Shell (India)

Nowadays, due to the increasing number of cores and sockets in modern multiprocessor servers, it is essential to coschedule multiple applications simultaneously to maximize system utilization. However, the performance and energy efficiency of coexecuting applications are highly sensitive to thread placement, core allocation, and both core and uncore frequency settings. Existing dynamic resource allocation solutions often rely on model-based approaches, require intrusive modifications to the parallel runtime, or lack a unified framework that can utilize multiple energy-saving mechanisms while supporting different parallel programming models. This paper presents Harmonizer, a novel dynamic resource optimization library for co-running applications on multiprocessor systems to improve overall system throughput and energy efficiency. Harmonizer is oblivious to the parallel programming model and requires no information from prior executions. It periodically profiles each application's CPU, cache, and memory usage by utilizing hardware performance monitoring counters and uses this data to determine optimal thread placement, core allocation, and frequency settings. We evaluate Harmonizer using several co-running mixes of exascale proxy applications on a four-socket, 72-core Intel Cooper Lake processor. Our results show that Harmonizer reduces energy consumption by 8.8% to 35% (20.5% geometric mean) and improves throughput by 4.8% (geometric mean) compared to the default Linux scheduler. Compared to two state-of-the-art approaches, it achieves up to 28.6% energy savings and 15.8% higher throughput (geometric mean).

Biochar (BC) is an eco-friendly soil conditioner that mitigates climate change and promotes sustainable agriculture. However, selecting the appropriate commercial biochar and its application for specific soil and crop types requires research. The present work explores the effects of two commercial biochars (BC-I and BC-II) on sorghum plants grown in sandy loam soil under greenhouse conditions. The study characterized the morphological and chemical properties of BCs, including structure, surface area, porosity, elemental composition, and functional groups. Sorghum plants were cultivated in soil amended with biochar at varying application rates of 2.5%, 5%, 10%, and 15%, corresponding to 7, 14, 28, and 42 tons per hectare, respectively. Plant performance was evaluated using chlorophyll content (SPAD), relative water content (RWC), Leaf area (LA) and biomass. Soil carbon content was analyzed using elemental and total organic carbon analysis. Macro- and micronutrients content in soils were also determined. Our findings suggest that compared to BC-II, BC-I exhibited a higher abundance of surface functional groups, greater micropore volume, and a significantly larger pore surface area, indicating its superior physicochemical properties. Except for the 2.5% application rate, all other biochar (BC) rates (namely, 5%, 10%, and 15%) significantly enhanced soil carbon content. Notably, the 5% application rate resulted in the most substantial improvement in soil nutrient levels, including calcium (Ca), potassium (K), magnesium (Mg), and phosphorus (P), compared to the other rates. Sorghum plants treated with lower application rates (2.5% and 5%) of both BC-I and BC-II exhibited significantly enhanced RWC, chlorophyll content, and shoot biomass compared to those treated with higher rates (10% and 15%). Among these, the 5% composite BC-I application demonstrated the most consistent improvement in plant physiological traits (RWC and SPAD) and growth parameters (LA and biomass). However, neither BC variants significantly improved soil nitrogen levels. Our findings indicate that a 5% application rate of composite BC-I provides the most effective balance between enhancing plant performance and improving soil quality in sandy loam soil. Future research on biochar production efforts should prioritize blending this biochar with nitrogen-rich organic fertilizers to address nitrogen limitations and further enhance soil fertility.

Abstract Physics‐Informed Neural Networks (PINNs) accelerate equation solving by merging physics‐based theories with machine learning. Yet, their application in multi‐staged computational workflows unveils reliability issues. This is demonstrated for Fischer‐Tropsch synthesis modeling, by leveraging PINNs for source terms evaluation in the finite‐difference method solving the coupled reaction‐diffusion equations. Subtle inaccuracies of PINNs approximating the functions in close vicinity of their input variables' ranges boundaries, while not captured by traditional neural network assessment methods, are shown to induce unphysical ultimate solutions and convergence failures. A problem‐specific PINN architecture is proposed that has a correct asymptotic behavior and resolves the revealed issues. Combined with a tailored initial guess generation scheme, the proposed modifications are shown to recover the overall stability of the simulations while preserving the speed‐up brought by PINNs as the workflow component. The possible applications of the proposed hybrid solver are discussed in the context of chemical reactor simulations.

