Searcharxiv⌕ Search

arXiv subjects

Arun Prakash

Publications and source records attributed to Arun Prakash.

11 recordsLinked to original sources

Procedure without Judgment: What Small Language Models Learn under an Exact Verifier

When an authorization policy blocks an agent's tool call, the audit log should name the facts that forced the decision and nothing more. For an executable policy over an enumerable context space this minimal sufficient reason can be computed exactly, so an explanation is scored against ground truth rather than a proxy for it. We use that exactness to ask what a model can learn from a verifier that has no error in it. The explainer sees one request and one decision, never the policy, and has thirty turns in which it may call the policy on altered requests and ask whether a candidate set of facts is sufficient. Frontier models find the exact reason about half the time with the tools alone, and one sentence telling them to shrink the answer lifts Haiku 4.5 from 58% to 75%. A 4B model fine-tuned on 2000 traces of the deletion algorithm reaches 92% on the policy it was trained on. Teacher kind is what moves this: at matched data size the algorithm-taught student leads the frontier-taught one in all three training seeds, and the two arms never overlap. What the student learns is a fixed schedule of sufficiency checks, and the schedule does not travel. On three public policies it never saw, its answers stay sufficient (94%) and stop being minimal (48%), while Haiku under the same sentence is exact on all 47 episodes. Reinforcement learning from the exact reward leaves held-out accuracy where it was and inverts the strategy, from probing the policy to spending the entire budget on sufficiency checks. The oracle made the effect of training legible, but it did not make training better.

cs.AI↗

Optimistic Verifiers Get Hacked: Reward Hacking Under an Exact Process Oracle

Reinforcement learning from verifiable rewards inherits whatever its verifier gets wrong. Outside mathematics and code the verifier is a proxy, a rubric, a learned judge or a hand-written checker, and its distance from the true objective is never measured, so a rising reward cannot be told apart from a hack. We build a setting where the true reward is exactly computable at the process level: executable authorization policies over 124,416 contexts admit an oracle that lists every minimal sufficient reason for every decision. We compile these policies into exact process rewards, degrade the verifier along seven realistic axes (outcome-only, sufficiency, recall-only, extraction noise, coarse taxonomy, stale policy, LLM judge), and train against each while the exact verifier scores every rollout. Four phenotypes appear. Optimistic verifiers are hacked by padding: the sufficiency proxy ends at 0.940 while true faithfulness falls from 0.731 to 0.667 on 96 held-out episodes. Outcome-only verifiers saturate and stop teaching, noise dilutes, and ranking-preserving gaps do no harm. A model-free audit of the verifier, its selection error and gap sign, together with a short saturation pilot, anticipates which phenotype a run will produce. Across nine configurations spanning two frameworks, two model vendors, two adapter sizes, three precisions, two initialisations and two seeds, the phenotypes recur, sufficiency damage is positive in all 10 replication cells and beyond two standard errors in 7, and optimistic gaps remain the steeper branch. The damage magnitude does not replicate, and five pre-registered attempts to explain it are reported as they fell. We release the testbed, the training instrumentation, and 14 real public policies (Cedar, OPA, Gatekeeper) ported to executable oracles.

cs.LG↗

First Plasma Commissioning and Operational Highlights from India's First Spherical Tokamak at IPR

A compact Spherical Tokamak(ST) is commissioned at Institute for Plasma Research (IPR) to explore low aspect ratio tokamak physics and technologies that complement to the existing high aspect ratio tokamaks namely ADITYA-U and SST-1 by enabling studies on non-inductive startup, current drive in over dense plasmas, and shaped plasma physics on a low cost platform. The device, India's first spherical tokamak has completed major mechanical, magnetic, and electrical integration, and the coil system has been successfully tested with series of integrated commissioning. First plasma experiments have been carried out with a modest Ohmic system assisted by a 2.45GHz microwave system, supported by a centralized control and data acquisition system. An initial diagnostic set comprising visible imaging, spectroscopy, magnetics, and radiation monitors required for machine operation has been installed. This paper presents the integrated commissioning experiences and first plasma experiments of the newly installed machine.

physics.plasm-ph↗

MA-SBI: Misspecification-Aware Simulation-Based Inference via Side-Channel Guidance

