arXiv subjects

Shreya Das

Publications and source records attributed to Shreya Das.

9 recordsLinked to original sources

arXiv2026

A Bayesian Filtering Approach for Learning Lagrangian Dynamics from Noisy Measurements

This paper proposes a Bayesian filtering-based approach for learning the dynamics of a physical system from partial, noisy measurements. We model the system dynamics using a Lagrangian mechanics formulation. As in Lagrangian neural networks (LNNs), we parameterize the kinetic and potential energies with neural networks. The unknown external forces in the Lagrangian formulation are modeled as white Gaussian noise. The corresponding Euler--Lagrange equations then yield a continuous-time stochastic state-space model (SSM) that describes the system dynamics. The neural network parameters and system states are then jointly learned via a maximum-likelihood method using Gaussian-approximation-based Bayesian filters. The effectiveness of the proposed method is demonstrated on pendulum and Duffing oscillator examples, and its performance is compared with conventional LNNs and with approximate Bayesian filters using known system models.

cs.LG↗

arXiv2026

Integrating Lagrangian Neural Networks into the Dyna Framework for Reinforcement Learning

Model-based reinforcement learning (MBRL) is sample-efficient but depends on the accuracy of the learned dynamics, which are often modeled using black-box methods that do not adhere to physical laws. Those methods tend to produce inaccurate predictions when presented with data that differ from the original training set. In this work, we employ Lagrangian neural networks (LNNs), which enforce an underlying Lagrangian structure to train the model within a Dyna-based MBRL framework. Furthermore, we train the LNN using stochastic gradient-based and state-estimation-based optimizers to learn the network's weights. The state-estimation-based method converges faster than the stochastic gradient-based method during neural network training. Simulation results are provided to illustrate the effectiveness of the proposed LNN-based Dyna framework for MBRL.

eess.SY↗

arXiv2026

Coexistence of Rashba and Ising Spin-Singlet Pairings in Two-Dimensional IrTe$_{2}$

Symmetry offers a useful approach to unfold the intertwined degrees of freedom. Thus it paves the way to resolve coexisting quantum orders into distinct symmetry sectors. Motivated by the recent observation of superconductivity in nano-flaked IrTe$_2$, we investigate the superconductivity in strain-stabilized two-dimensional (2D) limit of IrTe$_2$ by combining density-functional theory with mean-field solution of spin-fluctuation mediated pairing interaction on a symmetry-constrained ${\bf k}\cdot{\bf p}$ model. The spin-orbit coupled band structure shows $\Gamma$-centred Fermi sheets with coexistence of band-selective Rashba-like (in-plane) and Ising-like (out-of-plane) superconductivity. Remarkably, the superconducting gaps are odd in spin, orbital, and momentum channels despite the presence of global inversion symmetry. Fermi surface topologies and little-group symmetry enforce distinct irreducible representations to the Rashba and Ising channels, forbidding their mixing. Our findings open up a symmetry-based route to multichannel superconductivity in 2D transition-metal dichalcogenides with unique functionalities.

cond-mat.supr-con↗

arXiv2023

Tracking an Underwater Target with Unknown Measurement Noise Statistics Using Variational Bayesian Filters

This paper considers a bearings-only tracking problem using noisy measurements of unknown noise statistics from a passive sensor. It is assumed that the process and measurement noise follows the Gaussian distribution where the measurement noise has an unknown non-zero mean and unknown covariance. Here an adaptive nonlinear filtering technique is proposed where the joint distribution of the measurement noise mean and its covariance are considered to be following normal inverse Wishart distribution (NIW). Using the variational Bayesian (VB) method the estimation technique is derived with optimized tuning parameters i.e, the confidence parameter and the initial degree of freedom of the measurement noise mean and the covariance, respectively. The proposed filtering technique is compared with the adaptive filtering techniques based on maximum likelihood and maximum aposteriori in terms of root mean square error in position and velocity, bias norm, average normalized estimation error squared, percentage of track loss, and relative execution time. Both adaptive filtering techniques are implemented using the traditional Gaussian approximate filters and are applied to a bearings-only tracking problem illustrated with moderately nonlinear and highly nonlinear scenarios to track a target following a nearly straight line path. Two cases are considered for each scenario, one when the measurement noise covariance is static and another when the measurement noise covariance is varying linearly with the distance between the target and the ownship. In this work, the proposed adaptive filters using the VB approach are found to be superior to their corresponding adaptive filters based on the maximum aposteriori and the maximum likelihood at the expense of higher computation cost.

eess.SP↗

arXiv2023

Occupant's Behavior and Emotion Based Indoor Environment's Illumination Regulation

This paper presents an efficient approach for building occupancy modeling to reduce energy consumption. In this work, a novel approach to occupancy modeling based on the posture and comfort level of the occupant is developed, and subsequently, we report a new and efficient framework for detecting posture and emotion from skeleton joints and face points data respectively obtained from the Kinect sensor. The proposed approach is tested in terms of accuracy, region of convergence, and confusion matrix using several machine learning techniques. Out of all the techniques, random forest classifier gave the maximum blind test accuracy for multi-class classification of posture detection. Deep learning is used for emotion detection using several optimizers out of which Adadelta gave the maximum blind test accuracy for multi-class classification. Along with the Kinect sensor, several other sensors such as the magnetic door sensor, pyroelectric sensors, and illumination sensors are connected through a wireless network using Raspberry Pi Zero W. Thus creating an unmanned technique for illumination regulation.

