The Global Research Network for Verified Scholars

Perovskite solar cells (PSCs) are an intriguing field of study due to their unique characteristics. This study aims to utilize the SCAPS-1D simulator to study the behaviour of absorbers in perovskite photocells that incorporate silver cadmium fluoride (AgCdF3). In this study, AgCdF3-based absorber material combined with 4 distinct ETLs namely PC60BM, SnS2, ZnSe and CdZnS and C6TBTAPH2 HTL in the structural proposed configuration of FTO/ETL/AgCdF3/C6TBTAPH2/Au. The materials have been chosen based on the proper band alignments. The influence of thickness, acceptor density, defect density for both HTL and absorber layer in addition to the donor density, defect density, and thickness of the ETLs are observed carefully. Several parameters including the thickness of absorber and ETL and acceptor density of absorber layer is optimized to find best performances. After all the optimization, the FTO/PC60BM/AgCdF3/C6TBTAPH2/Au structure has provided the highest power conversion efficiency (PCE) of 31.32 %, along with 0.86 V of open circuit voltage (VOC), 41.90 mA/cm2 of short circuit current (JSC) and 86.65 % of fill factor (FF). This simulation research allows scientists to develop cost-effective and efficient PSCs, advancing solar technology.

Scaling a stateless service is easy: add pods when CPU rises, remove them when it falls. A MongoDB replica set does not work like that. A secondary that is still catching up on replication is not really usable capacity yet. Count it anyway, and you think you have headroom you do not. Scale down after a failure, and you can remove a replica before it is safe to do so. Our paper proposes CAAS, a consistency-aware autoscaler for stateful microservices. It counts only non-stale secondaries toward capacity, plans scale-up over the full initial-sync horizon, keeps apply-headroom so a lagging member can catch up, and blocks unsafe scale-down until failure recovery is actually complete. Evaluation is simulation-only, using a Python discrete-event simulator with 3 YCSB-inspired workloads, injected failures, 10 seeds, and 390 runs, compared against HPA, a predictive autoscaler, and HPA-MS. On those simulated workloads, SLO violations fell from 11.41% to 0.36% and stale reads from 5.68% to 0.03%. Results were strongest on read-dominated workloads; on write-heavy workloads the trade-off is real. The secure design elements were designed in the paper, not measured in the simulation. I would welcome discussion from researchers running stateful workloads in production: how do you gate scale-down on replication lag today, and what signal do you trust for "this secondary is ready"?

Selective harmonic elimination pulse-width modulation places exact spectral nulls at fundamental-frequency switching, yet its defining transcendental system admits no closed form and is resolved offline, leaving the deployed modulator without an optimality certificate for its angles and without any mechanism for restoring exactness at run time. A Chebyshev change of variables reduces the nine-level, four-angle system nulling the 3rd, 5th, and 7th harmonics to a single monic quartic whose coefficients are closed-form rational functions of the modulation index, certifying a unique angle set per index and a distortion minimum of 9.2115% at 𝑀=0.8047 within a feasibility island of width 3.4×10−3. This certified oracle then supervises a physics-informed distillation objective in which the governing equations enter the loss directly through equation-residual and distortion channels, compressing a teacher into a 389-parameter student (1.52 kiB, 352 multiply-accumulates, 22.6× fewer parameters) at 0.3208% mean distortion deviation. A cumulative-softmax head makes the ordering constraint structural, and the physics channels place the prediction inside the quadratic convergence basin, so three undamped Newton corrections reach a 2.9×10−13 median residual and 0.0010% deviation where a safeguarded cold start stalls near 10−1. Ten-seed ablation, closed-form gradient attribution, and MATLAB/Simulink validation at 9.21% distortion complete a certified, refinable, kilobyte-scale modulation pipeline.

Articular cartilage degeneration resulting from osteoarthritis, trauma, and age-related disorders is a major cause of disability worldwide. Although total joint replacement (TJR) remains the standard treatment for advanced joint degeneration, it is associated with long-term complications such as implant loosening, stress shielding, bone necrosis, wear debris, and the need for revision surgery. These limitations arise primarily because conventional implant materials fail to replicate the unique tribological and mechanical characteristics of natural cartilage. This research explores advanced biomaterials capable of functioning as cartilage substitutes, thereby reducing the extent of surgical intervention while restoring near-natural joint performance. The study investigates the tribological and nanomechanical behaviour of polymer composites and coated refractory metals with established biocompatibility. Experimental evaluation was carried out under simulated physiological conditions using lubricants formulated to mimic natural synovial fluid. Particular emphasis was placed on reproducing the complex lubrication mechanisms—boundary, elasto-hydrodynamic, boosted, and weeping lubrication—that enable healthy synovial joints to achieve an exceptionally low coefficient of friction (~0.01). Friction, wear resistance, and surface integrity were systematically assessed to determine the suitability of the candidate materials for cartilage replacement applications. The investigations identified both favourable and limiting characteristics of the tested materials. However, optimization through microstructural modification and the use of appropriate synthetic lubricants resulted in significant improvements in tribological performance. These findings demonstrate that engineering the surface characteristics and microstructure of biocompatible materials can substantially enhance their functional behaviour under physiological loading conditions.

