulation of style through SVD-based feature reconstruction, and Structural Attention Correction, which stabilizes structural consistency by leveraging content-guided attention correction across fine stages. Without any additional training, extensive experiments demonstrate that our method achieves competitive style fidelity and prompt fidelity compared to fine-tuned baselines, while offering faster inference and greater deployment flexibility.
cs.CV↗
Infinite-Story: A Training-Free Consistent Text-to-Image Generation
We present Infinite-Story, a training-free framework for consistent text-to-image (T2I) generation tailored for multi-prompt storytelling scenarios. Built upon a scale-wise autoregressive model, our method addresses two key challenges in consistent T2I generation: identity inconsistency and style inconsistency. To overcome these issues, we introduce three complementary techniques: Identity Prompt Replacement, which mitigates context bias in text encoders to align identity attributes across prompts; and a unified attention guidance mechanism comprising Adaptive Style Injection and Synchronized Guidance Adaptation, which jointly enforce global style and identity appearance consistency while preserving prompt fidelity. Unlike prior diffusion-based approaches that require fine-tuning or suffer from slow inference, Infinite-Story operates entirely at test time, delivering high identity and style consistency across diverse prompts. Extensive experiments demonstrate that our method achieves state-of-the-art generation performance, while offering over 6X faster inference (1.72 seconds per image) than the existing fastest consistent T2I models, highlighting its effectiveness and practicality for real-world visual storytelling.
cs.CV↗
Rate-Adaptive Semantic Communication via Multi-Stage Vector Quantization
This paper proposes a novel framework for rate-adaptive semantic communication based on multi-stage vector quantization (VQ), termed \textit{MSVQ-SC}. Unlike conventional single-stage VQ approaches, which require exponentially larger codebooks to achieve higher fidelity, the proposed framework decomposes the quantization process into multiple stages and dynamically activates both stages and individual VQ modules. This design enables fine-grained rate adaptation under varying bit constraints while mitigating computational complexity and the codebook collapse problem. To optimize performance, we formulate a module selection problem that minimizes task loss subject to a rate constraint and solve it using an incremental allocation algorithm. Furthermore, we extend the framework by incorporating entropy coding to exploit non-uniform codeword distributions, further reducing communication overhead. Simulation results on the CIFAR-10 dataset demonstrate that the proposed framework outperforms existing digital semantic communication methods, achieving superior semantic fidelity with lower complexity while providing flexible and fine-grained rate control.
eess.SP↗
Conical Kähler-Einstein metrics on K-unstable del Pezzo surfaces
We establish the optimal upper bounds for cone angles of Kähler-Einstein metrics with conical singularities along smooth anticanonical divisors on smooth K-unstable del Pezzo surfaces.
math.AG↗
Importance-Aware Semantic Communication in MIMO-OFDM Systems Using Vision Transformer
This paper presents a novel importance-aware quantization, subcarrier mapping, and power allocation (IA-QSMPA) framework for semantic communication in multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) systems, empowered by a pretrained Vision Transformer (ViT). The proposed framework exploits attention-based importance extracted from a pretrained ViT to jointly optimize quantization levels, subcarrier mapping, and power allocation. Specifically, IA-QSMPA maps semantically important features to high-quality subchannels and allocates resources in accordance with their contribution to task performance and communication latency. To efficiently solve the resulting nonconvex optimization problem, a block coordinate descent algorithm is employed. The framework is further extended to operate under finite blocklength transmission, where communication errors may occur. In this setting, a segment-wise linear approximation of the channel dispersion penalty is introduced to enable efficient joint optimization under practical constraints. Simulation results on a multi-view image classification task using the MVP-N dataset demonstrate that IA-QSMPA significantly outperforms conventional methods in both ideal and finite blocklength transmission scenarios, achieving superior task performance and communication efficiency.
eess.SP↗
Style-Editor: Text-driven object-centric style editing
We present Text-driven object-centric style editing model named Style-Editor, a novel method that guides style editing at an object-centric level using textual inputs. The core of Style-Editor is our Patch-wise Co-Directional (PCD) loss, meticulously designed for precise object-centric editing that are closely aligned with the input text. This loss combines a patch directional loss for text-guided style direction and a patch distribution consistency loss for even CLIP embedding distribution across object regions. It ensures a seamless and harmonious style editing across object regions. Key to our method are the Text-Matched Patch Selection (TMPS) and Pre-fixed Region Selection (PRS) modules for identifying object locations via text, eliminating the need for segmentation masks. Lastly, we introduce an Adaptive Background Preservation (ABP) loss to maintain the original style and structural essence of the image's background. This loss is applied to dynamically identified background areas. Extensive experiments underline the effectiveness of our approach in creating visually coherent and textually aligned style editing.
cs.CV↗
Probing $p$-wave superconductivity in UTe$_2$ via point-contact junctions
Uranium ditelluride (UTe$_2$) is the strongest contender to date for a $p$-wave superconductor in bulk form. Here we perform a spectroscopic study of the ambient pressure superconducting phase of UTe$_2$, measuring conductance through point-contact junctions formed by metallic contacts on different crystalline facets down to 250 mK and up to 18 T. Fitting a range of qualitatively varying spectra with a Blonder-Tinkham-Klapwijk(BTK) model for $p$-wave pairing, we can extract gap amplitude and interface barrier strength for each junction. We find good agreement with the data for a $p_y$ -wave gap function with amplitude in 0.26 $\pm$ 0.06 meV. Our work provides spectroscopic evidence for a gap structure consistent with the proposed spin-triplet pairing in the superconducting state of UTe$_2$.
cond-mat.supr-con↗
Superconducting phase diagram in Bi$_x$Ni$_{1-x}$ thin films$\colon$ the effects of Bi stoichiometry on superconductivity
The Bi${-}$Ni binary system has been of interest due to possible unconventional superconductivity aroused therein, such as time-reversal symmetry breaking in Bi/Ni bilayers or the coexistence of superconductivity and ferromagnetism in Bi$_3$Ni crystals. While Ni acts as a ferromagnetic element in such systems, the role of strong spin-orbit-coupling element Bi in superconductivity has remained unexplored. In this work, we systematically studied the effects of Bi stoichiometry on the superconductivity of Bi$_x$Ni$_{1-x}$ thin films (${x} \approx$ 0.5 to 0.9) fabricated via a composition-spread approach. The superconducting phase map of Bi$_x$Ni$_{1-x}$ thin films exhibited a superconducting composition region attributable to the intermetallic Bi$_3$Ni phase with different amount of excess Bi, revealed by synchrotron X-ray diffraction analysis. Interestingly, the mixed phase region with Bi$_3$Ni and Bi showed unusual increases in the superconducting transition temperature and residual resistance ratio as more Bi impurities were included, with the maximum ${T}_{c}$ ($=$ 4.2 K) observed at $x \approx$ 0.79. A correlation analysis of structural, electrical, and magneto-transport characteristics across the composition variation revealed that the unusual superconducting $"$dome$"$ is due to two competing roles of Bi$\colon$ impurity scattering and carrier doping. We found that the carrier doping effect is dominant in the mild doping regime (0.74 $\leq {x} \leq$ 0.79), while impurity scattering becomes more pronounced at larger Bi stoichiometry.
cond-mat.supr-con↗