Traffic speed prediction is known as an important but challenging problem. In this paper, we propose a novel model, called LC-RNN, to achieve more accurate traffic speed prediction than existing solutions. It takes advantage of both RNN and CNN models by a rational integration of them, so as to learn more meaningful time-series patterns that can adapt to the traffic dynamics of surrounding areas. Furthermore, since traffic evolution is restricted by the underlying road network, a network embedded convolution structure is proposed to capture topology aware features. The fusion with other information, including periodicity and context factors, is also considered to further improve accuracy. Extensive experiments on two real datasets demonstrate that our proposed LC-RNN outperforms six well-known existing methods.

In this paper, density functional theory (DFT) was performed to study the adsorption properties of ornidazole on anatase TiO 2 (101) and (001) crystal facets under vacuum, neutral and acid-base conditions. We calculated the adsorption structure of ornidaozle on the anatase TiO 2 surface, optimal adsorption sites, adsorption energy, density of states, electronic density and Milliken atomic charge under different conditions. The results show that when the N(3) atom on the imidazole ring is adsorbed on the Ti(5) atom, the largest adsorption energy and the most stable adsorption configuration could be achieved. According to the analysis of the adsorption configuration, we found that the stability of C(2)-N(3) bond showed a weakening trend. The adsorption wavelengths of the electronic transition between the valence band and conduction band of ornidazole on the TiO 2 surface were in the visible light wavelengths range, showing that the TiO 2 crystal plane can effectively make use of visible light under different conditions. We speculate the possibility of ornidazole degradation on the surface of TiO 2 and found that the reactive site is the C-N bond on the imidazole ring. These discoveries explain the photocatalytic degradation of ornidazole by TiO 2 and reveal the microscopic nature of catalytic degradation.

No abstract

We present the encoder-forecaster convolutional long short-term memory (LSTM) deep-learning model that powers Microsoft Weather's operational precipitation nowcasting product. This model takes as input a sequence of weather radar mosaics and deterministically predicts future radar reflectivity at lead times up to 6 hours. By stacking a large input receptive field along the feature dimension and conditioning the model's forecaster with predictions from the physics-based High Resolution Rapid Refresh (HRRR) model, we are able to outperform optical flow and HRRR baselines by 20-25% on multiple metrics averaged over all lead times.Preprint. Under review.