Search for: All records

Earth’s inner core is predominantly composed of solid iron (Fe) and displays intriguing properties such as strong shear softening and an ultrahigh Poisson’s ratio. Insofar, physical mechanisms to explain these features coherently remain highly debated. Here, we have studied longitudinal and shear wave velocities of hcp-Fe (hexagonal close-packed iron) at relevant pressure–temperature conditions of the inner core using in situ shock experiments and machine learning molecular dynamics (MLMD) simulations. Our results demonstrate that the shear wave velocity of hcp-Fe along the Hugoniot in the premelting condition, defined asT/T_m(T_m: melting temperature of iron) above 0.96, is significantly reduced by ~30%, while Poisson’s ratio jumps to approximately 0.44. MLMD simulations at 230 to 330 GPa indicate that collective motion with fast diffusive atomic migration occurs in premelting hcp-Fe primarily along [100] or [010] crystallographic direction, contributing to its elastic softening and enhanced Poisson’s ratio. Our study reveals that hcp-Fe atoms can diffusively migrate to neighboring positions, forming open-loop and close-loop clusters in the inner core conditions. Hcp-Fe with collective motion at the inner core conditions is thus not an ideal solid previously believed. The premelting hcp-Fe with collective motion behaves like an extremely soft solid with an ultralow shear modulus and an ultrahigh Poisson’s ratio that are consistent with seismic observations of the region. Our findings indicate that premelting hcp-Fe with fast diffusive motion represents the underlying physical mechanism to help explain the unique seismic and geodynamic features of the inner core.

 

With the recent advances in data science, machine learning has been increasingly applied to convection and cloud parameterizations in global climate models (GCMs). This study extends the work of Han et al. (2020,https://doi.org/10.1029/2020MS002076) and uses an ensemble of 32‐layer deep convolutional residual neural networks, referred to as ResCu‐en, to emulate convection and cloud processes simulated by a superparameterized GCM, SPCAM. ResCu‐en predicts GCM grid‐scale temperature and moisture tendencies, and cloud liquid and ice water contents from moist physics processes. The surface rainfall is derived from the column‐integrated moisture tendency. The prediction uncertainty inherent in deep learning algorithms in emulating the moist physics is reduced by ensemble averaging. Results in 1‐year independent offline validation show that ResCu‐en has high prediction accuracy for all output variables, both in the current climate and in a warmer climate with +4K sea surface temperature. The analysis of different neural net configurations shows that the success to generalize in a warmer climate is attributed to convective memory and the 1‐dimensional convolution layers incorporated into ResCu‐en. We further implement a member of ResCu‐en into CAM5 with real world geography and run the neural‐network‐enabled CAM5 (NCAM) for 5 years without encountering any numerical integration instability. The simulation generally captures the global distribution of the mean precipitation, with a better simulation of precipitation intensity and diurnal cycle. However, there are large biases in temperature and moisture in high latitudes. These results highlight the importance of convective memory and demonstrate the potential for machine learning to enhance climate modeling.

 

As one of the popular deep learning methods, deep convolutional neural networks (DCNNs) have been widely adopted in segmentation tasks and have received positive feedback. However, in segmentation tasks, DCNN-based frameworks are known for their incompetence in dealing with global relations within imaging features. Although several techniques have been proposed to enhance the global reasoning of DCNN, these models are either not able to gain satisfying performances compared with traditional fully-convolutional structures or not capable of utilizing the basic advantages of CNN-based networks (namely the ability of local reasoning). In this study, compared with current attempts to combine FCNs and global reasoning methods, we fully extracted the ability of self-attention by designing a novel attention mechanism for 3D computation and proposed a new segmentation framework (named 3DTU) for three-dimensional medical image segmentation tasks. This new framework processes images in an end-to-end manner and executes 3D computation on both the encoder side (which contains a 3D transformer) and the decoder side (which is based on a 3D DCNN). We tested our framework on two independent datasets that consist of 3D MRI and CT images. Experimental results clearly demonstrate that our method outperforms several state-of-the-art segmentation methods in various metrics. 

We report a single atom Rh₁/CeO₂catalyst prepared by the high temperature (800 °C) atom trapping (AT) method which is stable under both oxidative and reductive conditions. Infrared spectroscopic and electron microscopy characterization revealed the presence of exclusively ionic Rh species. These ionic Rh species are stable even under reducing conditions (CO at 300 °C) due to the strong interaction between Rh and CeO₂achieved by the AT method, leading to high and reproducible CO oxidation activity regardless of whether the catalyst is reduced or oxidized. In contrast, ionic Rh species in catalysts synthesized by a conventional impregnation approach (e. g., calcined at 350 °C) can be readily reduced to form Rh nanoclusters/nanoparticles, which are easily oxidized under oxidative conditions, leading to loss of catalytic performance. The single atom Rh₁/CeO₂catalysts synthesized by the AT method do not exhibit changes during redox cycling hence are promising catalysts for emission control where redox cycling is encountered, and severe oxidation (fuel cut) leads to loss of performance.

 

We present a comparative study of two nearby type Ia supernovae (SNe Ia), 2018xx and 2019gbx, that exploded in NGC 4767 and MCG-02-33-017 at a distance of 48 Mpc and 60 Mpc, respectively. The B -band light curve decline rate for SN 2018xx is estimated to be 1.48 ± 0.07 mag and for SN 2019gbx it is 1.37 ± 0.07 mag. Despite the similarities in photometric evolution, quasi-bolometric luminosity, and spectroscopy between these two SNe Ia, SN 2018xx has been found to be fainter by about ∼0.38 mag in the B -band and has a lower 56 Ni yield. Their host galaxies have similar metallicities at the SN location, indicating that the differences between these two SNe Ia may be associated with the higher progenitor metallicity of SN 2018xx. Further inspection of the near-maximum-light spectra has revealed that SN 2018xx has relatively strong absorption features near 4300 Å relative to SN 2019gbx. The application of the code TARDIS fitting to the above features indicates that the absorption features near 4300 Å appear to be related to not only Fe  II /Mg  II abundance but possibly to the other element abundances as well. Moreover, SN 2018xx shows a weaker carbon absorption at earlier times, which is also consistent with higher ejecta metallicity.