11563 research outputs found
Sort by
Filling streamflow data gaps in Indian catchments using machine learning
No full textComplete hydrological time series are critical for effective water resource management, flood and drought forecasting, hydroelectric power optimization, irrigation planning, ecological preservation, and climate change impact assessments. However, significant data gaps in streamflow and water level observations, compounded by extreme hydroclimatic events and quality control issues, hinder accurate modeling and informed decision-making in Indian catchments. The current challenges are particularly pronounced in regions with high climatic variability, where missing data spans 6 to 12 months. To address this, we employed geomorphological, meteorological, and hydrological parameters in combination with the Random Forest method to gap-fill streamflow data at 352 stations across India, except the transboundary basins. To enhance model accuracy and training, we categorized stations into similar-behaving classes using a k-means clustering algorithm based on catchment characteristics. This clustering increased the availability of training data for machine learning models. Streamflow data from each class was trained with 80% of the available data and validated on the remaining 20%. Our results indicate that clustering significantly improves performance, with over 100 stations reporting a >25% increase in Nash-Sutcliffe Efficiency (NSE). Model performance was evaluated for continuous data gaps of 1 week, 1 month, 3 months, 6 months, and 1 year, revealing a decline in accuracy with longer gaps. Despite this, the mean NSE exceeded 0.85 across all clusters. The gap-filled datasets provide robust hydrographs, enabling precise streamflow variability modeling, climate-hydrology interaction evaluation, and improved water resource management strategies
Scalable free energy computation of polymers in explicit solvent using TMMC and pre-generated conformation libraries
No full textWe present a novel application of the Transition Matrix Monte Carlo (TMMC) algorithm to compute the relative free energies of polymers in explicit solvents as a function of a selected order parameter. Our method leverages a pre-generated library of polymer conformations in vacuum, coupled with explicit solvent environments using the Growth Expanded Ensemble (GEE) framework. The integration of TMMC within GEE addresses sampling challenges by introducing bias in Monte Carlo simulations while enabling the computation of unbiased probability distributions and relative free energies. A key advantage of our approach is its flexibility—the polymer conformation library can be generated using any sampling technique, including Molecular Dynamics or Monte Carlo simulations, in implicit or explicit solvents and at different temperatures. The method is adaptable to any collective variable (CV) and can be extended to compute free energies as a function of multiple CVs. Furthermore, its parallelizable structure makes it highly scalable on multi-core central processing units and graphics processing unit architectures. To demonstrate its applicability, we apply it to a fully flexible polymer model consisting of Lennard-Jones particles connected via a harmonic potential, immersed in an explicit solvent of Lennard-Jones particles. The relative free energies are computed as a function of the radius of gyration. Results for three different solvents, obtained by varying the polymer-solvent interaction strength, reveal that the polymer preferentially adopts an extended conformation in good solvents and a collapsed conformation in poor solvents, consistent with theoretical expectations. Our method provides a computationally efficient and scalable framework for free energy calculations, with broad applications in polymer physics and macromolecular thermodynamics
UNDRAINED INSTABILITY RESPONSE OF GRANULAR MATERIAL IN FLEXIBLE BOUNDARY PLANE STRAIN CONDITIONS
No full textInstabilities in granular materials are marked by the development of heterogeneous deformations in "single-element" experiments. However, with the onset of instability, it becomes a boundary value problem. The present study uses a flexible boundary plane-strain (FB-PS) apparatus to characterise the instability behaviour under undrained conditions. Consolidated undrained tests are performed with two different boundary conditions, i.e., flexible and mixed boundaries, to study the influence of boundary rigidity. Flexible boundary signifies employing butyl rubber flexible membranes along all sides. While in mixed boundary conditions, additional rigidity is provided along the principal loading direction (σ1) using thin aluminium sheets with flexible butyl rubber membranes. The initiation of instability is delayed with relatively flexible boundary conditions
Cohen–Macaulay binomial edge ideals in terms of blocks with whiskers
No full textFor a graph G and the binomial edge ideal JG of G, Bolognini et al. have proved the following: JG is strongly unmixed ⇒JG is Cohen–Macaulay ⇒G is accessible. Moreover, they have conjectured that the converse of these implications is true. Accessible and strongly unmixed properties are purely combinatorial. We give some motivations to focus only on blocks with whiskers for the characterization of all G with Cohen–Macaulay JG. We show that accessible and strongly unmixed properties of G depend only on the corresponding properties of its blocks with whiskers and vice versa. We give a new family of graphs whose binomial edge ideals are Cohen–Macaulay, and from that family, we classify all r-regular r-connected graphs, with the property that, after attaching some special whiskers to it, the binomial edge ideals become Cohen–Macaulay. To prove the Cohen–Macaulay conjecture, it is enough to show that every non-complete accessible graph G has a cut vertex v such that G / {v} is accessible. We show that any non-complete accessible graph G having at most three cut vertices has a cut vertex v for which G / {v} is accessible