arge language models (LLMs) do not read words; they read tokens, and the tokenizer decides how many tokens a language needs to say the same thing. This paper introduces the notion of a Sindhi token tax and tests it empirically for Pakistan’s English–Sindhi context. Using 1,004 parallel sentences from the FLORES-200 benchmark in English, Sindhi and Urdu, the study measured five widely deployed tokenizers (GPT-2, Llama 2, Llama 3, Qwen2 and Command-R) with a reproducible Python pipeline. A convergent mixed-methods design combined quantitative indices with a rule-assisted qualitative analysis of how frequent Sindhi words are segmented. Quantitatively, Sindhi required 2.87 to 4.98 times as many tokens as English for the same content in total (Wilcoxon signed-rank, r = .87, p < .001), and a 4,096-token context window held about 770 to 1,335 Sindhi words against 2,890 to 3,330 English words. A new Orthographic Penalty Index showed that the 18 letters unique to the Sindhi alphabet, such as ڪ, ٿ, ڻ and ڳ, were split into byte fragments in 100% of their occurrences in four of the five tokenizers, while letters shared with Arabic and Urdu were almost never split. Qualitatively, five patterns emerged, including the fragmentation of core grammatical words, the “Urdu shortcut” by which Urdu code points are cheaper than Sindhi ones, and the invisibility of Sindhi morphology to subword boundaries. The study also found Unicode inconsistency in digital Sindhi, with 238 of 1,004 benchmark sentences mixing forms of the letter heh. The paper proposes the Script-Aware Tokenization Equity Framework (SATEF) and argues that script-level design, not only data scarcity, makes AI costlier and weaker for Sindhi speakers, with implications for language policy, education and technology in Pakistan.

Significant advancements in machine learning algorithms have the potential to aid in the early detection and prevention of cancer, a devastating disease. However, traditional research methods face obstacles, and the amount of cancer-related information is rapidly expanding. The authors have developed a helpful support system using three distinct deep-learning models, ResNet-50, EfficientNet-B3, and ResNet-101, along with transfer learning, to predict lung cancer, thereby contributing to health and reducing the mortality rate associated with this condition. This offer aims to address the issue effectively. Using a dataset of 1,000 DICOM lung cancer images from the LIDC-IDRI repository, each image is classified into four different categories. Although deep learning is still making progress in its ability to analyse and understand cancer data, this research marks a significant step forward in the fight against cancer, promoting better health outcomes and potentially lowering the mortality rate. The Fusion Model, like all other models, achieved 100% precision in classifying squamous cells. The Fusion Model and ResNet-50 achieved a precision of 90%, closely followed by EfficientNet-B3 and ResNet-101 with slightly lower precision. To prevent overfitting and improve data collection and planning, the authors implemented a data extension strategy. The relationship between acquiring knowledge and reaching specific scores was also connected to advancing and addressing the issue of imprecise accuracy, ultimately contributing to advancements in health and a reduction in the mortality rate associated with lung cancer.

LC continues to be the most prevalent cause of cancer deaths worldwide, which calls for sophisticated detection strategies. The present study investigates 34 demographic, behavioural, and haematological risk factors based on a sample of 2,000 patient data records. A multi-model machine learning approach compares nine algorithms: KNN, AdaBoost (AB), logistic regression (LR), random forest (RF), SVM, naive Bayes (NB), decision tree (DT), gradient boosting (GB), and stochastic gradient descent (SGD). Performance measures (accuracy, sensitivity, specificity, F1-score, AUC) identify quantitative differences: GB had the best F1-scores (0.953) and NB had the second-best F1-score (0.945), while GB had the best sensitivity (99.1%). The KNN-AB hybrid model reported the highest accuracy with 99.5%, while RF reported the highest AUC with a value of 0.92. Ensemble approaches (RF, GB) showed robust predictive performance across measures through integration of complementary strengths of base models. Lasso and ridge regression were able to minimise overfitting, making them easier to interpret. Therapeutic uses include integration into electronic health records (EHRs) for computerised risk stratification, early LC screening, and public health interventions in high-risk subjects (smokers with abnormal haematologic markers). The research highlights the value of hybrid ML models to integrate behavioural and biological data to effectively predict LC. Subsequent work can expand predictive capabilities through imaging data and genomics data incorporation and continue to advance early identification and patient-specific therapy options. This is an intersection of computational advances and clinical translation, providing scalable solutions for global LC diagnosis.

Mobile-mediated communication has reshaped multilingual practices in Pakistan, yet digital interaction among Sindhi university students remains under-researched. This study examines the frequencies, drivers, and attitudes regarding Sindhi–Urdu–English code-switching on WhatsApp among 280 postsecondary students in Sindh. Framed by Myers-Scotton’s Markedness Model, this convergent mixed-methods inquiry evaluated survey responses using SPSS (v26), alongside a 42,150-word chat corpus and 15 semi-structured interviews via NVivo (v14). Quantitative findings reveal Sindhi–English as the most prevalent combination (34.6%), followed by Urdu–English (29.3%) and trilingual integration (27.5%). Gender patterns show female participants switching predominantly for relational and identity functions, whereas male peers favor technical and topical switches. Overall, respondents expressed favorable attitudes, highlighting ease of communication and vocabulary bridging. Qualitative thematic synthesis yielded six core motivations: convenience, identity construction, peer solidarity, status, lexical gaps, and humor. The study contends that WhatsApp functions as a distinct sociolinguistic domain where youth perform a dynamic, technology-mediated identity that both resists and reinforces English-dominated linguistic hierarchies. To better explain digital language selection, the author proposes "digilingual markedness" as an extended adaptation of the Markedness Model. These insights provide significant contributions to sociolinguistic theory, educational policy, and English language pedagogy in Sindh.