Searcharxiv⌕ Search

arXiv subjects

Tao Su

Publications and source records attributed to Tao Su.

17 recordsLinked to original sources

SFU-Sim: Bit-Exact Reverse Engineering of NVIDIA MUFU and Validation of PTX Paths

Reproducing the numerical behavior of a GPU requires predicting the exact output of special function instructions. Vendor documents give only the function semantics and error bounds. These do not fix the output bit pattern. We present SFU-Sim, which builds bit-exact models of NVIDIA multi-function unit (MUFU) instructions by black-box reverse engineering on real GPUs. We first model a single bare MUFU instruction. We then add the surrounding operations found in the disassembly to model full PTX paths. The model covers 12 common MUFU forms and two BF16 forms on the tested Blackwell GPU. Using SIN/COS as a complete case study, we show how we analyze the PTX-to-SASS compilation path and recover the segment structure, the equivalent coefficient tables, and the finite-width arithmetic. On NVIDIA GPUs of three architectures (Ampere, Ada Lovelace, and Blackwell), we exhaustively compare every tested PTX path over its specified input domain. FP16 and FP32 paths cover all $2^{16}$ and $2^{32}$ input encodings, respectively. The model output matches the hardware bit for bit, with zero mismatches. SFU-Sim computes each output directly from its input. It needs neither a lookup table of all inputs nor access to a physical GPU. It can serve as a bit-exact special function component for instruction simulators. It can also serve as an arithmetic reference for hardware design and functional verification that target the same numerical behavior. SFU-Sim and its validation programs are open source at https://github.com/hzX40000000/SFU-Sim

cs.AR↗

MDA framework: multi-architecture domain adaptation for unsupervised person re-identification

Unsupervised person re-identification not only faces the challenge of domain gaps, but existing methods also rarely support multiple network architectures. To tackle both issues, this paper proposes a multi-architecture auto-adaptive domain adaptation method: Mamba augmented multi-architecture domain adaptation framework. Specifically, this paper constructs the multi-architecture domain adaptation capability at two levels. At the first level, we propose a composite operator(MDA operator) that encapsulates Mamba as a component for enhancing domain adaptation. At the second level, based on this operator, we propose a Mamba-enhanced multi-architecture Domain Adaptation framework(MDA framework). Comprehensive experiments on person re-identification benchmarks validate the consistency and flexibility of the proposed framework. The MDA framework supports versatile deployment strategies without being confined to a single integration point. Under a single deployment configuration, it surpasses several SOTA methods. Moreover, it demonstrates substantial performance improvements across both CNN and Transformer architectures. When embedded into a CNN backbone, the framework achieves a 10.2% mAP improvement on the adaptation task from CUHK03 to Market-1501. Integrated with a Transformer backbone, it yields a 19.8% mAP improvement on the same task.

cs.CV↗

Integral cohomology of dual boundary complexes is motivic

In this note, we give a motivic characterization of the integral cohomology of dual boundary complexes of smooth quasi-projective complex algebraic varieties. As a corollary, the dual boundary complex of any variety stably isomorphic to $\mathbb{G}_m^r\times \mathbb{A}^s$ with $s>0$ is contractible. In a separate paper [Su26], this corollary has been used by the author in his proof of the weak geometric P=W conjecture for very generic $GL_n(\mathbb{C})$-character varieties over any punctured Riemann surfaces.