Simulation-based inference (SBI) of latent parameters is often hindered by simulator misspecification, the mismatch between simulated and real-world observations caused by inherent modeling simplifications. RoPE, the recent state-of-the-art for robust SBI, addresses this through optimal transport between learned representations of real and simulated observations, but requires ground-truth parameter calibration pairs that are typically unavailable in the very settings where SBI is needed. What practitioners do have is unstructured side-information such as regime labels, instruction text, and policy bulletins. We propose Misspecification-Aware Simulation-Based Inference (MA-SBI), a calibration-free framework that turns this side-channel into a posterior correction. A learned corrector maps side-channel text to an observation-space shift applied before any pre-trained amortized posterior, requiring no retraining and no parameter ground-truth. Our main theorem bounds achievable bias reduction by the mutual information between misspecification and side-channel, with a non-vacuous constant that extends to all sub-Gaussian noise via Donsker-Varadhan. On hide-the-calibration benchmarks, MA-SBI with text alone matches the oracle posterior across 10 seeds and two backbones (TOST equivalence), while RoPE given more data does not. The two approaches are complementary: where misspecification is structural and recoverable from parameter pairs, RoPE dominates, as the theory predicts. A stochastic variant improves posterior-predictive log-likelihood on real COVID and OxCGRT epidemiological data, and correctly leaves the posterior unchanged on a well-specified cognitive-science corpus.

cs.AI↗

RAID: Semantic Graph Diffusion for True Cold-Start and Cross-Lingual Forecasting

Time-series foundation models show strong transfer performance when given a non-empty history window. However, true cold-start scenarios, where a new item has no prior observations, violate this assumption. We propose RAID (Retrieval-Augmented Iterative Diffusion) a framework, which replaces history-based correlation learning with metadata-driven semantic retrieval and graph-conditioned diffusion. RAID maps textual metadata into a shared semantic space using a frozen multilingual embedding model and constructs an inductive retrieval graph that extends naturally to unseen items. It first forms a base forecast by aggregating information from semantically related neighbors, then refines this forecast with a gated diffusion module to model residual uncertainty. Under a strict true cold-start protocol, RAID outperforms strong foundation models and competitive baselines on both forecasting accuracy and prediction interval coverage, while reducing inference latency by an order of magnitude through non-autoregressive decoding. The shared semantic space also enables zero-shot cross-lingual transfer, allowing a model trained on English descriptions to generalize to items described in other languages without direct supervision.

cs.AI↗

Analog Solver Circuit for Linear Symmetric Positive-Definite Systems at a Complexity Independent of Matrix Size

Accelerating the solution of linear systems of equations is critical due to their central role in numerous applications, such as numerical simulations, data analytics, and machine learning. This paper presents an analog solver circuit designed to accelerate the solution of symmetric positive definite (SPD) linear systems of equations. The proposed design leverages noninverting operational amplifier configurations to create a negative resistance circuit, effectively modeling any symmetric system. The paper details the principles behind the design, optimizations of the system architecture, and numerical results that demonstrate the robustness of the design. The findings reveal that the proposed system solves symmetric diagonally dominant (SDD) matrices with O(1) complexity, achieving the theoretical maximum speed as the circuit relies solely on resistors. For non-diagonally dominant SPD systems, the solution speed depends on matrix properties, specifically eigenvalues and diagonal dominance deviation, but remains independent of the size of the matrix.

cs.AR↗

Scalar auxiliary variable (SAV) stabilization of implicit-explicit (IMEX) time integration schemes for nonlinear structural dynamics

Implicit-explicit (IMEX) time integration schemes are well suited for nonlinear structural dynamics because of their low computational cost and high accuracy. However, stability of IMEX schemes cannot be guaranteed for general nonlinear problems. In this article, we present a scalar auxiliary variable (SAV) stabilization of high-order IMEX time integration schemes that leads to unconditional stability. The proposed IMEX-BDFk-SAV schemes treat linear terms implicitly using kth-order backward difference formulas (BDFk) and nonlinear terms explicitly. This eliminates the need for iterations in nonlinear problems and leads to low computational cost. Truncation error analysis of the proposed IMEX-BDFk-SAV schemes confirms that up to kth-order accuracy can be achieved and this is verified through a series of convergence tests. Unlike existing SAV schemes for first-order ordinary differential equations (ODEs), we introduce a novel SAV for the proposed schemes that allows direct solution of the second-order ODEs without transforming them to a system of first-order ODEs. Finally, we demonstrate the performance of the proposed schemes by solving several nonlinear problems in structural dynamics and show that the proposed schemes can achieve high accuracy at a low computational cost while maintaining unconditional stability.

math.NA↗

Flow-driven spectral chaos (FSC) method for long-time integration of second-order stochastic dynamical systems