MedFocusCLIP : Improving few shot classification in medical datasets using pixel wise attention
No full textWith the popularity of foundational models, parameter efficient fine tuning has become the defacto approach to leverage pretrained models to perform downstream tasks. Taking inspiration from recent advances in large language models, Visual Prompt Tuning, and similar techniques, learn an additional prompt to efficiently finetune a pretrained vision foundational model. However, we observe that such prompting is insufficient for fine-grained visual classification tasks such as medical image classification, where there is large inter-class variance, and small intra-class variance. Hence, in this paper we propose to leverage advanced segmentation capabilities of Segment Anything Model 2 [1] (SAM2) as a visual prompting cue to help visual encoder in the CLIP [2] (Contrastive Language-Image Pretraining) by guiding the attention in CLIP visual encoder to relevant regions in the image. This helps the model to focus on highly discriminative regions, without getting distracted from visually similar background features, an essential requirement in a fewshot, finegrained classification setting. We evaluate our method on diverse medical datasets including X-rays, CT scans, and MRI images, and report an accuracy of (71%, 81%, 86%, 58%) from the proposed approach on (COVID, lung-disease, brain-tumor, breast-cancer) datasets against (66%, 70%, 68%, 29%) from a pretrained CLIP model after fewshot training. The proposed approach also allows to obtain interpretable explanation for the classification performance through the localization obtained using segmentation. For demonstrations and visualizations, please visit https://aadya-arora.github.io/MedFocusClip/
Dissecting Parasitic Capacitance in Nanosheet FETs: An Analytical Perspective
No full textThis work presents an approach to extract and analytically model the parasitic capacitance components in Nanosheet FETs. Along with parallel and fringing components, the junction capacitance which is a significant contributor to the total parasitic capacitance is accurately modeled for the first time. The fringing parasitic capacitance components are modeled using the Elliptical Integral Method. The model uses only one fitting parameter and is accurate across the device structural variations with only 1.2% error. � 2024 Elsevier B.V., All rights reserved
Effect of Methyl Cellulose on Cement Hydration and Pore Creation in High-Strength Mixes at Elevated Temperatures
No full textConcrete is known for its inherent fire resistance in addition to several other desirable properties, making it a highly used construction material. Newer scenarios have given rise to enhanced use of high-strength concrete (HSC). However, HSC possesses a dense microstructure because of the lower water-to-binder ratio and finer filler materials. HSC thus has a tendency of explosive spalling when subjected to a rapid increase in temperature. Fibers such as steel and polypropylene (PP) increase the tensile strength and create microchannels in HSC to mitigate explosive spalling, respectively. However, these fibers possess the challenge of lower workability and fiber agglomeration during the mixing of fresh HSC mixes. Water-soluble polymers have been shown to reduce spalling in high-strength mixes at elevated temperatures without compromising on workability. The present work investigates the effect of methyl cellulose (MC) polymer on the hydration of cement and its potential to create additional pores in high-strength mortar mixes at elevated temperatures. The addition of 0.5% MC increased the initial and final setting time of ordinary portland cement by 16% and 10%, respectively. Bound water calculations showed retardation in the initial hydration rate after 1 day of hydration. However, thermal characterization and Fourier transform infrared spectroscopy (FTIR) results demonstrated that the addition of MC polymer did not alter the rate and degree of hydration of cement pastes in the long term. MC polymer showed high pore creation ability when subjected to elevated temperatures. Higher mass loss and water absorption capacity of MC-mixed mortar samples indicate the formation of interconnected pores that can effectively mitigate explosive spalling. The formation of pores was also confirmed through ultrasonic pulse velocity measurement. The present work is expected to form the basis, and future adoption of MC as a cost-effective water-soluble polymer as admixtures in HSC to reduce the susceptibility to spalling at high temperatures
Bicriteria FPT-Approximation Algorithms for Vertex Deletion to Bounded Degeneracy Graphs
No full textIn this work, we consider the optimization problem of finding a minimum-weight subset of vertices of a given undirected graph on n vertices whose deletion results in a d-degenerate graph. For d≥2, this problem is known to be constant-factor inapproximable implying that one cannot hope for anything better than bicriteria approximation algorithms. Towards this end, we give a randomized polynomial-time algorithm that for any value of the bicriteria approximation trade-off parameter α>1 and confidence parameter δ∈(0,1), returns a 2αd-degeneracy modulator whose weight is at most (1+δ)·2αα-1 times the weight of an optimum solution with high probability. Then, we move on to the decision problem of determining if a graph G on n vertices has a d-degeneracy modulator of size at most k. For each d≥2, this problem is known to be W[P]-hard with respect to k and we give three FPT-approximation algorithms for solving it. These algorithms return a 2αd-degeneracy modulator whose size is at most k (if a k-sized d-degeneracy modulator exists) for any α>1. All our algorithms can be tuned to return a 2d-degeneracy modulator of size at most k (if a k-sized d-degeneracy modulator exists) by setting α appropriately