arXiv subjects

Minkyu Kim

Publications and source records attributed to Minkyu Kim.

At least 19 recordsLinked to original sources

arXiv2026

Exact $\ell_p$-CVP via Euclidean SVP in Polynomial Space

We give a randomized, rank-preserving reduction from exact $\ell_p$-CVP to exact Euclidean SVP for every integer $p\ge2$. For rational inputs of bit length $\ell$ on full-rank lattices $L\subset\mathbb R^N$, the reduction uses $(Cp)^N\poly(p,\ell)$ oracle calls and additional bit operations, where $C$ is an absolute constant. It uses $\poly(p,\ell)$ working space outside the oracle, and all oracle queries have rank $N$ and $\poly(p,\ell)$ bit complexity. For $p=2$, the reduction transfers polynomial-space exact SVP algorithms to exact CVP with $2^{O(N)}\poly(\ell)$ time overhead. Instantiating it with the SVP algorithms of Kannan and of Feng, Gao and Liu yields randomized polynomial-space CVP algorithms running in time $N^{N/(2e)+o(N)}$ and $N^{N/(4e)+o(N)}$, respectively. For every fixed integer $p\ge3$, the same reduction yields exact $\ell_p$-CVP algorithms for arbitrary rational targets either in randomized time and space $2^{O(N)}$, or in randomized time $N^{N/(4e)+o(N)}$ and polynomial space. These algorithmic bounds are up to polynomial factors in $\ell$. The reduction performs local search sequentially over the cosets of $2L$, where $L$ is the input lattice, while tracking the excess of the current objective value over the optimum. Within a coset containing a closest vector, midpoint inequalities bound the unknown correction in terms of this excess, which in turn allows us to construct many short corrections that reduce the excess bound by a factor of four. A counting argument shows that these improving corrections form a sufficiently large fraction of the lattice points in a known search region, and a sparsification-based sampler assigns sufficiently large probability to every nonzero vector in that region, using at most one SVP call per sample. For $p\ge3$, the same framework is extended using a diagonal weighted norm computed from the current point.

cs.DS↗

arXiv2026

FinAutoRubric: Expert-Guided Automatic Rubric Generation for Evaluating Financial Research Agents

Evaluating finance research agents requires rubrics that reflect expert standards and fix the values correct as of an information cutoff. Expert-reviewed finance benchmarks rely on fixed, per-item rubrics, which are costly to extend and cannot encode each institution's own standard. In FinAutoRubric, experts specify reusable evaluation guidance, while agents and code carry out query-specific rubric generation, review, and validation. This expert guidance governs every agent, as prompts and as rules that code enforces, and a Task Bank of reusable criteria carries it across tasks. In long-horizon loops that follow the expert guidance, a writer agent researches every expected value and a reviewer agent verifies it, and failures escalate to a human. On three expert-authored finance benchmarks, its rubrics track expert scoring as closely as the strongest evaluated generator while stating the expert rubric's expected value for more criteria, their scores agree with human grading, and in-house analysts prefer them in a blind review. The released 100-query FinAutoRubric Benchmark, built from in-house analysts' key questions across 78 tasks and eight asset classes, shows that rubrics from an earlier model generation still leave headroom for a later one.

cs.AI↗

arXiv2026

ManiFusion: Unlocking the High-Throughput Generation via Superposition in Manifold Space

Diffusion models achieve high-quality image generation but remain computationally expensive due to iterative denoising across many timesteps. Existing acceleration methods mainly reduce per-sample latency, but gains along this axis are becoming saturated. We instead explore an orthogonal throughput-oriented direction by processing multiple samples within a single denoising evaluation. To this end, we propose ManiFusion, a backbone-agnostic framework for multi-sample generation through fused-space computation. ManiFusion jointly processes multiple samples by fusing them before the denoising backbone and separating them afterward through unfusion. We provide a geometric perspective on how multiple samples can share a fused representation while retaining the information needed to guide their individual generation trajectories. To better align fusion allocation with denoising stages, ManiFusion uses Trajectory-Fusion Scheduling (TFS), which dynamically adjusts fusion factors throughout sampling. ManiFusion generalizes across diffusion families, including both U-Net and transformer backbones in pixel and latent spaces. We further show that ManiFusion composes with existing acceleration methods and reaches throughput regimes unattainable by prior approaches with minimal quality degradation.