For decades, uncertainty quantification techniques based on the spectral approach have been demonstrated to be computationally more efficient than the Monte Carlo method for a wide variety of problems, particularly when the dimensionality of the probability space is relatively low. The time-dependent generalized polynomial chaos (TD-gPC) is one such technique that uses an evolving orthogonal basis to better represent the stochastic part of the solution space in time. In this paper, we present a new numerical method that uses the concept of 'enriched stochastic flow maps' to track the evolution of the stochastic part of the solution space in time. The computational cost of this proposed flow-driven stochastic chaos (FSC) method is an order of magnitude lower than TD-gPC for comparable solution accuracy. This gain in computational cost is realized because, unlike most existing methods, the number of basis vectors required to track the stochastic part of the solution space, and consequently the computational cost associated with the solution of the resulting system of equations, does not depend upon the dimensionality of the probability space. Four representative numerical examples are presented to demonstrate the performance of the FSC method for long-time integration of second-order stochastic dynamical systems in the context of stochastic dynamics of structures.

math.NA↗

Flow-driven spectral chaos (FSC) method for simulating long-time dynamics of arbitrary-order non-linear stochastic dynamical systems

Uncertainty quantification techniques such as the time-dependent generalized polynomial chaos (TD-gPC) use an adaptive orthogonal basis to better represent the stochastic part of the solution space (aka random function space) in time. However, because the random function space is constructed using tensor products, TD-gPC-based methods are known to suffer from the curse of dimensionality. In this paper, we introduce a new numerical method called the 'flow-driven spectral chaos' (FSC) which overcomes this curse of dimensionality at the random-function-space level. The proposed method is not only computationally more efficient than existing TD-gPC-based methods but is also far more accurate. The FSC method uses the concept of 'enriched stochastic flow maps' to track the evolution of a finite-dimensional random function space efficiently in time. To transfer the probability information from one random function space to another, two approaches are developed and studied herein. In the first approach, the probability information is transferred in the mean-square sense, whereas in the second approach the transfer is done exactly using a new theorem that was developed for this purpose. The FSC method can quantify uncertainties with high fidelity, especially for the long-time response of stochastic dynamical systems governed by ODEs of arbitrary order. Six representative numerical examples, including a nonlinear problem (the Van-der-Pol oscillator), are presented to demonstrate the performance of the FSC method and corroborate the claims of its superior numerical properties. Finally, a parametric, high-dimensional stochastic problem is used to demonstrate that when the FSC method is used in conjunction with Monte Carlo integration, the curse of dimensionality can be overcome altogether.

math.NA↗

Multi-element flow-driven spectral chaos (ME-FSC) method for uncertainty quantification of dynamical systems

The flow-driven spectral chaos (FSC) is a recently developed method for tracking and quantifying uncertainties in the long-time response of stochastic dynamical systems using the spectral approach. The method uses a novel concept called 'enriched stochastic flow maps' as a means to construct an evolving finite-dimensional random function space that is both accurate and computationally efficient in time. In this paper, we present a multi-element version of the FSC method (the ME-FSC method for short) to tackle (mainly) those dynamical systems that are inherently discontinuous over the probability space. In ME-FSC, the random domain is partitioned into several elements, and then the problem is solved separately on each random element using the FSC method. Subsequently, results are aggregated to compute the probability moments of interest using the law of total probability. To demonstrate the effectiveness of the ME-FSC method in dealing with discontinuities and long-time integration of stochastic dynamical systems, four representative numerical examples are presented in this paper, including the Van-der-Pol oscillator problem and the Kraichnan-Orszag three-mode problem. Results show that the ME-FSC method is capable of solving problems that have strong nonlinear dependencies over the probability space, both reliably and at low computational cost.

math.NA↗

Sintering of Alumina Nanoparticles: Comparison of Interatomic Potentials, Molecular Dynamics Simulations, and Data Analysis

Sintering of alumina nanoparticles is of interest both from the view of fundamental research as well as for industrial applications. Atomistic simulations are tailor-made for understanding and predicting the time- and temperature-dependent sintering behaviour. However, the quality and predictability of such analysis is strongly dependent on the performance of the underlying interatomic potentials. In this work, we investigate and benchmark four empirical interatomic potentials and discuss the resulting properties and drawbacks based on experimental and density functional theory data from the literature. The potentials, which have different origins and formulations, are then used in molecular dynamics simulations to perform a systematic study of the sintering process. To analyse the results, we develop a number of tailored data analysis approaches that are able to characterise and quantify the sintering process. Subsequently, the disparities in the sintering behaviour predicted by the potentials are critically discussed. Finally, we conclude by providing explanations for the differences in performance of the potentials, together with recommendations for molecular dynamics sintering simulations of alumina.

cond-mat.mtrl-sci↗