math.AG↗

Cell decomposition and dual boundary complexes of character varieties

The weak geometric P=W conjecture of Katzarkov--Noll--Pandit--Simpson asserts that for any smooth Betti moduli space $\mathcal{M}_B$ of complex dimension $d$ associated to a punctured Riemann surface, the dual boundary complex $\mathbb{D}\partial\mathcal{M}_B$ is homotopy equivalent to a $(d-1)$-sphere. Here, $\mathcal{M}_B$ is a generic $GL_n(\mathbb{C})$-character variety of genus $g$, with prescribed semisimple local monodromies at $k$ punctures. In this article, we prove the weak geometric P=W conjecture for all \emph{very generic} $\mathcal{M}_B$, namely when at least one conjugacy class is regular semisimple. A key step is a strengthening of Mellit's cell decomposition theorem: we decompose $\mathcal{M}_B$ itself (without passing to a vector bundle) into locally closed subvarieties which are stably isomorphic to $(\mathbb{C}^{\times})^{d-2b}\times \mathbb{C}^b$. A second ingredient is a motivic characterization of the integral cohomology of dual boundary complexes, developed in [Su24]. Following Simpson's strategy, the proof reduces to an inductive computation of dual boundary complexes arising from this cell decomposition.

math.AG↗

ASARL: Autonomous Social-Aware Relevance Learning for QQ Search

The rapid growth of online social platforms has transformed communication and information retrieval, giving rise to social search, where queries-titles are typically expressed in informal, community-specific language. While large language models provide strong general-purpose semantic understanding, their effectiveness in social search is constrained by contextual discrepancy, data scarcity, and behavior-driven dynamics. To address these challenges, we propose the Autonomous Social-Aware Relevance Learning (ASARL), a fully automated framework that integrates multi-agent data curation with staged model training. ASARL leverages a collaborative agent system: ReasonAgent generates interpretable relevance labels grounded in social attributes, CriticAgent validates and ensures logical consistency, and GenAgent augments long-tail data through synthetic query-title pairs. Building on the curated dataset, ASARL employs three-stage training: Social Context Training (SCT) to capture social language patterns, Preference-Guided Optimization (PGO) to align model predictions with behavioral signals, and Social Distillation (SD) to transfer these improvements into compact models for efficient deployment. Extensive offline and online experiments on the QQ search platform demonstrate significant improvements in both offline relevance metrics and online user engagement indicators, along with enhanced annotation efficiency. These results validate the effectiveness of combining autonomous, socially grounded data governance with preference-aligned training in practical search systems.

cs.IR↗

Log-concavity from enumerative geometry of planar curve singularities

We propose a log-concavity conjecture for BPS invariants arising in the enumerative geometry of planar curve singularities, identified with the local Euler obstructions of Severi strata in their versal deformations. We further extend this conjecture to ruling polynomials of Legendrian links and to E-polynomials of character varieties. We establish these conjectures for irreducible weighted-homogeneous singularities (torus knots) and for ADE singularities, and prove a multiplicative property for ruling polynomials compatible with log-concavity.

math.AG↗

AnalogMaster: Large Language Model-based Automated Analog IC Design Framework from Image to Layout

Design automation has the potential to substantially improve the efficiency of analog integrated circuit (IC) design. However, existing algorithms and tools typically focus on individual stages, such as device sizing, placement, or routing, and still require significant manual intervention to complete the full design flow. While large language models (LLMs) have recently demonstrated remarkable success in automating digital IC design workflows, these advances cannot be directly transferred to analog IC design. Key challenges include strongly coupled performance metrics, the predominance of unstructured circuit schematic images, and the fact that most prior approaches address only isolated stages of the analog design process, limiting their ability to capture end-to-end performance impact. To address these challenges, we propose AnalogMaster, an extensible, LLM-based framework that enables end-to-end automation of analog IC design through a unified pipeline spanning circuit image-to-netlist generation, parameter optimization, placement, and routing. AnalogMaster integrates a joint reasoning mechanism that leverages in-context learning and intent reasoning to achieve accurate and robust image-to-netlist conversion. A parameter search agent integrating self-enhanced prompt engineering and context truncation is developed for effective device sizing and downstream physical design. Experimental evaluations on 15 representative circuits with varying levels of complexity demonstrate strong and consistent performance across multiple models. In particular, GPT-5 achieves success rates of 92.9% and 99.9% on Pass@1 and Pass@5, respectively. These results validate the effectiveness and robustness of the proposed framework and establish a practical paradigm for applying LLMs to full-stack analog IC design automation.

cs.AR↗

Purity of generalized affine Springer fibers from generic planar curve singularities