cs.CV↗

arXiv2026

AtomMOF: All-Atom Flow Matching for MOF-Adsorbate Structure Prediction

Metal-organic frameworks (MOFs) are promising materials for applications such as direct air capture, where performance depends on how adsorbates bind within the framework. Accurately predicting MOF-adsorbate configurations is therefore important for screening candidate materials. However, traditional approaches are computationally expensive and require a known host structure, while existing generative models rely on rigid-body assumptions and do not explicitly model adsorbates. We introduce AtomMOF, a scalable all-atom flow-matching model that jointly predicts MOF-adsorbate structures from discrete building blocks and adsorbates. Built on a Diffusion Transformer with a building block-based pairwise attention bias, AtomMOF operates in an unconstrained all-atom space and exhibits clear scaling behavior. To improve the structural validity of flexible all-atom models, we also propose Feynman-Kac (FK) steering with machine-learned interatomic potentials (MLIPs). On the BW dataset, AtomMOF achieves a 58.17% relative increase in match rate and a 31.84% relative reduction in RMSD compared with prior work. MLIP-guided steering further improves validity by 28.7% and reduces formation energy error by 86.5%. On ODAC25, AtomMOF generates adsorption configurations faster than GCMC and, when combined with MLIP relaxation, identifies lower-energy configurations than those in the reference dataset.

cond-mat.mtrl-sci↗

arXiv2026

Regular Syndrome Decoding Is Easier Than Previously Estimated

The Regular Syndrome Decoding (RSD) problem is a structured variant of classical Syndrome Decoding and underlies a growing number of efficient cryptographic constructions, including zero knowledge protocols, pseudorandom correlation generators, and MPC in the Head signatures. Although RSD is generally believed to be computationally hard, recent works have shown that algebraic attacks can outperform information set decoding on important RSD parameter regimes. Building on this line of work, we revisit the concrete hardness of RSD and show that many widely used parameter sets are substantially easier to solve than previously estimated. This improvement stems from a refined hybrid algebraic attack that exposes and removes a previously overlooked source of concrete complexity: the joint blockwise placement of zero coordinate guesses and syndrome based variable eliminations. Rather than treating the two guessing stages and linear elimination as largely separate choices, our method coordinates them across all regular blocks so that the guessing penalty is balanced while sufficient equations are retained for linearization. This gives a systematic block balancing principle for the entire guess reduce solve pipeline. We reassess several RSD based applications, including Ferret, Wolverine, and committed VOLE. Our estimates place at least one parameter set per application below its 128 bit target, with losses approaching 67 bits. On PCG benchmarks, we improve previous best attack costs by up to 29.77 and 25.91 bits over non binary and binary fields, respectively. Moreover, for the revised 128 bit PCG parameter sets of Liu et al. evaluated here, estimated attack costs are all below $2^{111}$ field operations. These results substantially revise the concrete security landscape of RSD and indicate that several existing parameter choices overestimate the hardness of their underlying RSD instances.

cs.CR↗

arXiv2026

Policy-Aligned Predictive Verification for Adaptive VLA Action-Chunk Execution

Vision-language-action (VLA) policies generate action sequences that reduce repeated inference, but deciding how much of each sequence to execute is critical to task success. Existing adaptive methods make this decision using generation stability or prediction agreement, yet these signals do not establish whether executing a candidate prefix preserves the ability to complete the task. We introduce a policy-aligned predictive verifier (PAV) that anticipates the consequences of candidate action prefixes and evaluates their compatibility with successful continuation before execution. Rather than using future prediction to generate actions, our method evaluates existing proposals in the VLA's internal task-conditioned representation space. To learn local judgments from terminal task outcomes, we combine trajectory-level survival supervision with a localization prior that preserves useful behavior within failed rollouts. The resulting validity scores guide execution-horizon selection. Evaluations with pi_0 and pi_0.5 across LIBERO, RoboTwin 2.0, and real-world manipulation show improved success-query trade-offs. These findings point toward a different role for future prediction in robotics, using it to assess how far to trust a policy's proposed actions rather than to generate new ones.

