arXiv subjects

Subham Ghosh

Publications and source records attributed to Subham Ghosh.

9 recordsLinked to original sources

arXiv2026

MatMMExtract: An Open-Source Pipeline for Panel-Level Extraction of Grounded Image-Text Pairs from Materials Science Literature

The materials science literature encodes decades of experimental knowledge in figures, yet this visual record remains locked away and inaccessible to AI at scale. The core difficulty is structural: most scientific figures are compound, with a single caption describing multiple sub-panels simultaneously, making direct image-text pairing unreliable. We present MatMMExtract, an end-to-end open-source pipeline that resolves this by decomposing compound figures into individual sub-panels and generating structured, grounded annotations using a large language model guided by a curated materials science taxonomy. Applied to 14,810 open-access articles, MatMMExtract produces MatSciFig; 391,606 panel-level image-text pairs from 180,571 figures, each annotated with a sub-caption, a two-level visualisation category spanning 19 classes and over 100 subtypes, and a scientific summary. To enable accurate panel localisation, we introduce MaterialScope, a domain-specific detection dataset of 2,811 manually annotated materials science figures, on which a fine-tuned YOLO12-m detector achieves mAP_50 of 0.9227. Among six benchmarked language models, Gemini 3.1 Flash Lite delivers the best cost-quality trade-off for annotation generation, with 82% of outputs rated good and a hallucination rate of 4.8%. A dual-encoder retrieval baseline on MatSciFig achieves a 4.4 times improvement in R@1 over zero-shot CLIP, demonstrating the dataset's immediate utility for vision-language learning. All resources are released openly to the community.

cs.CV↗

arXiv2025

Automated Extraction of Material Properties using LLM-based AI Agents

The rapid discovery of materials is constrained by the lack of large, machine-readable datasets that couple performance metrics with structural context. Existing databases are either small, manually curated, or biased toward first principles results, leaving experimental literature underexploited. We present an agentic, large language model (LLM)-driven workflow that autonomously extracts thermoelectric and structural-properties from about 10,000 full-text scientific articles. The pipeline integrates dynamic token allocation, zeroshot multi-agent extraction, and conditional table parsing to balance accuracy against computational cost. Benchmarking on 50 curated papers shows that GPT-4.1 achieves the highest accuracy (F1 = 0.91 for thermoelectric properties and 0.82 for structural fields), while GPT-4.1 Mini delivers nearly comparable performance (F1 = 0.89 and 0.81) at a fraction of the cost, enabling practical large scale deployment. Applying this workflow, we curated 27,822 temperature resolved property records with normalized units, spanning figure of merit (ZT), Seebeck coefficient, conductivity, resistivity, power factor, and thermal conductivity, together with structural attributes such as crystal class, space group, and doping strategy. Dataset analysis reproduces known thermoelectric trends, such as the superior performance of alloys over oxides and the advantage of p-type doping, while also surfacing broader structure-property correlations. To facilitate community access, we release an interactive web explorer with semantic filters, numeric queries, and CSV export. This study delivers the largest LLM-curated thermoelectric dataset to date, provides a reproducible and cost-profiled extraction pipeline, and establishes a foundation for scalable, data-driven materials discovery beyond thermoelectrics.

cs.LG↗

arXiv2024

Magnetic reconnection: an alternative explanation of radio emission in galaxy clusters

Observations of galaxy clusters show radio emission extended over almost the system scale, necessitating mechanisms for particle acceleration. Previous models for acceleration such as diffusive shock acceleration and that due to turbulence can fall short in terms of efficiency. In this letter, we propose the possibility of acceleration via magnetic reconnection. In particular, we invoke the plasmoid instability which has been previously applied to understand particle energization in high energy systems. Turbulence in galaxy clusters lead to fluctuation dynamos that are known to generate magnetic fields structures consisting of sharp reversals. These form natural sites of reconnection. We perform Particle-In-Cell (PIC) simulations of the plasmoid instability in collisionless and nonrelativistic plasmas. We show that the resulting electron energy spectra have power law indices that are consistent with that inferred from observations. Our estimates show that the acceleration timescales are much smaller than the lifetime of the reconnecting magnetic structures indicating the feasibility of our model. The synchrotron radio luminosity estimate is about $10^{41}$ ergs/s, agreeing with observations. Finally, we find that the maximum achievable Lorentz factor can go upto $10^5$ indicating that acceleration due magnetic reconnection is a promising avenue for understanding the origin of nonthermal emission in galaxy clusters.