We prove the cohomological purity of punctual Hilbert schemes of points on generic irreducible planar curve singularities, by constructing an explicit affine paving. Via their identification with generalized $GL_N$-affine Springer fibers attached to the direct sum of the adjoint and standard representations, this establishes a new case of the purity conjecture for generalized affine Springer fibers. The combinatorics of the paving - cell indices and dimensions - are controlled by $(dn,dm)$-Dyck paths extending results of Gorsky-Mazin-Oblomkov on compactified Jacobians. As a byproduct, we also give a simpler proof of their bijection between admissible $(dn,dm)$-invariant subsets and $(dn,dm)$-Dyck paths.

math.AG↗

Model Predictive Control For Mobile Manipulators Based On Neural Dynamics(Extended version)

This article focuses on the trajectory tracking problem of mobile manipulators (MMs). Firstly, we construct a position and orientation model predictive tracking control (POMPTC) scheme for mobile manipulators. The proposed POMPTC scheme can simultaneously minimize the tracking error, joint velocity, and joint acceleration. Moreover, it can achieve synchronous control for the position and orientation of the end-effector. Secondly, a finite-time convergent neural dynamics (FTCND) model is constructed to find the optimal solution of the POMPTC scheme. Then, based on the proposed POMPTC scheme, a non-singular fast terminal sliding model (NFTSM) control method is presented, which considers the disturbances caused by the base motion on the manipulator at the dynamic level. It can achieve finite-time tracking performance and improve the anti-disturbances ability. Finally, simulation and experiments show that the proposed control method has the advantages of strong robustness, fast convergence, and high control accuracy.

cs.RO↗

Embedded Silicon-Organic Integrated Neuromorphic System

The development of artificial intelligence (AI) and robotics are both based on the tenet of "science and technology are people-oriented", and both need to achieve efficient communication with the human brain. Based on multi-disciplinary research in systems neuroscience, computer architecture, and functional organic materials, we proposed the concept of using AI to simulate the operating principles and materials of the brain in hardware to develop brain-inspired intelligence technology, and realized the preparation of neuromorphic computing devices and basic materials. We simulated neurons and neural networks in terms of material and morphology, using a variety of organic polymers as the base materials for neuroelectronic devices, for building neural interfaces as well as organic neural devices and silicon neural computational modules. We assemble organic artificial synapses with simulated neurons from silicon-based Field-Programmable Gate Array (FPGA) into organic artificial neurons, the basic components of neural networks, and later construct biological neural network models based on the interpreted neural circuits. Finally, we also discuss how to further build neuromorphic devices based on these organic artificial neurons, which have both a neural interface friendly to nervous tissue and interact with information from real biological neural networks.

q-bio.NC↗

Dual boundary complexes of Betti moduli spaces over the two-sphere with one irregular singularity

The weak geometric P=W conjecture of L. Katzarkov, A. Noll, P. Pandit, and C. Simpson states that, a smooth Betti moduli space of complex dimension $d$ over a punctured Riemann surface has the dual boundary complex homotopy equivalent to a sphere of dimension $d-1$. Via a microlocal geometric perspective, we verify this conjecture for a class of rank $n$ wild character varieties over the two-sphere with one puncture, associated with any Stokes Legendrian link defined by an $n$-strand positive braid.

math.AG↗

Practical Frequency-Hopping MIMO Joint Radar Communications: Design and Experiment

Joint radar and communications (JRC) can realize two radio frequency (RF) functions using one set of resources, greatly saving hardware, energy and spectrum for wireless systems needing both functions. Frequency-hopping (FH) MIMO radar is a popular candidate for JRC, as the achieved communication symbol rate can greatly exceed radar pulse repetition frequency. However, practical transceiver imperfections can fail many existing theoretical designs. In this work, we unveil for the first time the non-trivial impact of hardware imperfections on FH-MIMO JRC and analytically model the impact. We also design new waveforms and, accordingly, develop a low-complexity algorithm to jointly estimate the hardware imperfections of unsynchronized receiver. Moreover, employing low-cost software-defined radios and commercial off-the-shelf (COTS) products, we build the first FH-MIMO JRC experiment platform with radar and communications simultaneously validated over the air. Corroborated by simulation and experiment results, the proposed designs achieves high performances for both radar and communications.

