Search

The optimal performance of the Kalman filters is highly dependent on the measurement and process noise characteristics, making the whole system unable to achieve the desired estimation in the presence of non-Gaussian mean noise distribution and high initial uncertainties. Recently, the H-infinity filter, as a robust algorithm, has been broadly used, as it is not being dependent on the pre-knowledge of the noise nature; however, making a balance between high robustness and estimation accuracy is a challenging issue. Hence, to overcome this problem, a new adaptive H-infinity extended Kalman filter (AHEKF) was designed in this paper, which benefits from both high robustness and precision. The suggested algorithm contains two adaptive sections to achieve high accuracy as well as controlling the effects of time-varying noise characteristics, high initial uncertainties, and abnormal data that can degrade the accuracy of state estimation in an integrated navigation system. The presented algorithm was used to integrate data from two independent sensors data. The simulation results for an inertial navigation system (INS)/global positioning system (GPS) sensor fusion are presented and compared with the standard H-infinity filter, extended Kalman filter (EKF), and unscented Kalman filter (UKF) to show the effectiveness of the proposed algorithm. Evaluations demonstrate that the AHEKF achieves over 50% higher accuracy and robustness, and over 2.5 times faster convergence of estimation errors than the standard H-infinity filter. [ABSTRACT FROM AUTHOR]

Single particle imaging at atomic resolution is perhaps one of the most desired goals for ultrafast X-ray science with X-ray free-electron lasers. Such a capability would create great opportunity within the biological sciences, as high-resolution structural information of biosamples that may not crystallize is essential for many research areas therein. In this paper, we report on a comprehensive computational study of diffraction image formation during single particle imaging of a macromolecule, containing over one hundred thousand non-hydrogen atoms. For this study, we use a dedicated simulation framework, SIMEX, available at the European XFEL facility. Our results demonstrate the full feasibility of computational single-particle imaging studies for biological samples of realistic size. This finding is important as it shows that the SIMEX platform can be used for simulations to inform relevant single-particle-imaging experiments and help to establish optimal parameters for these experiments. This will enable more focused and more efficient single-particle-imaging experiments at XFEL facilities, making the best use of the resource-intensive XFEL operation., (© 2024. The Author(s).)

X-ray diffraction of silicon irradiated with tightly focused femtosecond x-ray pulses (photon energy, 11.5 keV; pulse duration, 6 fs) was measured at various x-ray intensities up to 4.6×10^{19}  W/cm^{2}. The measurement reveals that the diffraction intensity is highly suppressed when the x-ray intensity reaches of the order of 10^{19}  W/cm^{2}. With a dedicated simulation, we confirm that the observed reduction of the diffraction intensity can be attributed to the femtosecond change in individual atomic scattering factors due to the ultrafast creation of highly ionized atoms through photoionization, Auger decay, and subsequent collisional ionization. We anticipate that this ultrafast reduction of atomic scattering factor will be a basis for new x-ray nonlinear techniques, such as pulse shortening and contrast variation x-ray scattering.

The noise caused by sample heterogeneity (including sample solvent) has been identified as one of the determinant factors for a successful X-ray single-particle imaging experiment. It influences both the radiation damage process that occurs during illumination as well as the scattering patterns captured by the detector. Here, we investigate the impact of water layer thickness and radiation damage on orientation recovery from diffraction patterns of the nitrogenase iron protein. Orientation recovery is a critical step for single-particle imaging. It enables to sort a set of diffraction patterns scattered by identical particles placed at unknown orientations and assemble them into a 3D reciprocal space volume. The recovery quality is characterized by a "disconcurrence" metric. Our results show that while a water layer mitigates protein damage, the noise generated by the scattering from it can introduce challenges for orientation recovery and is anticipated to cause problems in the phase retrieval process to extract the desired protein structure. Compared to these disadvantageous effects due to the thick water layer, the effects of radiation damage on the orientation recovery are relatively small. Therefore, minimizing the amount of residual sample solvent should be considered a crucial step in improving the fidelity and resolution of X-ray single-particle imaging experiments., (© 2023. Springer Nature Limited.)

The interaction of a high intensity x-ray pulse with matter causes ionization of the constituent atoms through various atomic processes, and the system eventually goes through a complex structural dynamics. Understanding this whole process is important from the perspective of structure determination of molecules using single particle imaging. XMDYN, which is a classical molecular dynamics-Monte Carlo based hybrid approach, has been successful in simulating the dynamical evolution of various systems under intense irradiation over the past years. The present study aims for extending the XMDYN toolkit to treat chemical bonds using the reactive force field. In order to study its impact, a highly intense x-ray pulse was made to interact with the simplest amino acid, glycine. Different model variants were used to highlight the consequences of charge rearrangement and chemical bonds on the time evolution. The charge-rearrangement-enhanced x-ray ionization of molecules effect is also discussed to address the capability of a classical MD based approach, i.e., XMDYN, to capture such a molecular phenomenon., (© 2022 Author(s).)

High energy density (HED) matter exists extensively in the Universe, and it can be created with extreme conditions in laboratory facilities such as x-ray free-electron lasers (XFEL). In HED matter, the electronic structure of individual atomic ions is influenced by a dense plasma environment, and one of the most significant phenomena is the ionization potential depression (IPD). Incorporation of the IPD effects is of great importance in accurate modeling of dense plasmas. All theoretical treatments of IPD so far have been based on the assumption of local thermodynamic equilibrium, but its validity is questionable in ultrafast formation dynamics of dense plasmas, particularly when interacting with intense XFEL pulses. A treatment of transient IPD, based on an electronic-structure calculation of an atom in the presence of a plasma environment described by classical particles, has recently been proposed [Phys. Rev. E 103, 023203 (2021)2470-004510.1103/PhysRevE.103.023203], but its application to and impact on plasma dynamics simulations have not been investigated yet. In this work, we extend XMDYN, a hybrid quantum-classical approach combining Monte Carlo and molecular dynamics, by incorporating the proposed IPD treatment into plasma dynamics simulations. We demonstrate the importance of the IPD effects in theoretical modeling of aluminum dense plasmas by comparing two XMDYN simulations: one with electronic-structure calculations of isolated atoms (without IPD) and the other with those of atoms embedded in a plasma (with IPD). At equilibrium, the mean charge obtained in the plasma simulation with IPD is in good agreement with the full quantum-mechanical average-atom model. The present approach promises to be a reliable tool to simulate the creation and nonequilibrium evolution of dense plasmas induced by ultraintense and ultrashort XFEL pulses.

In this work, we report on incorporating for the first time tree-algorithm based solvers into the molecular dynamics code, XMDYN. XMDYN was developed to describe the interaction of ultrafast X-ray pulses with atomic assemblies. It is also a part of the simulation platform, SIMEX, developed for computational single-particle imaging studies at the SPB/SFX instrument of the European XFEL facility. In order to improve the XMDYN performance, we incorporated the existing tree-algorithm based Coulomb solver, PEPC, into the code, and developed a dedicated tree-algorithm based secondary ionization solver, now also included in the XMDYN code. These extensions enable computationally efficient simulations of X-ray irradiated large atomic assemblies, e.g., large protein systems or viruses that are of strong interest for ultrafast X-ray science. The XMDYN-based preparatory simulations can now guide future single-particle-imaging experiments at the free-electron-laser facility, EuXFEL.