Abstract Stress Corrosion Cracking (SCC) poses a significant threat to production systems, arising from the interaction of tensile stresses and corrosive environments. Sulfide Stress Cracking (SSC), particularly associated with hydrogen sulfide (H2S) gas, is highly relevant in oil and gas production. Corrosion‐resistant alloys, such as Duplex Stainless Steel (DSS), help mitigate this issue. However, understanding the impact of environmental conditions and loads on SSC in DSS remains challenging. Existing standards lack insights into specific environmental factors. Modeling SSC using physics‐based approaches is computationally intensive. To address this, a novel machine learning (ML) framework utilizing decision tree‐based models and probabilistic graphical models (Bayesian network, BN) is developed. The dataset for DSS is curated from published literature, and data imbalance is addressed using advanced data curation methods. The framework aims to unravel the intricate factors driving SSC in DSS, providing an accurate predictive tool for the oil and gas industry.

Generative AI and Agent-Based Models (ABMs) are transforming marketing with the ability to develop accurate, adaptive, and consumer-centered strategies. ABMs model interactions among agents such as consumers, firms, and influencers to reveal behavioral patterns and campaign optimization. At the same time, Large Language Models (LLMs) are elevating content creation, customer experience, and campaign management. This survey addresses where these technologies overlap to power contemporary marketing innovations, resolving for privacy, scalability, and ethics. Drawing on empirical case studies and current advances in multimodal and reinforcement learning, the book targets hybrid frameworks as well as future paths in ethical AI, green algorithms, and additional personalization.

Slow strain rate testing has been used to investigate the susceptibility of 316L stainless steel to stress corrosion cracking (SCC) in high chloride and low dissolved oxygen (DO) brines. Tests have been performed for various temperatures and ammonium chloride (NH4Cl) concentrations. The susceptibility to SCC was assessed in terms of ductility loss, detailed fractography, and cross-sectional inspections to identify the damage mechanisms. It is challenging to maintain long-term SCC experiments with very low DO levels. In this work, we showed that 3 h of sparging with high-purity nitrogen at a flow rate of 0.2 L/min was sufficient to reduce the DO in a 0.6 L test solution to approximately 10 ppb. However, over the full test duration of 7 d or 8 d with continuous nitrogen purging of the solution, the mean and maximum DO values were 17.4 ppb and 34.4 ppb in 40 wt% NH4Cl, and 16.5 ppb and 41.8 ppb in 30 wt% NH4Cl solutions at 95°C. For 316L stainless steel at the open circuit in ≥30 wt% (i.e., 6.1 M) ammonium chloride solutions with these very low DO levels, pitting corrosion was not seen at 60°C, became evident at 80°C and was severe at 95°C and above, while SCC was not seen at 60°C or 80°C, was possibly initiating at 80°C without propagating significantly and was severe at 95°C.

Recent advancements in AI-driven weather forecasting models provide faster predictions with lower computational costs compared to traditional Numerical Weather Prediction (NWP) models. However, like NWP models, data-driven models are also prone to persistent systematic biases. This study presents a multi-model ensemble technique designed to enhance forecast accuracy by leveraging the complementary strengths of state-of-the-art AI models combined using the Extreme Gradient Boosting algorithm. Our ensemble approach significantly reduced model biases, particularly improving forecast accuracy at longer lead times. These findings suggest that multi-model AI ensembles offer a promising, computationally efficient alternative to traditional methods for real-time weather forecasting, with potential applications in fields requiring high-precision weather predictions.