eess.SP↗

Radar Accurate Localization of UAV Swarms Based on Range Super-Resolution Method

In radar accurate localization of unmanned aerial vehicle (UAV) swarms, the high density, similar motion parameters, small radar cross-section (RCS), strong noise and far range put forward high requirements on radar resolution and transmitting power. In this paper, by using advantages of the long-time integration (LTI) technique and gridless sparse method, we construct a super-resolution framework for radar accurate localization of UAV swarms without changing radar hardware and system parameters. Thereafter, based on this framework, a range super-resolution method is proposed to realize the radar accurate localization of UAV swarms. Mathematical analyses and numerical simulations are performed and demonstrate that, compared to the keystone transform (KT)-based LTI method, MUSIC-based method and reweighted atomic-norm minimization (RAM)-based method, the range super-resolution method is more robust and practical for radar accurate localization of UAV swarms under the noisy environment. Additionally, the real experiment with X-band radar is also conducted to verify the effectiveness of the range super-resolution method.

eess.SP↗

Augmentations are sheaves for Legendrian graphs

In this article, associated to a (bordered) Legendrian graph, we study and show the equivalence between two categorical Legendrian isotopy invariants: the augmentation category, a unital $A_{\infty}$-category, which lifts the set of augmentations of the associated Chekanov-Eliashberg DGA, and a DG category of constructible sheaves on the front plane, with micro-support at contact infinity controlled by the (bordered) Legendrian graph. In other words, generalizing [21], we prove "augmentations are sheaves" in the singular case.

math.SG↗

Augmentations and ruling polynomials for Legendrian graphs

In this article, associated to a (bordered) Legendrian graph, we study and show the equivalence between two Legendrian isotopy invariants: augmentation number via point-counting over a finite field, for the augmentation variety of the associated Chekanov-Eliashberg differential graded algebra, and ruling polynomial via combinatorics of the decompositions of the associated front projection.

math.SG↗

Ruling polynomials and augmentations for Legendrian tangles

Associated to Legendrian links in the standard contact three-space, Ruling polynomials are Legendrian isotopy invariants, which also compute augmentation numbers, that is, the points-counting of augmentation varieties for Legendrian links (up to a normalized factor) \cite{HR15}. In this article, we generalize this picture to Legendrian tangles, which are morally the pieces obtained by cutting Legendrian link fronts along 2 vertical lines. Moreover, we show that the Ruling polynomials for Legendrian tangles satisfy the composition axiom. In the special case of Legendrian knots, our arguments provide new proofs to the main results in \cite{HR15}. In the end, we also introduce generalized Ruling polynomials for Legendrian tangles, to account for non-acyclic augmentations in the "Ruling polynomials compute augmentation numbers" picture.

math.SG↗

Overcrowding drives the unjamming transition of gap-free monolayers

Collective cell motility plays central roles in various biological phenomena such as inflammatory response, wound healing, cancer metastasis and embryogenesis. These are biological demonstrations of the unjamming transition. However, contradictory to the typical density-driven jamming processes in particulate assemblies, cellular systems often get unjammed in highly packed, sometimes overcrowding tissue environments. In this work, we report that overcrowding can unjam gap-free monolayers through increasing isotropic compression. The transition boundary is determined by the isotropic compression and the cell-cell adhesion. We explicitly construct the free energy landscape for the T1 topological transition during monolayer rearrangement, and find that it evolves from single-barrier shape to double-barrier shape upon completion of the unjamming process. Our analyses reveal that the overcrowding and adhesion induced unjamming transition reflects the mechanical yielding of the highly deformable monolayer, which differs from those caused by loosing up a packed particulate assembly.

physics.bio-ph↗