澳大利亚国立大学

A comparison of Precambrian striate, discoid acritarchs with Circulisporites parvus de Jersey originally described from the Triassic of Queensland, significantly expands the stratigraphic distribution of a morphotype previously shown to be aligned with cysts of species belonging to the Euglenophyceae. These organic walled microfossils (owms) are described systematically and placed into three species of Ouroboros gen. nov.: O. soroboruo gen. et sp. nov., O. vulgaris gen et sp. nov. and O. torus gen. et sp. nov. These species are widespread in the 1078 Ma Nonesuch Formation in the United States with minor occurrences in the Torridonian Sequence of northwest Scotland. These microfossil cysts were originally buried in the bottom sediments of a rift valley lake, and they share a common ecology with modern freshwater euglenids, including an association with cyanobacteria. As a member of the Discobid clade, the euglenids are positioned close to the base of the eukaryote tree in some molecular phylogenetic studies, so their occurrence as documented here would appear to support models of earlier rather than later Neoproterozoic diversification of eukaryote lineages. Ouroboros gen. nov. becomes one of only a handful of examples of Precambrian owms that have found a place in crown group eukaryotic lineages.

This study systematically investigates copper gettering in silicon by heavily doped n+ and p+ diffused layers fabricated on high-resistivity float-zone silicon wafers of both polarities, following intentional copper contamination of the wafer bulk. Experimental carrier lifetime analysis and photoluminescence imaging before and after dopant diffusion, corroborated by numerical gettering simulations based on the reported copper solubilities and diffusivities, demonstrate the gettering effect of n+ and p+ diffused layers for copper. The gettering effectiveness increases with increasing doping, as expected. Lightly doped regions exhibit limited gettering effects, for instance, an n+ region with a sheet resistance of 150 Ω/□ can only getter 10% of the bulk Cu. This implies that such lightly doped surface regions may be insufficient barriers against copper in-diffusion from metallized layers to the silicon wafer bulk. Furthermore, the impact of gettered Cu on the surface passivation quality of SiNx-coated n+ and Al2O3-coated p+ regions was evaluated. For heavily doped samples, Auger recombination remains the dominant effect that masks potential Cu-induced degradation. For lightly doped samples, surface recombination was mostly limited by the applied spin-on-dopant rather than the gettered Cu.

Trait anxiety is typically associated with impaired performance on cognitive tasks. In contrast, here we found that trait cognitive anxiety was associated with superior performance in detecting rare visual search targets. Specifically, the risk of missing searched-for targets is substantially higher when they appear infrequently (i.e., under low prevalence conditions) compared with when the same targets appear frequently (i.e., under high prevalence conditions). This phenomenon is called the Low Prevalence Effect (LPE). In a sample of 315 adults, we assessed the relationship between anxiety (cognitive and somatic, trait and state) and visual search performance under low and high prevalence conditions. Higher trait cognitive anxiety predicted a reduced LPE, driven specifically by a lower miss rate under low prevalence conditions. This relationship was mediated by a longer quitting threshold, such that individuals with higher levels of trait cognitive anxiety spent longer searching each array for the target under low prevalence conditions compared with individuals with lower levels of trait cognitive anxiety. Importantly, their time to detect present targets under high prevalence conditions was not impaired. These findings indicate that trait cognitive anxiety can be associated with superior task performance under conditions in which a more cautious task strategy is advantageous.

Graphical user interface (GUI) agents autonomously complete tasks across platforms (\eg, Linux) by sequentially decomposing user instructions into action proposals that iteratively interact with visual elements in the evolving environment. However, two main challenges arise: i) planning (\ie, the action proposal sequence) under expansive action space, where selecting an appropriate plan is non-trivial, as many valid ones may exist; ii) accurately grounding actions in complex and high-resolution interfaces, \ie, precisely interacting with visual targets. This paper investigates the aforementioned challenges with our \textbf{G}UI \textbf{T}est-time Scaling \textbf{A}gent, namely \name. First, we conduct test-time scaling to select the most appropriate action proposal: at each step, multiple candidate proposals are sampled and evaluated and selected by a judge model. It trades off computation for better decision quality by concurrent sampling. Second, we propose a model that improves grounding of the selected action proposals to its corresponding visual elements. Our key insight is that reinforcement learning (RL) facilitates grounding through inherent objective alignments, rewarding successful clicks on interface elements. Experimentally, \name achieves state-of-the-art performance on both grounding and agent task execution benchmarks.