Discover and publish cutting edge, open research.

Coastal ecosystems are exposed to both global and local stressors operating across multiple scales. However, research rarely considers how their combined effects propagate over time and across levels of biological organisation. Here, we employ an in situ warming experiment across two rocky shore sites with contrasting sewage pollution to quantify independent and interactive effects of warming and nutrient pollution from genes to communities. Passively warmed and control settlement plates were deployed at polluted and non-polluted sites and surveyed across a summer to quantify temporal dynamics in the responses of key intertidal taxa. Barnacles were further employed as a model for comparing responses across biological levels, including body size, stable isotope analysis, and RNA sequencing. Pollution consistently increased invertebrate abundance and macroalgal cover, alongside transient positive effects on barnacle size, whereas warming reduced barnacle abundance and suppressed macroalgal cover late in the season. Warming and pollution interacted synergistically on barnacle abundance, with pollution remaining the dominant stressor. Microphytobenthos groups similarly showed distinct pollution-driven increases with warming primarily modifying temporal trajectories; cyanobacteria showed both date-specific and season-wide synergistic interactions, against a backdrop of temporal variability across stressor treatments. Consistent with these patterns, pollution shifted barnacle δ13C and δ15N toward values indicative of greater assimilation of sewage-derived material, while warming increased elemental C:N ratios, consistent with altered nutritional stress. Transcriptomic responses mirrored this dominance of pollution, broadly regulating gene expression linked to protein turnover, DNA repair, and protein folding; combined warming and pollution further intensified proteostasis-related changes and produced predominantly reversal and antagonistic interaction types. Our results show that sewage pollution can overwhelm and reshape warming effects over time and across biological levels, linking group-level responses with parallel shifts in trophic biomarkers and gene regulation. Our scalable field approach provides a template for in situ marine multiple stressor experiments across wider spatiotemporal scales.

Widespread acceptance of collaborative robots in human-involved scenarios requires accessible and intuitive interfaces for lay workers and non-expert users. Existing interfaces often rely on users to plan and issue low-level commands, necessitating extensive knowledge of robot control. This study proposes a multimodal Agentic AI framework integrating natural user interfaces (NUIs) to foster effortless human-like partnerships in Human-Robot Collaboration (HRC). It enhances intuitiveness and operational efficiency. First, it allows users to instruct robots using plain language verbally, coupled with gaze revealing objects precisely. Second, it offloads users’ workload for robot motion planning by understanding context and reasoning task decomposition. Third, coordinating with AI agents built on large-language models (LLMs), the system interprets users’ requests effectively and provides feedback to establish transparent communication. Experiments were conducted to demonstrate a practical implementation of the Agentic AI framework on a mobile manipulation robot in the collaborative task of human-robot wood assembly. In the implementation, seven participants were recruited to interact with this AI-integrated agentic robotic system. Task performance and user experience metrics were measured in terms of completion time, intervention rate, NASA TLX survey for workload, and valuable insights of practical applications were summarized through a qualitative analysis. This study highlights the potential of NUIs and Agentic AI-embodied robots to overcome existing HRC barriers and contributes to improving HRC intuitiveness and efficiency.

Recent advances in deep learning and scientific machine learning (SML) offer a wide spectrum of novel tools with the potential to radically transform various areas of academic research, industrial R&D, and manufacturing. These include inverse device design, data-driven equation discovery, rapid approximate partial differential equations solutions for exploring the design space, automated metrology and manufacturing process control, new materials design, and the creation of fast and realistic digital twins for virtual testing and design optimization. The purpose of this review is to assist R&D practitioners, who are not specialists in AI, in navigating this complex and dynamic landscape, enabling them to adopt modern machine learning (ML) methods in their work. We particularly emphasize the potential advantages of deep learning methods for the field of thin film device developing, highlighting the main approaches and points of their applications in R&D design and process. The review is organized into several sections. First we provide a brief overview of machine learning and deep learning, introducing basic neural network architectures, and describing their possible use cases relevant to industrial R&D. In the following section we introduce examples of ML approaches enabling the reduction of dependency on the amount of input data and improving generalization capabilities of neural networks through introduction of realistic inductive biases in the form of symmetries, conservation laws, physics equations, etc. Then we review some of most successful large scale SML models, including foundational materials simulation and generation models. Finally, we discuss existing and prospective applications of ML models in different aspects of thin film devices development.