cs.RO↗

arXiv2026

ACRO: Actor-Critic Rollout Orchestration

Executing a vision-language-action (VLA) policy involves deciding when to continue an attempt and when to intervene to improve its chances of success. We introduce ACRO, an Actor--Critic rollout orchestration layer that makes these decisions for a fixed VLA. ACRO seeks interventions that offer a more promising continuation, are physically feasible, and leave sufficient time to complete the task. A Critic learned from policy rollouts estimates success prospects to decide when to intervene and select a re-entry target along the executed trajectory. Trajectory Retraction shortcuts the visited path to return the robot to that target, where the same VLA resumes execution from a fresh observation. We evaluate ACRO with GR00T N1.7 and $π_{0.5}$ across SimplerEnv, RoboCasa, and real-robot manipulation. Under the same physical execution budget, including all return motion, ACRO improves success by 10.4, 7.3, and 21.7 percentage points, respectively. It also completes additional tasks as the execution budget grows. These results show that controlling when and where a policy resumes can turn its existing capabilities into more successful task executions. Project page: https://junhoo.me/acro

cs.RO↗

arXiv2026

Complementary Future Token Pruning for World Action Models

World action models (WAMs) jointly predict future visual states and robot actions, but processing dense future-frame tokens incurs substantial inference latency. We propose CF-Prune, a training-free framework for pruning future-frame tokens in pretrained WAMs. Our key insight is that future visual predictions support action generation and therefore need not preserve every visual detail required for high-fidelity video synthesis. CF-Prune combines instruction and action attention to identify task-relevant tokens during the initial denoising forward pass, even when future-frame and action representations remain highly noisy. An entropy-based phase identification mechanism further retains dense computation during precision phases to mitigate pruning-induced performance degradation. Experiments on RoboLab, RoboCasa, and real-world manipulation tasks demonstrate consistent inference speedups. At a 35% pruning ratio, CF-Prune achieves average speedups of 1.19-1.27x while improving task success rates. At a 50% pruning ratio, it achieves 1.29-1.33x average speedups on both simulation benchmarks, with overall success rates within 1.0 percentage point of the baseline.

cs.RO↗

arXiv2026

PXRD-Bench: How Close Are We to Determining Crystal Structures from Powder Diffraction?

Determining crystal structures from powder X-ray diffraction (PXRD) would enable routine characterization of newly synthesized materials, including those absent from reference databases. Generative models offer a promising approach, but existing evaluations remain fragmented and largely assume single-phase samples with known composition. To fill this gap, we introduce PXRD-Bench, a unified benchmark that extends evaluation to unknown per-element atom counts and recovery of both structures in two-phase mixtures. It provides approximately 4.1 million training structures with simulated PXRD, tests generalization to unseen compositions and structural prototypes, and evaluates recovery from experimental PXRD after excluding structurally matching training entries. To provide a strong reference baseline, we also introduce Bragglite, a PXRD-to-structure model developed through a systematic study of PXRD modeling choices. It achieves a 44.2\% Top-1 match rate on experimental RRUFF patterns with known composition, 24.6\% above the strongest prior baseline. Further analysis yields several key findings: some models fail to benefit from experimental PXRD, while generalization to unseen structural prototypes and recovery from two-phase mixtures remain challenging. Together, our benchmark, systematic studies, and Bragglite clarify the capabilities and limitations of current PXRD-to-structure models.

cs.AI↗

arXiv2026

Completion Aware Guidance for World Action Models

World Action Models (WAMs) predict visual futures and robot actions, yet they remain susceptible to task-incomplete imagination, where plausible, action-consistent predictions omit the transition needed for task completion. In this paper, we show that this failure is not inherent to the world model backbone, but emerges when adapted for short-chunk control, which can repeatedly favor plausible local continuations over task-completing transitions. To address this, we introduce Completion Aware Guidance (CAG), a training-free sampling method that guides generation toward task completion. Across representative WAMs, CAG improves success from 64% to 70% on a RoboTwin 2.0 subset and from 69% to 75% in zero-shot simulation, while reducing task-incomplete imagination from 79% to 40%.

