An efficient ptychography reconstruction strategy through fine-tuning of large pre-trained deep learning model

With pre-trained large models and their associated fine-tuning paradigms being constantly applied in deep learning, the performance of large models achieves a dramatic boost, mostly owing to the improvements on both data quantity and quality. Next-generation synchrotron light sources offer ultra-bright and highly coherent X-rays, which are becoming one of the largest data sources for scientific experiments. As one of the most data-intensive scanning-based imaging methodologies, ptychography produces an immense amount of data, making the adoption of large deep learning models possible. Here, we introduce and refine the architecture of a neural network model to improve the reconstruction performance, through fine-tuning large pre-trained model using a variety of datasets. The pre-trained model exhibits remarkable generalization capability, while the fine-tuning strategy enhances the reconstruction quality. We anticipate this work will contribute to the advancement of deep learning methods in ptychography, as well as in broader coherent diffraction imaging methodologies in future.

Regulating steric hindrance in difunctionalized porous aromatic frameworks for the selective separation of Pb(II)

Efficient and selective removal of Pb(II) from wastewater with complex matrix remains a challenging task. Porous aromatic frameworks (PAFs) with predesigned functional building blocks provide a favorable platform for the selective separation of Pb(II). Herein, the bifunctional SPAFs with the introduction of -OH and -SO3H were synthesized through rationally optimizing their steric hindrance. As a result, the SPAF-0.75 exhibits favorable adsorption capacity of Pb(II) (212.34 mg g-1), which is 22 times larger than pristine framework. Competition experiment indicates that SPAF-0.75 possess the selective removal of Pb(II) without interfering from co-existing metal ions. The removal rate of SPAF-0.75 still retain at 100% after six successive cycles. The DFT calculation illustrates that -OH and -SO3H are co-participate in the process of capturing Pb(II), revealing SPAF-0.75 preferred removal of Pb(II) owing to the lowest adsorption energy (ΔEab = -3.99 eV). This study extend the understanding of the structure-property relationship and facilitate new possibilities for PAFs.

Emerging d-d orbital coupling between non-d-block main-group elements Mg and I at high pressure

d-d orbital coupling, which increases anisotropic and directional bonding, commonly occurs between d-block transition metals. Here, we report an unexpected d-d orbital coupling in the non-d-block main-group element compound Mg₂I based on first-principles calculations. The unfilled d orbitals of Mg and I atoms under ambient conditions become part of the valence orbitals and couple with each other under high pressures, resulting in the formation of highly symmetric I-Mg-I covalent bonding in Mg₂I, which forces the valence electrons of Mg atoms into the lattice voids to form interstitial quasi-atoms (ISQs). In turn, the ISQs highly interact with the crystal lattice, contributing to lattice stability. This study greatly enriches the fundamental understanding of chemical bonding between non-d-block main-group elements at high pressures.

MDSuite: comprehensive post-processing tool for particle simulations

Particle-Based (PB) simulations, including Molecular Dynamics (MD), provide access to system observables that are not easily available experimentally. However, in most cases, PB data needs to be processed after a simulation to extract these observables. One of the main challenges in post-processing PB simulations is managing the large amounts of data typically generated without incurring memory or computational capacity limitations. In this work, we introduce the post-processing tool: MDSuite. This software, developed in Python, combines state-of-the-art computing technologies such as TensorFlow, with modern data management tools such as HDF5 and SQL for a fast, scalable, and accurate PB data processing engine. This package, built around the principles of FAIR data, provides a memory safe, parallelized, and GPU accelerated environment for the analysis of particle simulations. The software currently offers 17 calculators for the computation of properties including diffusion coefficients, thermal conductivity, viscosity, radial distribution functions, coordination numbers, and more. Further, the object-oriented framework allows for the rapid implementation of new calculators or file-readers for different simulation software. The Python front-end provides a familiar interface for many users in the scientific community and a mild learning curve for the inexperienced. Future developments will include the introduction of more analysis associated with ab-initio methods, colloidal/macroscopic particle methods, and extension to experimental data.