We present atmospheric chlorine inventories over twenty-one years (2004–2024) and five latitude bands (82–60°N, 60–30°N, 30°N–30°S, 30–60°S, 60–82°S) across altitudes from the surface up to 61 km. These inventories were calculated using the Atmospheric Chemistry Experiment-Fourier Transform Spectrometer (ACE-FTS) version 5.3 retrievals of the volume mixing ratios (VMRs) of 13 chlorine-containing species. Of these, five are product gases: HCl, HOCl, ClONO2, COClF, COCl2, and eight are source gases: CCl4, CH3Cl, CFC-11 (CCl3F), CFC-12 (CCl2F2), CFC-113 (CClF2CCl2F), HCFC-22 (CHF2Cl), HCFC-141b (C2H3Cl2F), and HCFC-142b (C2H3ClF2). Where necessary, ACE-FTS data were supplemented with data from the TOMCAT 3-D chemical transport model, the Aura Microwave Limb Sounder (MLS) instrument, and ground-based measurements from the National Oceanic and Atmospheric Administration (NOAA) and the Advanced Global Atmospheric Gases Experiment (AGAGE). Total chlorine (Cltot) profiles are dominated by source gas contributions in the troposphere and lower stratosphere. At lower altitudes, the primary contributions are chlorofluorocarbons (CFCs), chlorocarbons, and, increasingly in recent years, chlorine-containing very short-lived substances (Cl-VSLS). At higher altitudes, HCl becomes the dominant contributor, comprising up to 99% of Cltot by 61 km. The relative contribution of individual species to Cltot showed that Cl-VSLS and HCFC-22 have partially slowed the reduction in atmospheric chlorine achieved by the phase-out of CFC production and consumption. Between 2004 and 2024, the global mean Cltot time series decreased by 9.56 ± 0.28 ppt/year (0.28 ± 0.01 %/year), with values approaching 3.3 ppb in 2024. These findings demonstrate the significant impact of the Montreal Protocol in reducing emissions of substances that are both ozone-depleting and green-house gases.

Regulated debit routing sits at the intersection of EMV standards, Durbin routing rights, and acquirer engineering practice, yet in many gateways AID selection is buried in opaque configuration and ad hoc code. This makes it difficult to reason about merchant cost, acceptance rates, or even basic compliance with scheme and regulatory rules. This paper presents a cost aware and compliance safe AID debit routing engine for retail payment gateways. We first define a standards accurate policy model that encodes EMV debit routing constraints, Durbin style requirements for regulated domestic cards, and merchant routing preferences as machine readable rules. On top of this model, we design a constrained optimization algorithm that selects a primary network and fallback order to minimize expected routing cost subject to acceptance and tail latency constraints while respecting all policy guardrails. To support reproducibility, we release a synthetic AID routing dataset and a trace driven simulator with fault injection that enable controlled experiments over issuer mix, fee schedules, outages, and misconfigurations. Using this framework, we show that the proposed engine can reduce expected routing fees relative to a brand first baseline, keep p99 routing latency within the same budget, and eliminate simulated Durbin style violations even under adverse network and configuration scenarios.

Address verification and spend limit checks in card and instant payment systems expose cardholder ZIP codes and velocity histories to gateways and intermediaries, which increases the privacy impact of breaches. At the same time, issuers rely on these checks to control fraud within strict two to three hundred millisecond authorization budgets. This paper presents ZK-AVS, a design that replaces plaintext AVS and spend limit evaluation with zero knowledge proofs produced on the customer device and verified at the payment gateway. The issuer stores commitments to the cardholder ZIP and per window spend counters, and the device proves that its ZIP matches the committed value and that the proposed transaction keeps cumulative spend within the configured limit, without revealing the underlying values. We instantiate a synthetic workload of fifty thousand transactions and show that AVS mismatch and limit pressure carry useful fraud signal that can be exposed as proof outcomes. The design is structured to fit within sub three hundred millisecond budgets on commodity devices, while removing direct exposure of address and spend history data at the gateway and keeping changes largely at endpoints.

