Kaikki aineistot
Lisää
The transport properties of the strongly coupled quark-gluon plasma created in ultrarelativistic heavy-ion collisions are extracted by Bayesian parameter estimate methods with the latest collision beam energy data from the CERN Large Hadron Collider. This Bayesian analysis includes sophisticated flow harmonic observables for the first time. We found that the temperature dependence of specific shear viscosity appears weaker than in the previous studies. The results prefer a lower value of specific bulk viscosity and a higher switching temperature to reproduce additional observables. However, the improved statistical uncertainties both on the experimental data and hydrodynamic calculations with additional observables do not help to reduce the final credibility ranges much, indicating a need for improving the dynamical collision model before the hydrodynamic takes place. In addition, the sensitivities of experimental observables to the parameters in hydrodynamic model calculations are quantified. It is found that the analysis benefits most from the symmetric cumulants and nonlinear flow modes, which mostly reflect nonlinear hydrodynamic responses, in constraining the temperature dependence of the specific shear and bulk viscosities in addition to the previously used flow coefficients.
Abstract The wageneri species group of Gyrodactylus contains the following molecularly confirmed salmonid parasites in Asia: Gyrodactylus taimeni Ergens, 1971, Gyrodactylus magnus Konovalov, 1967, Gyrodactylus brachymystacis Ergens, 1978, and Gyrodactylus derjavini Mikhailov, 1975; in Europe it contains the following: Gyrodactylus derjavinoides Malmberg, Collins, Cunningham, and Jalali, 2007, Gyrodactylus truttae Gläser, 1974, Gyrodactylus teuchis Lautraite, Blanc, Thiery, Daniel, and Vigneulle, 1999, Gyrodactylus lavareti Malmberg, 1956, Gyrodactylus salvelini Kuusela, Ziętara, and Lumme, 2008 (presented herein as a junior synonym of Gyrodactylus salmonis), and Gyrodactylus salaris Malmberg, 1957, with the lone confirmed North American exception being G. salmonis. The mitochondrial DNA (cox1, 1545 bp) of this group shows a star-like phylogenetic expansion that began 2.05 ± 0.4 million years ago (mya), estimated from the mean distance of the cox1 gene (dMCL = 0.267) using a tentative, potentially high-end, divergence rate of 0.13/Myr. European G. salaris on Thymallus thymallus and Asian G. magnus on Thymallus arcticus have been separated for 1.95 Myr (dMCL = 0.253). The nuclear ITS rDNA region (1,245 bp) of G. salmonis was nearly uniform among North American populations of Oncorhynchus mykiss, Oncorhynchus clarkii, Oncorhynchus nerka, Salvelinus fontinalis, and Salmo salar (and non-native Salmo trutta) as well as on Salvelinus alpinus (under the synonym G. salvelini) from Lake Inari, Finland. Gyrodactylus salmonis is distal in a monophyletic subclade labeled by an apomorphic 56 bp insertion in the ITS1, shared by the European parasites G. lavareti (host: Coregonus lavaretus), Gyrodactylus pomeraniae Kuusela, Ziętara, and Lumme, 2008 (host: Rutilus rutilus), and Gyrodactylus bliccensis Gläser, 1974 (host: Alburnus alburnus). This subphylogeny suggests that a particular host switch from cyprinids to salmonids may have occurred less than 1.8 mya in the Old World [dMCL = 0.234 G. pomeraniae vs (G. salmonis, G. lavareti)] and possibly again among coregonine hosts and Salvelinus 1.2 mya (dMCL = 0.156). Although hypothetical, a transition from coregonines to charr (notably the widely distributed and adaptable Salvelinus alpinus) potentially could have occurred in a proglacial refugium leading to circumpolar distribution of G. salmonis and a secondary transition to other North American hosts. The maximum cox1 genetic distance within G. salmonis on all hosts was dMCL = 0.032, at the same level as in multihosted European G. salaris (dMCL = 0.032), suggesting circa 250,000 yr of population expansion with these parasites since a temporal, coinciding bottleneck.