eess.SY↗

arXiv2023

Resolving Left-Right Ambiguity During Bearing Only Tracking of an Underwater Target Using Towed Array

In bearing only tracking using a towed array, the array can sense the bearing angle of the target but is unable to differentiate whether the target is on the left or the right side of the array. Thus, the traditional tracking algorithm generates tracks in both the sides of the array which create difficulties when interception is required. In this paper, we propose a method based on likelihood of measurement which along with the estimators can resolve left-right ambiguity and track the target. A case study has been presented where the target moves (a) in a straight line with a near constant velocity, (b) maneuvers with a turn, and observer takes a `U'-like maneuver. The method along with the various estimators has been applied which successfully resolves the ambiguity and tracks the target. Further, the tracking results are compared in terms of the root mean square error in position and velocity, bias norm, \% of track loss and the relative execution time.

eess.SP↗

arXiv2021

XY magnetism, Kitaev exchange, and long-range frustration in the $J_{\rm eff}=1/2$ honeycomb cobaltates

The quest for Kitaev quantum spin liquids has led to great interest in honeycomb quantum magnets with strong spin-orbit coupling. It has been recently proposed that even Mott insulators with $3d$ transition metal ions, having nominally weak spin-orbit coupling, can realize such exotic physics. Motivated by this, we study the rhombohedral honeycomb cobaltates CoTiO$_3$, BaCo$_2$(PO$_4$)$_2$, and BaCo$_2$(AsO$_4$)$_2$, using $\textit{ab initio}$ density functional theory, which takes into account realistic crystal field distortions and chemical information, in conjunction with exact diagonalization numerics. We show that these Co$^{2+}$ magnets host $j_\text{eff}=1/2$ local moments with highly anisotropic $g$-factors, and we extract their full spin Hamiltonians including longer-range and anisotropic exchange couplings. For CoTiO$_3$, we find a nearest-neighbor easy-plane ferromagnetic $XXZ$ model with additional bond-dependent anisotropies and interlayer exchange, which supports three-dimensional (3D) Dirac nodal line magnons. In contrast, for BaCo$_2$(PO$_4$)$_2$ and BaCo$_2$(AsO$_4$)$_2$, we find a strongly suppressed interlayer coupling, and significant frustration from additional third-neighbor antiferromagnetic exchange mediated by P/As. Such bond-anisotropic $J_1$-$J_3$ spin models can support collinear zig-zag or coplanar spiral ground states; we discuss their dynamical spin correlations which reveal a gapped Goldstone mode, and argue that the effective parameters of these pseudospin-$1/2$ models may be strongly renormalized by coupling to a low energy spin-exciton. Our results call for re-examining proposals for realizing Kitaev spin liquids in the honeycomb cobaltates.

cond-mat.str-el↗

arXiv2021

Understanding Magnetism in Double Double Perovskites: A Complex Multiple Magnetic Sublattice System

Understanding magnetism in multiple magnetic sublattice system, driven by the interplay of varied nature of magnetic exchanges, is on one hand challenging and on other hand intriguing. Motivated by the recent synthesis of AA'BB'O_6 double double perovskites with multiple magnetic ions both at A- and B-sites, we investigate the mechanism of magnetic behavior in these interesting class of compounds. We find that the magnetism in such multiple sublattice compounds is governed by the interplay and delicate balance between two distinct mechanisms, a) kinetic energy-driven multiple sublattice double exchange mechanism and b) the conventional super-exchange mechanism. The derived spin Hamiltonian based on first-principles calculations is solved by the classical Monte Carlo technique which reproduces the observed magnetic properties. Finally, the influence of off-stoichiometry, as in experimental samples, is discussed. Some of these double double perovskite compounds are found to possess large total magnetic moment and also are found to be half-metallic, which raises the hope of future applications of these large magnetic moment half-metallic oxides in spintronics and memory devices.

cond-mat.mtrl-sci↗

arXiv2020

Understanding the Curious Magnetic State of Sr$_{3}$OsO$_{6}$

Motivated by the recent report on high T$_c$ ferromagnetic insulating state of single transition metal containing double perovskite compound, Sr$_3$OsO$_6$ (Wakabayashi et. al., Nature Commun {\bf 10} 535, 2019), we study this curious behavior by employing first-principles calculations in conjunction with exact diagonalization of full $t_{2g}$ multiplet problem of two Os sites. Our analysis highlights the fact that stabilization of Sr$_3$OsO$_6$ in the cubic phase in epitaxially grown thin film is the key to both ferromagnetic correlation and high temperature scale associated to it. This also provides a natural explanation for the sister compound, Ca$_3$OsO$_6$ to exhibit low T$_N$ antiferromagnetism in its monoclinic structure. Further the insulating property is found to be driven by opening of Mott gap in the half filled spin-orbit coupled $j=3/2$ manifold of $d^{2}$ Os. We point out that Sr$_2$CaOsO$_6$ which naturally forms in the cubic phase would be worthwhile to explore as a future candidate to exhibit high T$_c$ ferromagnetic insulating state in bulk form.

cond-mat.str-el↗