Real time payment gateways handle authorization and clearing traffic for card and account based ecosystems under strict latency and availability targets. Most deployed gateways still rely on RSA and elliptic curve cryptography, which become vulnerable once large scale quantum computers can run Shor's algorithm. Simply swapping these algorithms for NIST standard post quantum schemes is not enough, because gateways must still meet service level agreements, maintain Payment Card Industry Data Security Standard (PCI DSS) compliance, and interoperate with acquirers and issuers that migrate at different speeds. This paper presents a Post Quantum Cryptography Migration Framework that gives an end to end architectural view of gateway, acquirer, and issuer communication paths. The framework introduces a crypto abstraction layer with pluggable algorithm providers, supports hybrid classical plus post quantum modes during transition, and uses policy driven selection of key encapsulation and signature schemes such as Kyber and Dilithium. It also defines a phased migration process from asset discovery and quantum risk assessment through controlled hybrid rollout to post quantum only operation. To validate the approach, we implement a gateway testbed that replays a PaySim transaction workload and measure the impact of post quantum schemes on authorization latency and derived throughput under classical only, hybrid, and PQC only modes. The experiments show that, with careful configuration and connection management, post quantum protection can be integrated into real time payment processing with performance overhead kept within operational limits.

Condition monitoring and predictive maintenance of large hydro and wind generators are essential for reducing operational costs and improving system availability. However, traditional monitoring approaches rely primarily on stator-side measurements and bearing vibrations, providing limited observability of rotor-side phenomena where many critical faults originate. This paper presents a comprehensive modular contactless telemetry system designed for direct rotor-side measurement of multiple physical quantities including magnetic flux density, temperatures, damper winding currents, and mechanical vibrations. The system architecture prioritizes flexibility and scalability, enabling easy integration of different sensor types with configurable sampling rates and resolutions. In a laboratory research environment, fixed sensor configurations and rigid sampling parameters often become limiting factors. The proposed modular design addresses these constraints by allowing users to dynamically add or remove acquisition modules, adapt sampling parameters for specific experiments, and perform on-rotor preprocessing to optimize data transmission bandwidth. Detailed descriptions of the hardware design, synchronization strategy, communication protocol, and software architecture are provided. Experimental validation is demonstrated on a 366 kVA research hydro generator, demonstrating the system's capability for reliable, synchronized multi-channel acquisition on a rotating platform with contactless data transmission.

Instant payment rails and mobile money platforms such as UPI, Pix, and FedNow require fraud decisions in a few hundred milliseconds, but most deployed and published fraud models are batch oriented and sequence only. They are trained and evaluated offline, treat each account in isolation, and provide limited visibility into mule networks and collusive cash out structures. This paper presents a graph based streaming risk scoring architecture that brings graph neural network (GNN) style inference into the authorization path while respecting a sub 300 ms end to end service level objective. The system maintains a dynamic transaction graph over accounts, devices, merchants, and institutions, updates this graph as payments arrive, and uses an online GNN to propagate risk so that emerging mule rings can be flagged even when individual transfers appear benign. Raw scores are calibrated into three operational actions, allow, step up, and block, under explicit fraud loss and customer friction budgets. Using a labeled synthetic mobile money dataset representative of instant payment rails, we build a real time risk engine around this architecture and quantify its latency, queueing behaviour, and decision quality. A strong tabular baseline representative of existing non graph deployments is implemented inside the same engine and evaluated side by side, providing a concrete comparative analysis between current practice and the proposed architecture.

We study the peeker's advantage in first-person shooters, a classic problem in multiplayer games. A player, the peeker, who first observes another player, the defender, has a strategic advantage. The peeker has the opportunity to aim and shoot first, as the defender is not aware of the peeker's presence for a short amount of time. Formally, the peeker's advantage is the length of time that the peeker sees the defender without being seen. This is an inherent problem of multiplayer games, where each player runs their own instance of the game and relies on updates from an authoritative game server to get informed about the global game state. In this article, we explore the peeker's advantage with extensive simulations, by varying the system delays, the network delays, and the reaction times of the players. First, we present our multiplayer abstraction that captures the essence of first-person shooters, including client and server logic as well as the exchanged messages. Next, we show that the game frame rate is the most influential factor, while the game server tick rate and network delays have minimal impact. Specifically, we show that by using 120 fps in the game and 60 ticks/s on the game server, the peeker's advantage can be dropped below 100 ms. Finally, using realistic, standards-compliant simulations, we show that this advantage is consistent across multiple configurations, both in local area networks and over the Internet.

A document by Lalitha Ramachandrappa. Click on the document to view its contents.

In this paper, we investigate Statistical Bit Error Rate (BER) analysis for low-loss short-reach chiplet interface and high-loss long-reach serial interface. We used jitter filtering to account for the residue jitter not tracked by a forwarded clock system and proposed a fast and exact Statistical BER method to account for the Tx jitter amplification effect in a high-loss channel. Our proposed method achieves a linear computation complexity.