arXiv subjects

Igor Ostanin

Publications and source records attributed to Igor Ostanin.

14 recordsLinked to original sources

arXiv2026

Glite ARF: Verifier-Driven Research with Parallel LLM Coding Agents

LLM coding agents make it tempting to automate empirical research by delegating experiments to them directly, but naive delegation does not scale to large projects: low-rate instruction lapses compound into broken, irreproducible artefacts. To address this problem, we present Glite ARF, an open-source Python framework for running many LLM coding agents in parallel on a research repository without sacrificing reproducibility or auditability. The framework defines a three-role stack: a human researcher chooses which hypotheses to test, coding agents (Claude Code, Codex CLI) implement individual tasks under a fixed structure, and deterministic Python verifier scripts enforce task isolation, immutability of completed work, a corrections overlay, and a materialised project overview. We call this verifier-driven research: the rules of the research process live in code that fails loudly when violated, not in prose that agents are merely asked to follow. Using Glite ARF, we developed our submission to the BEA 2026 vocabulary-difficulty shared task, placing first in the closed track and second in the open track on all three target languages (Spanish, German, Mandarin) and reducing the official baseline RMSE by 29.9% (closed) and 35.9% (open). The campaign comprised 273 tracked tasks (146 experiment runs) across 129 feature sets, run by up to twelve parallel agents orchestrated from a single laptop - with some model training on rented A100s - at approximately \$450 in LLM API spend (\$498 total third-party cost), and structured per-fold provenance let us catch and strip four target-leaking feature sets, correcting an implausible 0.609 RMSE to 0.802. Across three campaigns in three domains, the framework's structural machinery adds only about 1% of wall-clock time. Framework and a public demo project accompany this paper.

cs.MA↗

arXiv2026

Granular mixing and flow dynamics in horizontal stirred bed reactors

Horizontal stirred bed reactors (HSBRs) are used in gas--phase polyolefin production, where efficient solids mixing and controlled residence time distributions are essential for product quality and stability. Despite their industrial relevance, the influence of operating conditions on granular flow and mixing in HSBRs is not well understood. Discrete Element Method (DEM) simulations are used to study the effects of rotation speed and fill level on particle motion, mixing, and axial transport in a lab--scale HSBR. An industrial--grade polypropylene powder is modelled using calibrated contact parameters. Mixing is quantified using the Lacey index in axial (z) and cross--sectional (xy) directions. Particle circulation is characterised via cycle--time analysis and a coarse--grained angular velocity field. Axial dispersion coefficients are obtained from particle trajectories using both Einstein--type and cycle--based approaches, and validated with a diffusion model predicting the axial Lacey index. Results show that axial mixing depends strongly on rotation speed and fill level: higher rotation speeds accelerate homogenization, while higher fill levels slow mixing. Cross--sectional mixing is mainly sensitive to rotation speed, with fill--level effects diminishing at higher speeds. Cycle time decreases with increasing rotation speed and fill level, indicating enhanced circulation. Axial dispersion increases with rotation speed but decreases with fill level, with consistent results across methods. These findings reveal trade--offs between axial mixing, circulation, and dispersion, highlighting the need to balance operating conditions and demonstrating the capability of DEM to support HSBR optimisation.

physics.comp-ph↗

arXiv2023

Arbitrary Controlled Re-Orientation of a Spinning Body by Evolving its Tensor of Inertia

Bodies with the nonspherical tensor of inertia exhibit a variety of rotational motion patterns, including chaotic motion, stable periodic (quasi-periodic) rotation, unstable rotation around the direction close to the body's second principal axis, featuring a well-known tennis-racket (also known as Garriott-Dzhanibekov) effect -- series of seemingly spontaneous 180 degrees flips. These patterns are even more complex if the body's tensor of inertia (TOI) is changing with time. Changing a body's TOI has been discussed recently as a tool to perform controllable Garriott-Dzhanibekov flips and similar maneuvers. In this work, the optimal control of the TOI of the body (spacecraft, or any other device that admits free rotation in three dimensions) is used as a means to perform desirable re-orientations of a body with respect to its angular velocity. Using the spherical TOI as the initial and final point of the maneuver, we optimize the parameters of the maneuver to achieve and stabilize the desired orientation of the body's principal axes with respect to spin angular velocity. It appears that such a procedure allows for finding arbitrarily complex maneuver trajectories of a spinning body. In particular, intermediate axis instability can be used to break the alignment of the body's principal axis and the axis of rotation. Such maneuvers do not require utilization of propellants and could be straightforwardly used for attitude control of a spin-stabilized spacecraft. The capabilities of such a method of angular maneuvering are demonstrated in numerical simulations.

nlin.CD↗

arXiv2023

How Fast are Domino Waves?