Abstract Participatory design (PD) research has historically strongly focused on the reporting of design events (e.g. workshops and prototyping activities with participants), where issues such as ‘involving users’, including the users’ point of view, and participation as a matter of mutual learning have been in the foreground. The need to further problematise and critically examine participation is nonetheless apparent. This special issue aims to shed light on participation as it unfolds over time during, between and beyond participatory events such as these. Here, we build an overview of existing directions taken by researchers to address the unfolding of participation in IT design over time. We do this by examining existing PD literature and the four contributions to this special issue. We identify two common temporalities in PD, the future-oriented and the project-based, and propose five lenses that may aid researchers in exploring and understanding the temporal dimensions of participation in their projects: the phasic, emergent, retrospective, prospective and longterm lenses. We end with propositions and opportunities for future research directions in PD, highlighting the multi-faceted nature of the temporality of participation.
The transport properties of quark-gluon plasma created in relativistic heavy-ion collisions are quantified by an improved global Bayesian analysis using the CERN Large Hadron Collider Pb–Pb data at sNN=2.76 and 5.02 TeV. The results show that the uncertainty of the extracted transport coefficients is significantly reduced by including new sophisticated collective flow observables from two collision energies for the first time. This work reveals the stronger temperature dependence of specific shear viscosity, a lower value of specific bulk viscosity, and a higher hadronization switching temperature than in the previous studies. The sensitivity analysis confirms that the precision measurements of higher-order harmonic flow and their correlations are crucial in extracting accurate values of the transport properties.
Abstract To cope with the lack of on-device machine learning samples, this article presents a distributed data augmentation algorithm, coined federated data augmentation (FAug). In FAug, devices share a tiny fraction of their local data, i.e., seed samples, and collectively train a synthetic sample generator that can augment the local datasets of devices. To further improve FAug, we introduce a multihop-based seed sample collection method and an oversampling technique that mixes up collected seed samples. Both approaches enjoy the benefit from the crowd of devices, by hiding data privacy from preceding hops and feeding diverse seed samples. In the image classification tasks, simulations demonstrate that the proposed FAug frameworks yield stronger privacy guarantees, lower communication latency, and higher on-device ML accuracy.
Abstract On-device machine learning (ML) has brought about the accessibility to a tremendous amount of data from the users while keeping their local data private instead of storing it in a central entity. However, for privacy guarantee, it is inevitable at each device to compensate for the quality of data or learning performance, especially when it has a non-IID training dataset. In this paper, we propose a data augmentation framework using a generative model: multi-hop federated augmentation with sample compression (MultFAug). A multi-hop protocol speeds up the end-to-end over-the-air transmission of seed samples by enhancing the transport capacity. The relaying devices guarantee stronger privacy preservation as well since the origin of each seed sample is hidden in those participants. For further privatization on the individual sample level, the devices compress their data samples. The devices sparsify their data samples prior to transmissions to reduce the sample size, which impacts the communication payload. This preprocessing also strengthens the privacy of each sample, which corresponds to the input perturbation for preserving sample privacy. The numerical evaluations show that the proposed framework significantly improves privacy guarantee, transmission delay, and local training performance with adjustment to the number of hops and compression rate.
Abstract This letter proposes a novel communication-efficient and privacy-preserving distributed machine learning framework, coined Mix2FLD. To address uplink-downlink capacity asymmetry, local model outputs are uploaded to a server in the uplink as in federated distillation (FD), whereas global model parameters are downloaded in the downlink as in federated learning (FL). This requires a model output-to-parameter conversion at the server, after collecting additional data samples from devices. To preserve privacy while not compromising accuracy, linearly mixed-up local samples are uploaded, and inversely mixed up across different devices at the server. Numerical evaluations show that Mix2FLD achieves up to 16.7% higher test accuracy while reducing convergence time by up to 18.8% under asymmetric uplink-downlink channels compared to FL.