cs.RO↗

arXiv2026

RobotUse: Allocating Computation, Context, and Decisions

Robot agents must connect their intended actions to observed outcomes while retaining the context needed to revise their choices over repeated attempts. Existing interfaces often leave these choices inside predefined tools or require agents to manage detailed execution code and its growing history. We introduce RobotUse, a robot agent harness that organizes computation, context, and decisions around specifying and revising physical actions. Agents visually select targets and poses, while the backend handles geometry, motion planning, and control. Subagents retain detailed interactions within each subgoal and return the information needed for subsequent decisions. Continual harnessing lets agents learn from execution by updating a persistent playbook. On RoboLab, RobotUse achieves 45% task success, outperforming CaP-X by 6.7 percentage points while maintaining compact decision contexts and reducing reliance on predefined action abstractions. Furthermore, we show that RobotUse learns from real-world execution despite imperfect feedback and transfers what it learns to subsequent tasks. Project page is available at https://robotuse-team.github.io/.

cs.RO↗

arXiv2026

Recursive Harness Distillation across Agents for Robot Manipulation

A central goal in robotics is to enable manipulation across changing tasks and environments. Vision-language-action (VLA) models provide broad manipulation capabilities but can struggle when execution requires diagnosing failures and adapting behavior. Strong agents can discover effective interventions through interaction with these policies. We propose Recursive Harness Distillation to accumulate this experience as reusable guidance across agents. A strong agent distills its experience into a playbook for a light agent, then recursively refines the playbook using the light agent's execution feedback. The resulting playbook enables agents to reuse accumulated intervention knowledge in new task instances without updating model parameters. In real-world manipulation, the harness improves success from 37.3% to 64.0%. On SimplerEnv Bridge, the light agent with the playbook achieves 66.7% success, compared with 41.7% for the GR00T-only baseline, and outperforms the strong agent without a playbook. The same playbook also benefits the strong agent, which reaches 79.2% success. These results demonstrate the feasibility of harness distillation for robotics: intervention experience can be accumulated, refined through execution, and reused across agents to improve manipulation.

cs.RO↗

arXiv2026

Bi-balanced shifts and variable specification

Examples of one-sided balanced shifts which do not have variable specification are provided. On the other hand, under synchronization condition, bi-balancedness is proven to be equivalent to variable specification.

math.DS↗

arXiv2026

When Summaries Distort Decisions: Information Fidelity in LLM-Compressed Financial Analysis

Financial decision-makers face more information than they can directly inspect, making context compression necessary. Yet when large language models (LLMs) compress financial source material, they can alter the investment judgment supported by the original source. We frame this problem as information fidelity: compression loses fidelity when it changes the decision induced by the source. In agentic systems, such losses may recur across intermediate steps and amplify throughout the decision process. Across financial filings and earnings-call transcripts, we find that LLM-based compression can produce fluent and factually plausible compressed contexts that nevertheless alter downstream decisions. We analyze two diagnostic patterns associated with fidelity loss: decontextualization, where salient evidence is retained but separated from the caveats and contextual qualifiers needed for correct interpretation, and model dependency, where different compressors expose different views of the same source. We then propose Agentic Context Compression, which generates multiple candidate compressions and audits their disagreements against the original source. Our results suggest that financial compression should be evaluated not only by efficiency or factuality, but also by its ability to preserve decision-relevant context.