astro-ph.CO↗

arXiv2024

Efficient Exact Algorithms for Minimum Covering of Orthogonal Polygons with Squares

Let $P$ be an orthogonal polygon of $n$ vertices, without holes. The Orthogonal Polygon Covering with Squares (OPCS) problem takes as input such an orthogonal polygon $P$ with integral vertex coordinates, and asks to find the minimum number of axis-parallel squares whose union is $P$ itself. [Aupperle et. al, 1988] provide an $\mathcal O(N^{1.5})$-time algorithm for OPCS, where $N$ is the number of integral lattice points lying in $P$. In their paper, designing algorithms for OPCS with a running time polynomial in $n$, was stated as an open question; $N$ can be arbitrarily larger than $n$. Output sensitive algorithms were known due to [Bar-Yehuda and Ben-Chanoch, 1994], but these fail to address the open question, as the output can be arbitrarily larger than $n$. We address this open question by designing a polynomial-time exact algorithm for OPCS with a worst-case running time of $\mathcal O(n^{10})$. We also consider the following structural parameterized version of the problem. Let a knob be a polygon edge whose both endpoints are convex polygon vertices. Given an input orthogonal polygon without holes that has $n$ vertices and at most $k$ knobs, we design an algorithm for OPCS with a worst-case running time $\mathcal O(n^2 + k^{10} \cdot n)$. This algorithm is more efficient than the former, whenever $k = o(n^{9/10})$. The problem of Orthogonal Polygon with Holes Covering with Squares (OPCSH) is also studied by [Aupperle et. al, 1988], where the input polygon could have holes. They claim a proof that OPCSH is NP-complete even when the input is the $N$ lattice points inside the polygon. We think there is an error in their proof, where an incorrect reduction from Planar 3-CNF is shown. We provide a correct reduction with a novel construction of one of the gadgets, and show how this leads to a correct proof of NP-completeness of OPCSH.

cs.CG↗

arXiv2022

The competition between the hydrodynamic instability from noise and magnetorotational instability in the Keplerian disks

We venture for the comparison between growth rates for magnetorotational instability (MRI) and hydrodynamics instability in the presence of an extra force in the local Keplerian accretion flow. The underlying model is described by the Orr-Sommerfeld and Squire equations in the presence of rotation, magnetic field and an extra force, plausibly noise with a nonzero mean. We obtain MRI using Wentzel-Kramers-Brillouin (WKB) approximation without extra force for purely vertical magnetic field and vertical wavevector of the perturbations. Expectedly, MRI is active within a range of magnetic field, which changes depending on the perturbation wavevector magnitude. Next, to check the effect of noise on the growth rates, a quartic dispersion relation has been obtained. Among those four solutions for growth rate, the one that reduces to MRI growth rate at the limit of vanishing mean of noise in the MRI active region of the magnetic field, is mostly dominated by MRI. However, in MRI inactive region, in the presence of noise the solution turns out to be unstable, which are almost independent of the magnetic field. Another growth rate, which is almost complementary to the previous one, leads to stability at the limit of vanishing noise. The remaining two growth rates, which correspond to the hydrodynamical growth rates at the limit of the vanishing magnetic field, are completely different from the MRI growth rate. More interestingly, the latter growth rates are larger than that of the MRI. If we consider viscosity, the growth rates decrease depending on the Reynolds number.

astro-ph.HE↗

arXiv2021

Hydrodynamical transport of angular momentum in accretion disks in the presence of nonlinear perturbations due to noise