The paper is concerned with the problem of toppling propagation velocity in elastic, domino-like mechanical systems. We build on the work of Efthimiou and Johnson, who developed the theory of perfectly elastic collisions of thin rigid dominoes on a frictional foundation. This theory has been criticised for the lack of correspondence with the experimental observations, in particular, prediction of infinite propagation velocity for zero spacing between dominoes, as well as the inability to represent the collective nature of collisions in real domino systems. In our work we consider a more realistic scenario of dominoes of finite stiffness and obtain a theory of fast elastic domino waves, taking into account a limit velocity of the perturbation propagation in the system of dominoes. Moreover, finite collision time allows to extract dynamic quantities of collisions and establish upper and lower borders for domino separations where the theory could still be applied. Our discrete element simulations support our theoretical findings and shed light on the nature of collective interactions in the nearly-elastic domino chains.

physics.class-ph↗

arXiv2023

Rigid Clumps in the MercuryDPM Particle Dynamics Code

Discrete particle simulations have become the standard in science and industrial applications exploring the properties of particulate systems. Most of such simulations rely on the concept of interacting spherical particles to describe the properties of particulates, although, the correct representation of the nonspherical particle shape is crucial for a number of applications. In this work we describe the implementation of clumps, i.e. assemblies of rigidly connected spherical particles, which can approximate given nonspherical shapes, within the \textit{MercuryDPM} particle dynamics code. \textit{MercuryDPM} contact detection algorithm is particularly efficient for polydisperse particle systems, which is essential for multilevel clumps approximating complex surfaces. We employ the existing open-source \texttt{CLUMP} library to generate clump particles. We detail the pre-processing tools providing necessary initial data, as well as the necessary adjustments of the algorithms of contact detection, collision/migration and numerical time integration. The capabilities of our implementation are illustrated for a variety of examples.

math.NA↗

arXiv2022

Local ultra-densification of single-walled carbon nanotube films: modeling and experiment

Fabrication of nanostructured metasurfaces poses a significant technological and fundamental challenge. Despite developing novel systems that support reversible elongation and distortion, their nanoscale patterning and control of optical properties remain an open problem. Herein we report the atomic force microscope lithography (AFML) application for nanoscale patterning of single-walled carbon nanotube films and the associated reflection coefficient tuning. We present models of bundling reorganization, formed-pattern stability, and energy distribution describing mechanical behavior with mesoscopic distinct element method (MDEM). All observed and calculated phenomena support each other and present a platform for developing AFML patterned optical devices using meshy nanostructured matter.

cond-mat.mtrl-sci↗

arXiv2022

Simulation-Guided Optimization of Granular Phononic Crystal Structure Using the Discrete Element Method

The paper describes a novel methodology of designing granular phononic crystals for acoustic wave manipulations. A discrete element method is utilized to model the dynamics of a pulse wave propagating through the densely packed assembly of elastic spherical particles with an embedded phononic crystal - the region consisting of a certain arrangement of particles with varying densities. We suggest an optimization strategy that extremizes the useful properties of a granular phononic crystal, which are described in terms of a noise-proof functional based on frequency-wavenumber summation of spectral energy density. Few types of efficient phononic crystals are identified. The suggested methodology is of interest for a number of applications, in particular, for seismic shielding and selective sound absorption.

cond-mat.mtrl-sci↗

arXiv2020

Densification of Single-Walled Carbon Nanotube Films: Mesoscopic Distinct Element Method Simulations and Experimental Validation

Nanometer thin single-walled carbon nanotube (CNT) films collected from the aerosol chemical deposition reactors have gathered attention for their promising applications. Densification of these pristine films provides an important way to manipulate the mechanical, electronic, and optical properties. To elucidate the underlying microstructural level restructuring, which is ultimately responsible for the change in properties, we perform large scale vector-based mesoscopic distinct element method simulations in conjunction with electron microscopy and spectroscopic ellipsometry characterization of pristine and densified films by drop-cast volatile liquid processing. Matching the microscopy observations, pristine CNT films with finite thickness are modeled as self-assembled CNT networks comprising entangled dendritic bundles with branches extending down to individual CNTs. Simulations of the film under uniaxial compression uncover an ultra-soft densification regime extending to a ~75% strain, which is likely accessible with the surface tensional forces arising from liquid surface tension during the evaporation. When removing the loads, the pre-compressed samples evolve into homogeneously densified films with thickness values depending on both the pre-compression level and the sample microstructure. The significant reduction in thickness, confirmed by our spectroscopic ellipsometry, is attributed to the underlying structural changes occurring at the 100 nm scale, including the zipping of the thinnest dendritic branches.

cond-mat.mtrl-sci↗

arXiv2019

Computational Study of Ultrathin CNT Films with the Scalable Mesoscopic Distinct Element Method

