Accurate and real-time traffic forecasting plays an important role in the Intelligent Traffic System and is of great significance for urban traffic planning, traffic management, and traffic control. However, traffic forecasting has always been considered an open scientific issue, owing to the constraints of urban road network topological structure and the law of dynamic change with time, namely, spatial dependence and temporal dependence. To capture the spatial and temporal dependence simultaneously, we propose a novel neural network-based traffic forecasting method, the temporal graph convolutional network (T-GCN) model, which is in combination with the graph convolutional network (GCN) and gated recurrent unit (GRU). Specifically, the GCN is used to learn complex topological structures to capture spatial dependence and the gated recurrent unit is used to learn dynamic changes of traffic data to capture temporal dependence. Then, the T-GCN model is employed to traffic forecasting based on the urban road network. Experiments demonstrate that our T-GCN model can obtain the spatio-temporal correlation from traffic data and the predictions outperform state-of-art baselines on real-world traffic datasets. Our tensorflow implementation of the T-GCN is available at https://github.com/lehaifeng/T-GCN.

The static dielectric constant of liquid water is computed using classical force field based molecular dynamics simulation at fixed electric displacement D. The method to constrain the electric displacement is the finite temperature classical variant of the constant-D method developed by Stengel, Spaldin and Vanderbilt [Nat. Phys. 5, 304, (2009)]. There is also a modification of this scheme imposing fixed values of the macroscopic field E. The method is applied to the popular SPC/E model of liquid water. We compare four different estimates of the dielectric constant, two obtained from fluctuations of the polarization at D = 0 and E = 0 and two from the variation of polarization with finite D and E. It is found that all four estimates agree when properly converged. The computational effort to achieve convergence varies however, with constant D calculations being substantially more efficient. We attribute this difference to the much shorter relaxation time of longitudinal polarization compared to transverse polarization accelerating constant D calculations.

Investigations of dielectric properties of liquid water in nanoconfinement are highly relevant for the energy storage in electrochemical systems, mineral-fluid interactions in geochemistry and microfluid based devices in biomedical analysis 1 . It has been reported that polarization shows a strong anisotropy at water interfaces 2,3 and the dielectric constant of nanoconfined water is surprisingly low ( ⊥ ∼ 10) 4 . Here, using a simple capacitor model, we show that the low dielectric constant of nanoconfined water can be largely accounted by the so-called dielectric dead-layer effect known for ferroelectric nanocapacitors 5 .Before talking about the effect of nanoconfinement, one needs to realize that the first effect of having an interface corresponds to a switch in the electric boundary condition. From classical electrodynamics, we know that the electric field E z is discontinuous at a dielectric interface, that is the reason why it is convenient to use the electric displacement D as the fundamental variable instead 6 . In the latter case, the polarization of dielectrics P z is added into the electric field. This makes D continuous in the direction perpendicular to an interface and leads to itsWhen the electric boundary condition is switched from constant electric field E to constant electric displacement D, the dielectric response will be different accordingly 7-10 .The difference in χ due to the electric boundary condition leads to differences in the fluctuation of polarization at zero field and in the corresponding relaxation time. This phenomenon is not limited to water in nanoconfinement where switching of electric boundary condition is enforced by introducing explicit interfaces 2,3 but can also be realized in bulk liquid water by turning on the constant electric displacement simulation 9,10 . For bulk liquid water, ⊥ = even though χ ⊥ is radically different from χ . Now let us go back to the original question: What accounts for the low dielectric constant ⊥ of water slab in nanoconfinement 4 . Is water in nanoconfinement completely different from that in the bulk ?To answer this question, we applied the constant electric displacement simulation with D = 0.6835V/Å to a water slab confined between two hydrophobic walls at a) Electronic mail: chao.zhang@kemi.uu. se FIG. 1. a) A snapshot of MD simulations of water slab confined between rigid walls under constant electric displacement D = 0.6835V/Å . The seperation distance between walls Lw is 30.77Å in this case.; b) The corresponding electrostatic potential profile ϕ(z) generated from the charge density. The slope gives the negative of the deploarization field 4πP ⊥ in the bulk water region. ambient conditions (Fig. 1a). Interactions between water molecules is described by simle point charge/extended (SPC/E) model 11 and the rigid hydrophobic walls are composed of atoms on a dense cubic lattice. All molecular dynamics (MD) simulations were performed with GROMACS 4 package 12 and technical settings are the same as described in the previous work 9 .From the ...

Key Points Question Are germline variants of natural killer (NK) cells associated with tumor immune microenvironment subtypes, cancer risk, prognosis, and immunotherapy? Findings This genetic association study analyzed functionally mutated genes in the germline genomes of 5883 patients with 13 common cancers and 4500 individuals with no cancer, finding that the number of functionally mutated genes in NK cell germlines was negatively associated with the abundance of tumor-infiltrating lymphocytes, clinical outcomes, and immunotherapy response but positively associated with cancer risk. Meaning Findings suggest that germline genetic variants in NK cells could help to identify individuals at risk of cancer and to improve existing immune checkpoint and chimeric antigen receptor–T cell therapies by adoptive transfer of healthy NK cells.

A micrometer-sized europium(iii)-organic framework with asymmetric binuclear metal subunits extended by 4,5-dichlorophthalaten (DCPA), [Eu(HO)(DCPA)], was easily obtained using a reverse microemulsion method. The framework exhibits good dispersibility, excellent thermal and environmental stability and easy regeneration ability. More importantly, the complex displays strong red emission and can selectively and sensitively detect both inorganic CrO anions (K = 8.7 × 10 M) and organic picric acid contaminants (K = 1.07 × 10 M) in water systems through fluorescence quenching. A luminescent film of 1 was further prepared and successfully used to detect the CrO anion in an aqueous system. These interesting results indicate that the well-dispersed europium(iii)-organic framework can serve as a promising dual-responsive luminescent sensor for environmental pollutant monitoring.