The origin of hydrodynamical instability and turbulence in the Keplerian accretion disk is a long-standing puzzle. The flow therein is linearly stable. Here we explore the evolution of perturbation in this flow in the presence of an additional force. Such a force, which is expected to be stochastic in nature hence behaving as noise, could result from thermal fluctuations (however small be), grain-fluid interactions, feedback from outflows in astrophysical disks, etc. We essentially establish the evolution of nonlinear perturbation in the presence of Coriolis and external forces, which is the modified Landau equation. We obtain that even in the linear regime, under suitable forcing and Reynolds number, the otherwise least stable perturbation evolves to a very large saturated amplitude, leading to nonlinearity and plausible turbulence. Hence, forcing essentially leads a linear stable mode to unstable. We further show that nonlinear perturbation diverges at a shorter time-scale in the presence of force, leading to a fast transition to turbulence. Interestingly, the emergence of nonlinearity depends only on the force but not on the initial amplitude of perturbation, unlike the original Landau equation-based solution.

astro-ph.HE↗

arXiv2021

Forced linear shear flows with rotation: rotating Couette-Poiseuille flow, its stability and astrophysical implications

We explore the effect of forcing on the linear shear flow or plane Couette flow, which is also the background flow in the very small region of the Keplerian accretion disk. We show that depending on the strength of forcing and boundary conditions suitable for the systems under consideration, the background plane shear flow and, hence, accretion disk velocity profile modifies to parabolic flow, which is plane Poiseuille flow or Couette-Poiseuille flow, depending on the frame of reference. In the presence of rotation, plane Poiseuille flow becomes unstable at a smaller Reynolds number under pure vertical as well as threedimensional perturbations. Hence, while rotation stabilizes plane Couette flow, the same destabilizes plane Poiseuille flow faster and forced local accretion disk. Depending on the various factors, when local linear shear flow becomes Poiseuille flow in the shearing box due to the presence of extra force, the flow becomes unstable even for the Keplerian rotation and hence turbulence will pop in there. This helps in resolving a long standing problem of sub-critical transition to turbulence in hydrodynamic accretion disks and laboratory plane Couette flow.

physics.flu-dyn↗

arXiv2020

Origin of hydrodynamic instability from noise: from laboratory flow to accretion disk

We attempt to address the old problem of plane shear flows: the origin of turbulence and hence transport of angular momentum in accretion flows as well as laboratory flows, such as plane Couette flow. We undertake the problem by introducing an extra force in Orr-Sommerfeld and Squire equations along with the Coriolis force mimicking the local region of the accretion disk. For plane Couette flow, the Coriolis term drops. Subsequently we solve the equations by WKB approximation method. We investigate the dispersion relation for the Keplerian flow and plane Couette flow for all possible combinations of wave vectors. Due to the very presence of extra force, we show that both the flows are unstable for a certain range of wave vectors. However, the nature of instability between the flows is different. We also study the Argand diagrams of the perturbation eigenmodes. It helps us to compare the different time scales corresponding to the perturbations as well as accretion. We ultimately conclude with this formalism that fluid gets enough time to be unstable and hence plausibly turbulent particularly in the local regime of the Keplerian accretion disks. Repetition of the analysis throughout the disk explains the transport of angular momentum and matter along outward and inward direction respectively.

astro-ph.HE↗

arXiv2020

Hydrodynamical instability with noise in the Keplerian accretion discs: Modified Landau equation

Origin of hydrodynamical instability and turbulence in the Keplerian accretion disc as well as similar laboratory shear flows, e.g. plane Couette flow, is a long standing puzzle. These flows are linearly stable. Here we explore the evolution of perturbation in such flows in the presence of an additional force. Such a force, which is expected to be stochastic in nature hence behaving as noise, could be result of thermal fluctuations (however small be), Brownian ratchet, grain-fluid interactions and feedback from outflows in astrophysical discs etc. We essentially establish the evolution of nonlinear perturbation in the presence of Coriolis and external forces, which is modified Landau equation. We show that even in the linear regime, under suitable forcing and Reynolds number, the otherwise least stable perturbation evolves to a very large saturated amplitude, leading to nonlinearity and plausible turbulence. Hence, forcing essentially leads a linear stable mode to unstable. We further show that nonlinear perturbation diverges at a shorter timescale in the presence of force, leading to a fast transition to turbulence. Interestingly, emergence of nonlinearity depends only on the force but not on the initial amplitude of perturbation, unlike original Landau equation based solution.

astro-ph.HE↗