arXiv subjects

Pankaj Mishra

Publications and source records attributed to Pankaj Mishra.

At least 19 recordsLinked to original sources

arXiv2025

Dispersion of active particles in oscillatory Poiseuille flow

Active particles exhibit complex transport dynamics in flows through confined geometries such as channels or pores. In this work, we employ a generalized Taylor dispersion (GTD) theory to study the long-time dispersion behavior of active Brownian particles (ABPs) in an oscillatory Poiseuille flow within a planar channel. We quantify the time-averaged longitudinal dispersion coefficient as a function of the flow speed, flow oscillation frequency, and particle activity. In the weak-activity limit, asymptotic analysis shows that activity can either enhance or hinder the dispersion compared to the passive case. For arbitrary activity levels, we numerically solve the GTD equations and validate the results with Brownian dynamics simulations. We show that the dispersion coefficient could vary non-monotonically with both the flow speed and particle activity. Furthermore, the dispersion coefficient shows an oscillatory behavior as a function of the flow oscillation frequency, exhibiting distinct minima and maxima at different frequencies. The observed oscillatory dispersion results from the interplay between self-propulsion and oscillatory flow advection -- a coupling absent in passive or steady systems. Our results show that time-dependent flows can be used to tune the dispersion of active particles in confinement.

physics.flu-dyn↗

arXiv2022

Masked Transformer for image Anomaly Localization

Image anomaly detection consists in detecting images or image portions that are visually different from the majority of the samples in a dataset. The task is of practical importance for various real-life applications like biomedical image analysis, visual inspection in industrial production, banking, traffic management, etc. Most of the current deep learning approaches rely on image reconstruction: the input image is projected in some latent space and then reconstructed, assuming that the network (mostly trained on normal data) will not be able to reconstruct the anomalous portions. However, this assumption does not always hold. We thus propose a new model based on the Vision Transformer architecture with patch masking: the input image is split in several patches, and each patch is reconstructed only from the surrounding data, thus ignoring the potentially anomalous information contained in the patch itself. We then show that multi-resolution patches and their collective embeddings provide a large improvement in the model's performance compared to the exclusive use of the traditional square patches. The proposed model has been tested on popular anomaly detection datasets such as MVTec and head CT and achieved good results when compared to other state-of-the-art approaches.

cs.CV↗

arXiv2021

Fairness based Multi-Preference Resource Allocation in Decentralised Open Markets

In this work, we focus on resource allocation in a decentralised open market. In decentralised open markets consists of multiple vendors and multiple dynamically-arriving buyers, thus makes the market complex and dynamic. Because, in these markets, negotiations among vendors and buyers take place over multiple conflicting issues such as price, scalability, robustness, delay, etc. As a result, optimising the resource allocation in such open markets becomes directly dependent on two key decisions, which are; incorporating a different kind of buyers' preferences, and fairness based vendor elicitation strategy. Towards this end, in this work, we propose a three-step resource allocation approach that employs a reverse-auction paradigm. At the first step, priority label is attached to each bidding vendor based on the proposed priority mechanism. Then, at the second step, the preference score is calculated for all the different kinds of preferences of the buyers. Finally, at the third step, based on the priority label of the vendor and the preference score winner is determined. Finally, we compare the proposed approach with two state-of-the-art resource pricing and allocation strategies. The experimental results show that the proposed approach outperforms the other two resource allocation approaches in terms of the independent utilities of buyers and the overall utility of the open market.

cs.GT↗

arXiv2021

VT-ADL: A Vision Transformer Network for Image Anomaly Detection and Localization

We present a transformer-based image anomaly detection and localization network. Our proposed model is a combination of a reconstruction-based approach and patch embedding. The use of transformer networks helps to preserve the spatial information of the embedded patches, which are later processed by a Gaussian mixture density network to localize the anomalous areas. In addition, we also publish BTAD, a real-world industrial anomaly dataset. Our results are compared with other state-of-the-art algorithms using publicly available datasets like MNIST and MVTec.