cs.AI↗

arXiv2026

Locality-Aware Redundancy Pruning for LLM Depth Compression

Large language models are known to contain representational redundancy across network depth, making depth pruning an effective approach for improving inference efficiency. Existing one-shot pruning methods rely on local layer importance or fixed redundancy assumptions across architectures. We propose Locality-Aware Redundancy Pruning (LoRP), a training-free one-shot depth pruning framework guided by representation locality. We show that inter-layer redundancy can be either localized or globally distributed depending on the LLM architecture. To characterize this phenomenon, we introduce Representation Locality Score (RLS), derived from global inter-layer hidden-state similarity. Using a small calibration set, LoRP computes pairwise layer similarity, clusters layers by representational similarity, and allocates pruning according to residual intra-cluster redundancy. Experiments across diverse LLM families show improvements in both perplexity and downstream task accuracy. Official github repository: https://github.com/daniel-eai/LoRP-Locality-Aware-Redundancy-Pruning/

cs.LG↗

arXiv2026

Not Safe for All: Auditing the Dialect Penalty in Text-to-Image Safety Pipelines

Text-to-image (T2I) safety guardrails fail to generalize equitably to non-standard dialects. Evaluating 23,080 paired prompts across five English dialects, we formalize this failure as the dialect penalty, where filters trigger based on linguistic surface features rather than semantic intent. Text-level filters fail in opposing directions: NSFW-T over-flags benign dialect prompts and LatentGuard over-flags toxic ones (bias gaps up to +28.29 pp), while the OpenAI Moderation API under-detects them. A controlled typo ablation confirms this penalty originates from flagging dialectal features, not generic out-of-distribution sensitivity. The pixel-level generator is largely dialect-agnostic; the penalty enters at text processing and cascades unevenly to post-hoc guardrails. We show this bias tracks training data imbalance and is mitigable via group-balanced retraining, with an ablation attributing the gain to balanced exposure rather than to the worst-group objective of GroupDRO (group distributionally robust optimization). Current pipelines systematically fail dialect speakers, an equity failure masked by mean accuracy benchmarks. Our official code and dataset are publicly available at https://github.com/minguinho26/dialect-penalty-t2i. Content Warning: This paper contains offensive, toxic, or disturbing text prompts and generated images.

cs.AI↗

arXiv2026

On integral extensions between the abelianization functor and its symmetric powers

This paper aims to study Ext-groups between certain functors defined on the category of finitely generated free groups. Rational Ext-groups between the abelianization functor and its symmetric powers are known, and are almost always equal to zero. Recently, using homotopical methods, Arone constructed an explicit bounded complex whose homology corresponds to the integral Ext-groups between the abelianization functor and its symmetric powers. The homology of this complex is far from being trivial. Using this complex, we explicitly calculate some of these Ext-groups. More precisely, we compute Ext^1, Ext^2, Ext^{d-1} and Ext^{d-2} between the abelianization functor and its dth symmetric power. We further explain how Arone's complex can be obtained from an explicit projective resolution of the abelianization functor. We compare our results with the computation of Ext-groups between functors from finitely generated free abelian groups, obtained by Franjou and Pirashvili. In particular, we obtain that the composition with the abelianization functor induces an isomorphism for the Ext^1 considered in this paper.

math.AT↗

arXiv2026

PROPs associated to Lawvere theories and their relation to polynomial functors

Several adjunctions between functor categories have been studied and applied previously. These include Powell's adjunction between functor categories on free groups and on the linear PROP associated with the Lie operad, as well as those implicit in the equivalence of Pirashvili between functors on projective modules and modules over wreath products. In this paper, for a Lawvere theory $\mathcal{C}$ with a zero object, we construct a natural linear PROP $\tildeΦ_{\mathcal{C}}$ together with a canonical adjunction between $\mathcal{C}$-modules and $\tildeΦ_{\mathcal{C}}$-modules. We show that, under suitable conditions on $\mathcal{C}$, the adjunction is compatible with polynomial degree, inducing a correspondence between polynomial $\mathcal{C}$-modules and truncated $\tildeΦ_{\mathcal{C}}$-modules. This framework recovers the constructions of Powell and Pirashvili, while also yielding new examples, including adjunctions for functor categories on modules over a ring, as well as on free nilpotent groups of class $\leq c$ and, more generally, on free $\mathcal{R}$-semisimple groups, where $\mathcal{R}$ is a radical functor for groups.

math.AT↗