Publications

Pt nanoparticle array model catalysts with 28 ± 2 nm diameters and 100 ± 2 nm interparticle spacing have been fabricated with electron beam lithography on alumina supports. A novel method for cleaning the Pt nanoparticle arrays, involving low dosages of NO2 and CO and mild temperature flashing, was established. This cleaning procedure was crucial for measuring reaction rates over the nanoparticle arrays. The reactivity of the Pt/Al2O3 arrays was compared to a Pt(111) single crystal for the ethylene hydrogenation reaction. The activation energy and the pressure dependence of the H2 and C2H4 on the nanoparticle array were in excellent agreement with the kinetic data on the Pt(111) single-crystal model catalyst. Because the ethylene hydrogenation reaction is structure insensitive, the rate equation for Pt(111) can be applied to the Pt nanoparticle arrays. The calculated turnover frequency led to a calculated active metal surface area that compared very well with an active metal surface area on the basis of geometry. This reaction can therefore be used to determine the active metal surface area of the Pt nanoparticle array model catalysts. The arrays were characterized with AFM, SEM, XPS, and AES before and after being exposed to reaction conditions.

Efficient luminescence extraction enables us to use the photovoltaic band gap as the spectral filter for thermophotovoltaics. Using this concept, we demonstrate a 28.2% conversion efficiency, higher than the previously reported value for the range of the emitter temperature (700oC–1250oC) used.

Modern datasets and models are notoriously difficult to explore and analyze due to their inherent high dimensionality and massive numbers of samples. Existing visualization methods which employ dimensionality reduction to two or three dimensions are often inefficient and/or ineffective for these datasets. This paper introduces T-SNE-CUDA, a GPU-accelerated implementation of t-distributed Symmetric Neighbour Embedding (t-SNE) for visualizing datasets and models. T-SNE-CUDA significantly outperforms current implementations with 50-700x speedups on the CIFAR-10 and MNIST datasets. These speedups enable, for the first time, visualization of the neural network activations on the entire ImageNet dataset - a feat that was previously computationally intractable. We also demonstrate visualization performance in the NLP domain by visualizing the GloVe embedding vectors. From these visualizations, we can draw interesting conclusions about using the L2 metric in these embedding spaces. T-SNE-CUDA is publicly available at https://github.com/CannyLab/tsne-cuda.

The Cadmium‐Lined Outer‐Core Irradiation Tube ( CLOCIT ) is a new irradiation facility for 40 Ar/ 39 Ar geochronology at the Oregon State University TRIGA ® reactor. We report fluence (i.e., time‐integrated flux) parameters from the first four CLOCIT irradiations and compare them with the existing Cadmium‐Lined Inner‐Core Irradiation Tube ( CLICIT ). CLOCIT provides an average neutron flux equivalent of 1.45–1.53 × 10 −4 J h −1 ; about 55% of CLICIT . Radial fluence gradients were on the order of 0.2–4.2% cm −1 . A planar fit of J ‐values results in residuals in the range of uncertainty in the J ‐value, but systematic deviations resolve a non‐planar component of the neutron flux field, which has also been observed in CLICIT . Axial neutron fluence gradients were 0.6–1% cm −1 , compared with 0.7–1.6% cm −1 for the CLICIT . Production rate ratios of interfering reactions were ( 40 Ar/ 39 Ar) K = (4 ± 6) × 10 −4 and ( 38 Ar/ 39 Ar) K = (1.208 ± 0.002) × 10 −2 , ( 36 Ar/ 37 Ar) Ca = (2.649 ± 0.014) × 10 −4 , ( 38 Ar/ 37 Ar) Ca = (3.33 ± 0.12) × 10 −5 and ( 39 Ar/ 37 Ar) Ca = (9.1 ± 0.28) × 10 −4 , similar to the CLICIT values.

<p>Supplementary Table 1. The Comparisons of Patient Characteristics in All (N = 178), CT (N = 170), DXA (N = 162), and D3Cr (N = 118) Groups</p>