Search arXiv⌕ Search

arXiv subjects

Dipankar Ray

Publications and source records attributed to Dipankar Ray.

6 recordsLinked to original sources
arXiv2025

Effect of kappa-modified polarization force on Jeans instability in nonthermal EiBI-gravitating dust clouds

A semi-analytic model is developed to study the effects of kappa-distributed lighter constituents and the associated kappa-modified polarization force on the classical Jeans instability in dust molecular clouds (DMCs). The constitutive electrons and ions are considered to follow a nonthermal kappa-velocity distribution law, while the constitutive massive dust grains are treated as the EiBI-gravitating fluids. A linearized quadratic dispersion relation is derived using spherical normal mode analysis. The resulting dispersion relation and its corresponding modified instability criteria are analyzed in the hydrodynamic and kinetic regimes. The oscillatory and propagating mode characteristics are illustratively analyzed. It is seen that the EiBI gravity introduces a new velocity term in the dispersion relation. In contrast, the nonthermal kappa-distributed constituents significantly enhance the polarization force against their respective Maxwellian counterparts. The kappa-modified polarization force and the negative EiBI gravity parameter have destabilizing influences, unlike the positive EiBI parameter. An enhanced polarization interaction parameter and a positive EiBI parameter reduce the real normalized frequency. Consequently, the phase velocity exhibits strong dispersion, increasing with wavenumber until reaching saturation, after which it transitions into a weakly dispersive regime. These findings provide new insights into the formation of smaller astrophysical structures via the non-local Jeans instability in the ultracompact HII regions of dense DMCs.

astro-ph.GA↗
arXiv2024

Pulsational mode stability in complex EiBI-gravitating polarized astroclouds with (r, q)-distributed electrons

The pulsational mode of gravitational collapse (PMGC) originating from the combined gravito-electrostatic interaction in complex dust molecular clouds (DMCs) is a canonical mechanism leading to the onset of astronomical structure formation dynamics. A generalized semi-analytic model is formulated to explore the effects of the Eddington-inspired Born-Infeld (EiBI) gravity, non-thermal (r, q)-distributed electrons, and dust-polarization force on the PMGC stability concurrently. The thermal ions are treated thermo-statistically with the Maxwellian distribution law and the non-thermal electrons with the (r, q)-distribution law. The constitutive partially ionized dust grains are modeled in the fluid fabric. A spherical normal mode analysis yields a generalized linear PMGC dispersion relation. Its oscillatory and propagation characteristics are investigated in a reasonable numerical platform. It is found that an increase in the polarization force and positive EiBI parameter significantly enhances the instability, causing the DMC collapse and vice versa. The electron non-thermality spectral parameters play as vital stabilizing factors, and so on. Its reliability and applicability are finally outlined in light of astronomical predictions previously reported in the literature.

astro-ph.GA↗
arXiv2022

ToxiGen: A Large-Scale Machine-Generated Dataset for Adversarial and Implicit Hate Speech Detection

Toxic language detection systems often falsely flag text that contains minority group mentions as toxic, as those groups are often the targets of online hate. Such over-reliance on spurious correlations also causes systems to struggle with detecting implicitly toxic language. To help mitigate these issues, we create ToxiGen, a new large-scale and machine-generated dataset of 274k toxic and benign statements about 13 minority groups. We develop a demonstration-based prompting framework and an adversarial classifier-in-the-loop decoding method to generate subtly toxic and benign text with a massive pretrained language model. Controlling machine generation in this way allows ToxiGen to cover implicitly toxic text at a larger scale, and about more demographic groups, than previous resources of human-written text. We conduct a human evaluation on a challenging subset of ToxiGen and find that annotators struggle to distinguish machine-generated text from human-written language. We also find that 94.5% of toxic examples are labeled as hate speech by human annotators. Using three publicly-available datasets, we show that finetuning a toxicity classifier on our data improves its performance on human-written data substantially. We also demonstrate that ToxiGen can be used to fight machine-generated toxicity as finetuning improves the classifier significantly on our evaluation subset. Our code and data can be found at https://github.com/microsoft/ToxiGen.

cs.CL↗
arXiv2021

Hydrodynamic self-similar cosmological models

Hydrodynamic self-similar solutions, as obtained by Chi [J. Math. Phys. 24, 2532 (1983)] have been generalized by introducing new variables in place of the old space and time variables. A systematic procedure of obtaining a complete set of solutions has been suggested. The Newtonian analogs of all homogeneous isotropic Friedmann dust universes with spatial curvature $k = 0, \pm 1$ have been given.

gr-qc↗
arXiv2013

Counterfactual Reasoning and Learning Systems

This work shows how to leverage causal inference to understand the behavior of complex learning systems interacting with their environment and predict the consequences of changes to the system. Such predictions allow both humans and algorithms to select changes that improve both the short-term and long-term performance of such systems. This work is illustrated by experiments carried out on the ad placement system associated with the Bing search engine.

cs.LG↗