cs.CV↗

arXiv2020

Image Anomaly Detection by Aggregating Deep Pyramidal Representations

Anomaly detection consists in identifying, within a dataset, those samples that significantly differ from the majority of the data, representing the normal class. It has many practical applications, e.g. ranging from defective product detection in industrial systems to medical imaging. This paper focuses on image anomaly detection using a deep neural network with multiple pyramid levels to analyze the image features at different scales. We propose a network based on encoding-decoding scheme, using a standard convolutional autoencoders, trained on normal data only in order to build a model of normality. Anomalies can be detected by the inability of the network to reconstruct its input. Experimental results show a good accuracy on MNIST, FMNIST and the recent MVTec Anomaly Detection dataset

cs.CV↗

arXiv2020

Horizon induces instability locally and creates quantum thermality

The classical Hamiltonian for a chargeless and massless particle in a very near horizon region is shown to be of the form $H\sim xp$ as long as radial motion is concerned. This is demonstrated explicitly for static spherically symmetric black hole and also found to be applicable for specific choice of radial trajectories in the Kerr case. Such feature of horizon leads to unavoidable "{\it local instability}" in the particle's radial motion as long as near horizon regime is concerned. We show that at the quantum level this provides thermality in the system. The temperature is found to be given by the Hawking expression. Finally, we conjecture that the automatic instability created by the horizon is responsible for its own temperature and consequently can be a possible physical mechanism for horizon temperature.

gr-qc↗

arXiv2020

Induction of chaotic fluctuations in particle dynamics in a uniformly accelerated frame

The ongoing conjecture that the presence of horizon may induce chaos in an integrable system, is further investigated from the perspective of a uniformly accelerated frame. Particularly, we build up a model which consists of a particle (massless and chargeless) trapped in harmonic oscillator in a uniformly accelerated frame (namely Rindler observer). Here the Rindler frame provides a Killing horizon without any intrinsic curvature to the system. This makes the present observations different from previous studies. We observe that for some particular values of parameters of the system (like acceleration, energy of the particle), the motion of the particle trapped in harmonic potential systematically goes from periodic state to the chaotic. This indicates that the existence of horizon alone, not the intrinsic curvature (i.e. the gravitational effect) in the background, is sufficient to induce the chaotic motion in the particle. We believe the present study further enlighten and balustrade the conjecture.

gr-qc↗

arXiv2019

Image anomaly detection with capsule networks and imbalanced datasets

Image anomaly detection consists in finding images with anomalous, unusual patterns with respect to a set of normal data. Anomaly detection can be applied to several fields and has numerous practical applications, e.g. in industrial inspection, medical imaging, security enforcement, etc.. However, anomaly detection techniques often still rely on traditional approaches such as one-class Support Vector Machines, while the topic has not been fully developed yet in the context of modern deep learning approaches. In this paper, we propose an image anomaly detection system based on capsule networks under the assumption that anomalous data are available for training but their amount is scarce.

cs.CV↗

arXiv2018

Presence of horizon makes particle motion chaotic

We analyze the motion of a {\it massless} and {\it chargeless} particle very near to the event horizon. It reveals that the radial motion has exponential growing nature which indicates that there is a possibility of inducing chaos in the particle motion of an integrable system when it comes under the influence of the horizon. This is being confirmed by investigating the Poincar$\acute{e}$ section of the trajectories with the introduction of a harmonic trap to confine the particle's motion. Two situations are investigated: (a) {\it any} static, spherically symmetric black hole and, (b) spacetime represents a stationary, axisymmetric black hole (e.g., Kerr metric). In both cases, the largest Lyapunov exponent has upper bound which is the surface gravity of the horizon. We find that the inclusion of rotation in the spacetime introduces more chaotic fluctuations in the system. The possible implications are finally discussed.

