arXiv subjects

Andrei Zakharov

Publications and source records attributed to Andrei Zakharov.

7 recordsLinked to original sources

arXiv2025

Hierarchical AI-Meteorologist: LLM-Agent System for Multi-Scale and Explainable Weather Forecast Reporting

We present the Hierarchical AI-Meteorologist, an LLM-agent system that generates explainable weather reports using a hierarchical forecast reasoning and weather keyword generation. Unlike standard approaches that treat forecasts as flat time series, our framework performs multi-scale reasoning across hourly, 6-hour, and daily aggregations to capture both short-term dynamics and long-term trends. Its core reasoning agent converts structured meteorological inputs into coherent narratives while simultaneously extracting a few keywords effectively summarizing the dominant meteorological events. These keywords serve as semantic anchors for validating consistency, temporal coherence and factual alignment of the generated reports. Using OpenWeather and Meteostat data, we demonstrate that hierarchical context and keyword-based validation substantially improve interpretability and robustness of LLM-generated weather narratives, offering a reproducible framework for semantic evaluation of automated meteorological reporting and advancing agent-based scientific reasoning.

cs.AI↗

arXiv2025

A Modular LLM-Agent System for Transparent Multi-Parameter Weather Interpretation

Weather forecasting is not only a predictive task but an interpretive scientific process requiring explanation, contextualization, and hypothesis generation. This paper introduces AI-Meteorologist, an explainable LLM-agent framework that converts raw numerical forecasts into scientifically grounded narrative reports with transparent reasoning steps. Unlike conventional forecast outputs presented as dense tables or unstructured time series, our system performs agent-based analysis across multiple meteorological variables, integrates historical climatological context, and generates structured explanations that identify weather fronts, anomalies, and localized dynamics. The architecture relies entirely on in-context prompting, without fine-tuning, demonstrating that interpretability can be achieved through reasoning rather than parameter updates. Through case studies on multi-location forecast data, we show how AI-Meteorologist not only communicates weather events but also reveals the underlying atmospheric drivers, offering a pathway toward AI systems that augment human meteorological expertise and support scientific discovery in climate analytics.

cs.CY↗

arXiv2023

Sequences of dislocation reactions and helicity transformations in tubular crystals

Freestanding tubular crystals offer a general description of crystalline order on deformable surfaces with cylindrical topology, such as single-walled carbon nanotubes, microtubules, and recently reported colloidal assemblies. These systems exhibit a rich interplay between the crystal's helicity on its periodic surface, the deformable geometry of that surface, and the motions of topological defects within the crystal. Previously, in simulations of tubular crystals as elastic networks, we found that dislocations in nontrivial patterns can co-stabilize with kinks in the tube shape, producing mechanical multistability. Here, we extend that work with detailed Langevin dynamics simulations, in order to explore defect dynamics efficiently and without the constraints imposed by elastic network models. Along with the predicted multistability of dislocation glide, we find a variety of irreversible defect transformations, including vacancy formation, particle extrusions, and "reactions" that reorient dislocation pairs. Moreover, we report spontaneous sequences of several such defect transformations, which are unique to tubular crystals. We demonstrate a simple method for controlling these sequences through a time-varying external force.

cond-mat.soft↗

arXiv2021

Shape multistability in flexible tubular crystals through interactions of mobile dislocations

We study avenues to shape multistability and shape-morphing in flexible crystalline membranes of cylindrical topology, enabled by glide mobility of dislocations. Using computational modeling, we obtain states of mechanical equilibrium presenting a wide variety of tubular crystal deformation geometries, due to an interplay of effective defect interactions with out-of-tangent-plane deformations that reorient the tube axis. Importantly, this interplay often stabilizes defect configurations quite distinct from those predicted for a two-dimensional crystal confined to the surface of a rigid cylinder. We find that relative and absolute stability of competing states depend strongly on control parameters such as bending rigidity, applied stress, and spontaneous curvature. Using stable dislocation pair arrangements as building blocks, we demonstrate that targeted macroscopic three-dimensional conformations of thin crystalline tubes can be programmed by imposing certain sparse patterns of defects. Our findings reveal a broad design space for controllable and reconfigurable colloidal tube geometries, with potential relevance also to carbon nanotubes and microtubules.

cond-mat.soft↗

arXiv2021

Modeling mechanochemical pattern formation in elastic sheets of biological matter

Inspired by active shape morphing in developing tissues and biomaterials, we investigate two generic mechanochemical models where the deformations of a thin elastic sheet are driven by, and in turn affect, the concentration gradients of a chemical signal. We develop numerical methods to study the coupled elastic deformations and chemical concentration kinetics, and illustrate with computations the formation of different patterns depending on shell thickness, strength of mechanochemical coupling and diffusivity. In the first model, the sheet curvature governs the production of a contractility inhibitor and depending on the threshold in the coupling, qualitatively different patterns occur. The second model is based on the stress--dependent activity of myosin motors, and demonstrates how the concentration distribution patterns of molecular motors are affected by the long-range deformations generated by them. Since the propagation of mechanical deformations is typically faster than chemical kinetics (of molecular motors or signaling agents that affect motors), we describe in detail and implement a numerical method based on separation of timescales to effectively simulate such systems. We show that mechanochemical coupling leads to long-range propagation of patterns in disparate systems through elastic instabilities even without the diffusive or advective transport of the chemicals.

cond-mat.soft↗

arXiv2020

Mechanochemical induction of wrinkling morphogenesis on elastic shells

Morphogenetic dynamics of tissue sheets require coordinated cell shape changes regulated by global patterning of mechanical forces. Inspired by such biological phenomena, we propose a minimal mechanochemical model based on the notion that cell shape changes are induced by diffusible biomolecules that influence tissue contractility in a concentration-dependent manner -- and whose concentration is in turn affected by the macroscopic tissue shape. We perform computational simulations of thin shell elastic dynamics to reveal propagating chemical and three-dimensional deformation patterns arising due to a sequence of buckling instabilities. Depending on the concentration threshold that actuates cell shape change, we find qualitatively different patterns. The mechanochemically coupled patterning dynamics are distinct from those driven by purely mechanical or purely chemical factors. Using numerical simulations and theoretical arguments, we analyze the elastic instabilities that result from our model and provide simple scaling laws to identify wrinkling morphologies.

cond-mat.soft↗

arXiv2019

Fiber-based shape-morphing architectures

We describe a combined experimental and theoretical investigation of shape-morphing structures assembled by actuating composite (Janus) fibers, taking into account multiple relevant factors affecting shape transformations, such as strain rate, composition, and geometry of the structures. Starting with simple bending experiments, we demonstrate the ways to attain multiple out-of-plane shapes of closed rings and square frames. Through combining theory and simulation, we examine how the mechanical properties of Janus fibers affect shape transitions. This allows us to control shape changes and to attain target 3D shapes by precise tuning of the material properties and geometry of the fibers. Our results open new perspectives of design of advanced mechanical metamaterials capable to create elaborate structures through sophisticated actuation modes.

cond-mat.soft↗