A chemical precursor mediated process was used to form catalyst nanoparticles (NPs) with an extremely high density (10(14) to 10(16) m(-2)), controllable size distribution (3-20 nm), and good thermal stability at high temperature (900 °C). This used metal cations deposited in layered double hydroxides (LDHs) to give metal catalyst NPs by reduction. The key was that the LDHs had their intercalated anions selected and exchanged by guest-host chemistry to prevent sintering of the metal NPs, and there was minimal sintering even at 900 °C. Metal NPs on MoO(4)(2-) intercalated Fe/Mg/Al LDH flakes were successfully used as the catalyst for the double helix growth of single-walled carbon nanotube arrays. The process provides a general method to fabricate thermally stable metal NPs catalysts with the desired size and density for catalysis and materials science.

Transformers, the default model of choices in natural language processing, have drawn scant attention from the medical imaging community. Given the ability to exploit long-term dependencies, transformers are promising to help atypical convolutional neural networks (convnets) to overcome its inherent shortcomings of spatial inductive bias. However, most of recently proposed transformer-based segmentation approaches simply treated transformers as assisted modules to help encode global context into convolutional representations without investigating how to optimally combine self-attention (i.e., the core of transformers) with convolution. To address this issue, in this paper, we introduce nnFormer (i.e., not-another transFormer), a powerful segmentation model with an interleaved architecture based on empirical combination of self-attention and convolution. In practice, nnFormer learns volumetric representations from 3D local volumes. Compared to the naive voxel-level self-attention implementation, such volume-based operations help to reduce the computational complexity by approximate 98% and 99.5% on Synapse and ACDC datasets, respectively. In comparison to prior-art network configurations, nnFormer achieves tremendous improvements over previous transformer-based methods on two commonly used datasets Synapse and ACDC. For instance, nnFormer outperforms Swin-UNet by over 7 percents on Synapse. Even when compared to nnUNet, currently the best performing fully-convolutional medical segmentation network, nnFormer still provides slightly better performance on Synapse and ACDC. Codes and models are available at https://github.com/282857341/nnFormer.

Laboratory experiments measured the velocity inside a model meadow of submerged, flexible vegetation under 1 and 2 s period waves. The model plant consisted of a rigid stem and strap‐like blades, similar to the seagrass Zostera marina and the freshwater eelgrass Vallisneria Americana. The ratio of wave excursion (Aw) to stem spacing (S) determined whether, or not, plant‐generated turbulence enhanced the turbulence level within the meadow, compared to bare bed. Specifically, near‐bed turbulence was enhanced for conditions with Aw/S > 0.5, and for these conditions the turbulence (TKE) normalized by the RMS wave velocity squared, TKE/Uw,RMS2, increased monotonically with the plant solid volume fraction, ϕ. The plant‐generated turbulence was greater in the stem region than in the blade region, and this was attributed to the greater relative motion between the waves and rigid stem, compared to the flexible blades. A model previously developed to predict TKE in unidirectional flow through a rigid emergent canopy was modified by replacing the time‐mean current with the RMS wave velocity. With a fitted scale coefficient, the modified model predicts TKE as a function of RMS wave velocity in the meadow, stem and blade geometry, and solid volume fraction. Wave decay was also measured and shown to have a linear correlation with the measured TKE within the canopy, providing a second method to predict meadow TKE in the field.