gr-qc↗

arXiv2016

Coupled dynamics for superfluid $^4$He in the channel

We study the coupled dynamics of normal and superfluid components of the superfluid $^4$He in the channel considering the counterflow turbulence with laminar normal component. In particular, we calculated profiles of the normal velocity, the mutual friction, the vortex line density and other flow properties and compared them to the case when the dynamic of the normal component is "frozen". We have found that the coupling between the normal and superfluid components leads to flattening of the normal velocity profile, increasingly more pronounced with temperature, as the mutual friction, and therefore coupling, becomes stronger. The commonly measured flow properties also change when the coupling between two components is taken into account.

cond-mat.other↗

arXiv2015

Consequences of Disorder on the Stability of Amorphous Solids

Highly acurate numerical simulations are employed to highlight the subtle but important differences in the mechanical stability of perfect crystalline solids versus amorphous solids. We stress the difference between strain values at which the shear modulus vanishes and strain values at which a plastic instability insues. The temperature dependence of the yield strain is computed for the two types of solids, showing different scaling laws: $γ_{_{Y}}\simeq γ_{_{Y}}^{0}-C_1 T^{1/3}$ for crystals versus $γ_{_{Y}}\simeq γ_{_{Y}}^{0}- C_2 T^{2/3}$ for amorphous solids.

cond-mat.dis-nn↗

arXiv2014

Mechanical Properties and Plasticity of a Model Glass Loaded Under Stress Control

Much of the progress achieved in understanding plasticity and failure in amorphous solids had been achieved using experiments and simulations in which the materials were loaded using strain control. There is paucity of results under stress control. Here we present a new method that was carefully geared to allow loading under stress control either at $T=0$ or at any other temperature, using Monte-Carlo techniques. The method is applied to a model perfect crystalline solid, to a crystalline solid contaminated with topological defects, and to a generic glass. The highest yield stress belongs to the crystal, the lowest to the crystal with a few defects, with the glass in between. Although the glass is more disordered than the crystal with a few defects, it yields stress is much higher than that of the latter. We explain this fact by considering the actual microscopic interactions that are typical to glass forming materials, pointing out the reasons for the higher cohesive nature of the glass. The main conclusion of this paper is that the instabilities encountered in stress-control condition are the identical saddle-node bifurcation seen in strain-control. Accordingly one can use the latter condition to infer about the former. Finally we discuss temperature effects and comment on the time needed to see a stress controlled material failure.

cond-mat.stat-mech↗

arXiv2013

On the Effect of Micro-alloying on the Mechanical Properties of Metallic Glasses

"Micro-alloying", referring to the addition of small concentration of a foreign metal to a given metallic glass, was used extensively in recent years to attempt to improve the mechanical properties of the latter. The results are haphazard and nonsystematic. In this paper we provide a microscopic theory of the effect of micro-alloying, exposing the delicate consequences of this procedure and the large parameter space which needs to be controlled. In particular we consider two very similar models which exhibit opposite trends for the change of the shear modulus, and explain the origins of the difference as displayed in the different microscopic structure and properties.

cond-mat.mtrl-sci↗

arXiv2013

Energy Fluxes during Dynamo Reversals

Using direct numerical simulations of the equations of magnetohydrodynamics, we study reversals of the magnetic field generated by the flow of an electrically conducting fluid in a sphere. We show that at low magnetic Prandtl numbers, Pm=0.5, the decrease of magnetic energy, ohmic dissipation and power of the Lorentz force during a reversal is followed by an increase of the power injected by the force driving the flow and an increase of viscous dissipation. Cross correlations show that the Lorentz energy flux is in advance with respect to the other energy fluxes. We also observe that during a reversal, the maximum of the magnetic energy density migrates from one hemisphere to the other and comes back to its initial position, in agreement with recent experimental observations. For larger magnetic Prandtl numbers (Pm= 1, 2), the magnetic field reversals do not display these trends and strongly differ one from another.