In this work we present a computational study of the small strain mechanics of freestanding ultrathin CNT films under in-plane loading. The numerical modeling of the mechanics of representatively large specimens with realistic micro- and nanostructure is presented. Our simulations utilize the scalable implementation of the mesoscopic distinct element method of the waLBerla multi-physics framework. Within our modeling approach, CNTs are represented as chains of interacting rigid segments. Neighboring segments in the chain are connected with elastic bonds, resolving tension, bending, shear and torsional deformations. These bonds represent a covalent bonding within CNT surface and utilize Enhanced Vector Model (EVM) formalism. Segments of the neighboring CNTs interact with realistic coarse-grained anisotropic vdW potential, enabling relative slip of CNTs in contact. The advanced simulation technique allowed us to gain useful insights on the behavior of CNT materials. In particular, it was established that the energy dissipation during CNT sliding leads to extended load transfer that conditions material-like mechanical response of the weakly bonded assemblies of CNTs.

cond-mat.mes-hall↗

arXiv2019

A parameteric class of composites with a large achievable range of effective elastic properties

In this paper we investigate numerically an instance of the problem of G-closure for two-dimensional periodic metamaterials. Specifically, we consider composites with isotropic homogenized elasticity tensor, obtained as a mixture of two isotropic materials, focusing on the case of a single material with voids. This problem is important, in particular, in the context of designing small-scale structures for metamaterials in the context of additive fabrication, as this type of metamaterials makes it possible to obtain a range of material properties using a single base material. We demonstrate that two closely related simple parametric families based on the structure proposed by O. Sigmund attain good coverage of the space of isotropic properties satisfying Hashin-Shtrikman bounds. In particular, for positive Poisson ratio, we demonstrate that Hashin-Shtrikman bound can be approximated arbitrarily well, within limits imposed by numerical approximation: a strong evidence that these bounds are achievable in this case. For negative Poisson ratios, we numerically obtain a bound which we hypothesize to be close to optimal, at least for metamaterials with rotational symmetries of a regular triangle tiling.

physics.comp-ph↗

arXiv2019

Optimization-based approach to the calibration of mesoscale mechanical models for carbon nanotube systems

A new mesoscale mechanical model, describing elastic interactions in carbon nanotubes (CNT) and other nanofilaments, is proposed. Functional form of the developed model is based on enhanced vector model (EVM) that describes basic types of bond deformations: tension, torsion, bending and shear. Calibration of bond stiffnesses is performed by adjusting EVM parameters to reproduce both CNT's deformation energies and shape observed in a full-atomistic simulation. The parameters obtained are compared with the ones obtained from Euler-Bernoulli beam theory considerations. It is found that after certain critical length of a tested CNT specimen its stiffness parameters become length-independent and can be used in mesoscale simulations of CNTs of arbitrary length.

cond-mat.mes-hall↗

arXiv2016

What Lies Beneath the Surface: Topological-Shape Optimization With the Kernel-Independent Fast Multipole Method

The paper presents a new method for shape and topology optimization based on an efficient and scalable boundary integral formulation for elasticity. To optimize topology, our approach uses iterative extraction of isosurfaces of a topological derivative. The numerical solution of the elasticity boundary value problem at every iteration is performed with the boundary element formulation and the kernel-independent fast multipole method. Providing excellent single node performance, scalable parallelization and the best available asymptotic complexity, our method is among the fastest optimization tools available today. The performance of our approach is studied on few illustrative examples, including the optimization of engineered constructions for the minimum compliance and the optimization of the microstructure of a metamaterial for the desired macroscopic tensor of elasticity.

math.OC↗

arXiv2016

Time- and memory-efficient representation of complex mesoscale potentials

We apply the modern technique of approximation of multivariate functions - tensor train cross approximation - to the problem of the description of physical interactions between complex-shaped bodies in a context of computational nanomechanics. In this note we showcase one particular example - van der Waals interactions between two cylindrical bodies - relevant to modeling of carbon nanotube systems. The potential is viewed as a tensor (multidimensional table) which is represented in compact form with the help of tensor train decomposition. The described approach offers a universal solution for the description of van der Waals interactions between complex-shaped nanostructures and can be used within the framework of such systems of mesoscale modeling as recently emerged mesoscopic distinct element method (MDEM).

physics.comp-ph↗

arXiv2015

Toward Fast Topological-Shape Optimization With Boundary Elements

Wide variety of engineering design tasks can be formulated as constrained optimization problems where the shape and topology of the domain are optimized to reduce costs while satisfying certain constraints. Several mathematical approaches were developed to address the problem of finding optimal design of an engineered structure. Recent works have demonstrated the feasibility of boundary element method as a tool for topological-shape optimization. However, it was noted that the approach has certain drawbacks, and in particular high computational cost of the iterative optimization process. In this short note we suggest ways to address critical limitations of boundary element method as a tool for topological-shape optimization. We validate our approaches by supplementing the existing complex variables boundary element code for elastostatic problems with robust tools for fast topological-shape optimization. The efficiency of the approach is illustrated with a numerical example.

math.OC↗