Abstract The first generation of IoT was developed and deployed all over the world by connecting devices with common functionalities that were not sufficiently efficient or reliable for use in dynamic situations that require adaptive solutions. However, these fundamental IoT functions and services mainly targeted stable environments; there is consequently a strong need for the next generation of IoT services to be smarter, faster, and more reliable. We believe that the hyper-connected IoT ecosystem on fog platforms with contextual AI technologies is a promising solution. In this work, we introduce the EiF, a flexible fog computing framework that runs on IoT gateways with adaptive AI services fostered on the cloud. Our approach can be viewed as an integration of three emerging technologies, namely IoT, fog, and AI. Generally, EiF virtualizes an IoT service layer platform for fog nodes, and provides functions to manage and orchestrate various fog nodes; upon service virtualization and orchestration, AI services are fostered within both the federated cloud and distributed edge side and are deployed on fog nodes. We demonstrate the feasibility of EiF via the example of intelligent traffic flow monitoring and management.
The Cyclotron Institute at Texas A&M is currently configuring a scheme for the production of radioactive-ion beams that incorporates a light-ion guide and a heavy-ion guide coupled with an electron-cyclotron-resonance ion source constructed for charge-breeding. This scheme is part of an upgrade to the facility and is intended to produce radioactive beams suitable for injection into the K500 superconducting cyclotron. The current status of the project and details on the ion sources used in the project is presented.
The Tonle Sap Lake in Cambodia is a dynamic flood-pulsed ecosystem that annually increases its surface area from roughly 2,500 km2 to over 12,500 km2 driven by seasonal flooding from the Mekong River. This flooding is thought to structure many of the critical ecological processes, including aquatic primary and secondary productivity. The lake also has a large fishery that supports the livelihoods of nearly 2 million people. We used a state-space oxygen mass balance model and continuous dissolved oxygen measurements from four locations to provide the first estimates of gross primary productivity (GPP) and ecosystem respiration (ER) for the Tonle Sap. GPP averaged 4.1±2.3 g O2 m−3 d−1 with minimal differences among sites. There was a negative correlation between monthly GPP and lake level (r = 0.45) and positive correlation with turbidity (r = 0.65). ER averaged 24.9±20.0 g O2 m−3 d−1 but had greater than six-fold variation among sites and minimal seasonal change. Repeated hypoxia was observed at most sampling sites along with persistent net heterotrophy (GPP<ER), indicating significant bacterial metabolism of organic matter that is likely incorporated into the larger food web. Using our measurements of GPP, we calibrated a hydrodynamic-productivity model and predicted aquatic net primary production (aNPP) of 2.0±0.2 g C m−2 d−1 (2.4±0.2 million tonnes C y−1). Considering a range of plausible values for the total fisheries catch, we estimate that fisheries harvest is an equivalent of 7–69% of total aNPP, which is substantially larger than global average for marine and freshwater systems. This is likely due to relatively efficient carbon transfer through the food web and support of fish production from terrestrial NPP. These analyses are an important first-step in quantifying the resource pathways that support this important ecosystem.
Among the environmental factors associated with type 1 diabetes (T1D), viral infections of the gut and pancreas has been investigated most intensely, identifying enterovirus infections as the prime candidate trigger of islet autoimmunity (IA) and T1D development. However, the association between respiratory tract infections (RTI) and IA/T1D is comparatively less known. While there are significant amounts of epidemiological evidence supporting the role of respiratory infections in T1D, there remains a paucity of data characterising infectious agents at the molecular level. This gap in the literature precludes the identification of the specific infectious agents driving the association between RTI and T1D. Furthermore, the effect of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections on the development of IA/T1D remains undeciphered. Here, we provide a comprehensive overview of the evidence to date, implicating RTIs (viral and non-viral) as potential risk factors for IA/T1D.
Abstract We are increasingly in situations of divided attention, subject to interruptions, and having to deal with an abundance of information. Our cognitive load changes in these situations of divided attention, task interruption or multitasking; this is particularly true for older adults. To help mediate our finite attention resources in performing cognitive tasks, we have to be able to measure the real-time changes in the cognitive load of individuals. This paper investigates how to assess real-time cognitive load based on psycho-physiological measurements. We use two different cognitive tasks that test perceptual speed and visio-spatial cognitive processing capabilities, and build accurate models that differentiate an individual’s cognitive load (low and high) for both young and older adults. Our models perform well in assessing load every second with two different time windows: 10 seconds and 60 seconds, although less accurately for older participants. Our results show that it is possible to build a realtime assessment method for cognitive load. Based on these results, we discuss how to integrate such models into deployable systems that mediate attention effectively.