physics.flu-dyn↗

arXiv2013

Shear localization in 3-Dimensional Amorphous Solids

In this paper we extend the recent theory of shear-localization in 2-dimensional amorphous solids to 3-D. In 2-D the fundamental instability of shear-localization is related to the appearance of a line of displacement quadrupoles, that makes an angle of 45 degrees with the principal stress axis. In 3-D the fundamental plastic instability is also explained by the formation of a lattice of anisotropic elastic inclusions. In the case of pure external shear stress, we demonstrate that this is a 2-dimensional triangular lattice of similar elementary events. It is shown that this lattice is arranged on a plane, that, similarly to the 2-D case, makes an angle of 45 degrees with respect to the principal stress axis. This solution is energetically favorable only if the external strain exceeds a yield-strain value, which is determined by the strain parameters of the elementary events and the Poisson ratio. The predictions of the theory are compared to numerical simulations and very good agreement is observed.

cond-mat.soft↗

arXiv2013

A Mass-Conserving 4D XCAT Phantom for Dose Calculation and Accumulation

The XCAT phantom is a realistic 4D digital torso phantom that is widely used in imaging and therapy research. However, lung mass is not conserved between respiratory phases of the phantom, making detailed dosimetric simulations and dose accumulation unphysical. A framework is developed to correct this issue by enforcing local mass conservation in the XCAT lung. Dose calculations are performed to assess the implications of neglecting mass conservation, and to demonstrate an application of the phantom to calculate the accumulated delivered dose in an irregularly breathing patient. Monte Carlo methods are used to simulate conventional and SBRT treatment delivery. The spatial distribution of the lung dose was qualitatively changed by the use of mass conservation; however the corresponding DVH did not change significantly. Comparison of the delivered dose with 4DCT-based predictions shows similar lung metric results, however dose differences of 10% can be seen in some spatial regions. Using this tool to simulate patient treatments reveals differences between the planned dose and the calculated delivered dose for the full treatment. The software is freely available from the authors.

physics.med-ph↗

arXiv2013

Micro-alloying and the Toughness of Glasses: Modeling with Pinned Particles

The usefulness of glasses, and particularly of metallic glasses, in technological applications is often limited by their toughness, which is defined as the area under the stress vs. strain curve before plastic yielding. Recently toughness was found to increase significantly by the addition of small concentrations of foreign atoms that act as pinning centers. We model this phenomenon at zero temperature and quasi-static straining with randomly positioned particles that participate in the elastic deformation but are pinned in the non-affine return to mechanical equilibrium. We find a very strong effect on toughness via the increase of both the shear modulus and the yield stress as a function of the density of pinned particles. Understanding the results calls for analyzing separately the elastic, or "Born term" and the contributions of the "excess modes" that result from glassy disorder. Finally we present a scaling theory that collapses the data on one universal curve as a function of rescaled variables.

cond-mat.mtrl-sci↗

arXiv2007

Pair Correlation Functions and a Free-Energy Functional for the Nematic Phase

In this paper we have presented the calculation of pair correlation functions in a nematic phase for a model of spherical particles with the long-range anisotropic interaction from the mean spherical approximation(MSA) and the Percus-Yevick (PY) integral equation theories. The results found from the MSA theory have been compared with those found analytically by Holovko and Sokolovska (J. Mol. Liq. $\bf 82$, 161(1999)). A free energy functional which involves both the symmetry conserving and symmetry broken parts of the direct pair correlation function has been used to study the properties of the nematic phase. We have also examined the possibility of constructing a free energy functional with the direct pair correlation function which includes only the principal order parameter of the ordered phase and found that the resulting functional gives results that are in good agreement with the original functional. The isotropic-nematic transition has been located using the grand thermodynamic potential. The PY theory has been found to give nematic phase with pair correlation function harmonic coefficients having all the desired features. In a nematic phase the harmonic coefficient of the total pair correlation function $h({\bf x_1},{\bf x_2})$ connected with the correlations of the director transverse fluctuations should develop a long-range tail. This feature has been found in both the MSA and PY theories.

cond-mat.soft↗