First-principles calculations of hematite (α-Fe(2)O(3)) by self-consistent DFT+U+V

Owing to the confined Fe-3d orbitals and self-interaction error of exchange-correlation functionals, approximate DFT fails to describe iron oxides electronic structure and magnetic properties accurately. Hybrid DFT or DFT + U can solve these problems, but the former is expensive, and the latter only considers on-site interactions. Here, we used DFT + U + V, a DFT + U extension including inter-site interactions, to simulate the structural, magnetic, and electronic properties, along with Fe and O K-edge XAS spectra of α-Fe₂O₃. Two types of atomic orbital projectors were studied, orthogonalized and non-orthogonalized. DFT + U + V improves the description of the structural, magnetic, and electronic properties of α-Fe₂O₃ compared to approximate DFT. The accuracy of the correction depends on the orbital projector used. DFT + U + V with orthogonalized projectors achieves the best experimental agreement at a fraction of hybrid DFT cost. This work emphasizes the importance of inter-site interactions and the type of atomic orbital projectors used in the theoretical research of α-Fe₂O₃.

Investigation of the effect of air gap size on the spatial resolution in proton- and helium radio- and tomography

PURPOSE

Proton computed (transmission) tomography (pCT) refers to the process of imaging an object by letting protons pass through it, while measuring their energy after, and their position and (optionally) direction both before and after their traversal through that object. The so far experimental technique has potential to improve treatment planning of proton therapy by enabling the direct acquisition of a proton stopping power map of tissue, thus removing the need to obtain it by converting X-ray CT attenuation data and thereby eliminating uncertainties which arise in the mentioned conversion process. The image reconstruction in pCT requires accurate estimates of the proton trajectories. In experimental pCT detector setups where the direction of the protons is not measured, the air gap between the detector planes and the imaged object worsens the spatial resolution of the image obtained. In this work we determined the mean proton paths and the corresponding spatial uncertainty, taking into account the presence of the air gap.

METHODS

We used Monte Carlo simulations of radiation transport to systematically investigate the effect of the air gap size between detector and patient on the spatial resolution of proton (ion) computed tomography for protons with an energy of 200MeV and 250MeV as well as for helium ions (He-4) with an energy of 798MeV. For the simulations we used TOPAS which itself is based on Geant4.

RESULTS

For all particles, which are detected at the same entrance and exit coordinate, the average ion path and the corresponding standard deviation was computed. From this information, the dependence of the spatial resolution on the air gap size and the angular confusion of the particle beam was inferred.

CONCLUSION

The presence of the airgap does not pose a problem for perfect fan beams. In realistic scenarios, where the initial angular confusion is around 5mrad and for typical air gap sizes up to 10cm, using an energy of 200MeV a spatial resolution of about 1.6mm can be achieved. Using protons with E=250MeV a spatial resolution of about 1.1mm and using helium ions (He-4) with E=798MeV even a spatial resolution below 0.7mm respectively is attainable.

Vlasov methods in space physics and astrophysics

This paper reviews Vlasov-based numerical methods used to model plasma in space physics and astrophysics. Plasma consists of collectively behaving charged particles that form the major part of baryonic matter in the Universe. Many concepts ranging from our own planetary environment to the Solar system and beyond can be understood in terms of kinetic plasma physics, represented by the Vlasov equation. We introduce the physical basis for the Vlasov system, and then outline the associated numerical methods that are typically used. A particular application of the Vlasov system is Vlasiator, the world’s first global hybrid-Vlasov simulation for the Earth’s magnetic domain, the magnetosphere. We introduce the design strategies for Vlasiator and outline its numerical concepts ranging from solvers to coupling schemes. We review Vlasiator’s parallelisation methods and introduce the used high-performance computing (HPC) techniques. A short review of verification, validation and physical results is included. The purpose of the paper is to present the Vlasov system and introduce an example implementation, and to illustrate that even with massive computational challenges, an accurate description of physics can be rewarding in itself and significantly advance our understanding. Upcoming supercomputing resources are making similar efforts feasible in other fields as well, making our design options relevant for others facing similar challenges.

On the role of the second-order derivative term in the calculation of convergent beam diffraction patterns

The simulation of (scanning) transmission electron microscopy images and diffraction patterns is most often performed using the forward-scattering approximation where the second-order derivative term in z is assumed to be small with respect to the first-order derivative term in the modified Schrödinger equation. This assumption is very good at high incident electron energies, but breaks down at low energies. In order to study the differences between first- and second-order methods, convergent beam electron diffraction patterns were simulated for silicon at the [111] zone-axis orientation at 20 keV and compared using electron intensity difference maps and integrated intensity profiles. The geometrical differences in the calculated diffraction patterns could be explained by an Ewald surface analysis. Furthermore, it was found that solutions based on the second-order derivative equation contained small amplitude oscillations that need to be resolved in order to ensure numerical integration stability. This required the use of very small integration steps resulting in significantly increased computation time compared to the first-order differential equation solution. Lastly, the efficiency of the numerical integration technique is discussed.