SearcharxivSearch

arXiv subjects

Johannes Roth

Publications and source records attributed to Johannes Roth.

At least 19 recordsLinked to original sources

Conditional Normalizing Flow for Gas-Surface Scattering from Thermal to Hypersonic Velocities

Accurate aerodynamic modeling of satellites in very low Earth orbit (VLEO) requires gas-surface interaction (GSI) models that capture the full velocity spectrum from thermal to orbital speeds. Atmospheric particles initially strike spacecraft surfaces at hypersonic velocities of 6 000 - 10 000 m/s. Due to surface roughness and complex geometries, especially within air-breathing electric propulsion (ABEP) intake systems, multiple collisions occur, progressively reducing the particle velocities. A recent machine learning framework for deriving scattering kernels from molecular dynamics (MD) simulations has shown promise, but remains limited to high-velocity single impacts and possibly violates fundamental equilibrium principles such as detailed balance. This work extends this machine learning based scattering kernel to cover the complete velocity range using conditional normalizing flows trained with physics-informed constraints, enabling accurate modeling of multi-bounce scenarios in realistic VLEO applications. We train a conditional Real-valued Non-Volume Preserving (cRealNVP) model on expanded molecular dynamics simulations covering velocities from thermal to hypersonic speeds, incorporating a detailed balance loss term. The resulting model demonstrates improved accuracy compared to previous approaches even in the original high-velocity regime, while successfully capturing thermal-velocity scattering. Quantitative assessment shows that thermalization is approximated within acceptable tolerances. This framework provides essential capabilities for accurate ABEP intake optimization and VLEO mission planning while offering a general methodology applicable to broader rarefied gas dynamics problems requiring thermodynamic consistency.

physics.comp-ph

Characterizing Universal Object Representations Across Vision Models

Deep neural networks trained with different architectures, objectives, and datasets have been reported to converge on similar visual representations. However, what remains unknown is which visual properties models actually converge on and which factors may underlie this convergence. To address this, we decompose the object similarity structure of 162 diverse vision models into a small set of non-negative dimensions. To determine universal versus model-specific dimensions, we then estimate how often each dimension reappears across models. In contrast to model-specific dimensions, universal dimensions are more interpretable and more strongly driven by conceptual image properties, indicating the relevance of interpretability and semantic content as implicit factors driving universality across models. Differences in architecture, objective function, training data, model size, and model performance do not explain the emergence of universal dimensions. However, models with more universal dimensions also better predict macaque IT activity and human similarity judgments, suggesting that universality reflects representations relevant to biological vision. These findings have important implications for understanding the emergent representations underlying deep neural network models and their alignment with biological vision.

cs.CV

Energy conservation and pressure relaxation in an extended two-temperature model for copper with an electron temperature-dependent interaction potential

An implementation of an electron temperature-dependent interaction potential for copper in a two-temperature model-molecular dynamics framework is presented. An algorithm for enforcing energy conservation when using such an interaction is provided that is needed due to the changing interaction strength with the degree of excitation. Furthermore, focus is put on how to treat the pressure differences due to an electron temperature gradient following laser irradiation. The influence of various extensions is investigated in large-scale two-temperature model molecular dynamics simulations and compared to existing approaches.

cond-mat.mtrl-sci

Novel non-thermal Ablation Mechanics in the Laser Ablation of Silicon

We investigate the non-thermal material dynamics of strongly excited silicon during ultra-fast laser ablation. In contrast to metals, silicon shows strongly excitation-dependent interatomic bonding strengths, which gives rise to a number of unique material dynamics like non-thermal melting, Coulomb explosions and altered carrier heat conduction due to charge carrier confinement. In this study, we report novel non-thermal ablation mechanisms in the ultra-fast single shot laser ablation of silicon and perform large scale massive multi-parallel simulations on experimentally achievable length scales with atomistic resolution. For this, we model the ultra-fast carrier dynamics utilizing the Thermal-Spike-Model coupled to Molecular Dynamics simulations and include the accompanied excitation-dependent nonthermal bonding strength manipulation by application of the excitation-dependent modified Tersoff Potential. Further, we present first results on the systematic construction of the excitation-dependent phase diagram of silicon by thermodynamic integration.

cond-mat.mtrl-sci

Development and validation of an electron temperature-dependent interaction potential for silicon and copper for the use in atomistic simulations of laser ablation

Laser pulses with a duration of the order of femtoseconds lead to a strong excitation, heating and potentially to ablation of the irradiated material. During the time of strong excitation, the interaction of the atoms and thus the material dynamics can be strongly altered. To take this effect into account, an ip was developed for copper that takes the excitation of the electrons into account up to an electron temperature of 1.2 eV. Furthermore, several ways to identify non-thermal effects in Density Functional Theory calculations and how to incorporate and validate them in molecular dynamics simulations are presented. Explicitly, the free energy curves, elastic constants and phonon spectra are compared. Additionally, it is shown that the change of the melting temperature with the degree of excitation is consistent with all of these properties. Moreover, the behaviour of copper upon excitation is compared to silicon by using a similar potential that was previously developed by a different author.

cond-mat.mtrl-sci

A Machine Learning Framework for Scattering Kernel Derivation Using Molecular Dynamics Data in Very Low Earth Orbit

The free molecular flow regime in VLEO makes gas-surface interactions (GSIs) crucial for satellite aerodynamic modeling. The Direct Simulation Monte Carlo (DSMC) method is required to estimate aerodynamic forces due to the breakdown of the continuum assumption. In DSMC, the Maxwell model is the most widely used approach for GSI. It simplifies the process by treating it as a superposition of diffuse and specular reflections while assuming a constant accommodation coefficient. In reality, this coefficient is influenced by multiple factors, such as the angle and magnitude of the incident velocity. A high-precision GSI model could significantly improve satellite aerodynamics optimization and the design of efficient intakes for atmospheric breathing propulsion systems. This advancement would greatly refine mission planning and fuel requirement calculations, ultimately extending operational lifetimes and lowering costs. To gain a deep understanding of the GSI at the microscopic level, molecular dynamics (MD) simulations provide valuable insights into the physical processes involved. However, due to computational limitations, simulating an entire satellite is impractical. Instead, we use MD to analyze the impact of selected velocity vectors on a amorphous $\text{Al}_2\text{O}_3$ surface. The obtained scattering kernels for the respective velocity vectors are then used to train a conditional Variational Autoencoder (cVAE). This model is able to generate scattering kernels for any incident velocity vector and can be integrated into DSMC simulations, significantly enhancing their accuracy. Applications of this model on a flat plate have shown that the cVAE is able to predict the shift from diffuse to quasi-specular reflection with increasing polar angle. Additionally, the aerodynamic coefficients and molecular fluxes are considerably different from those obtained with the Maxwell model.

physics.comp-ph

Advanced Boundary Conditions for the Simulations of Laser Ablation

Irradiating materials with ultra-short laser pulses generates unwanted shock waves which distort the interesting physics. Advanced boundary conditions are presented to erase even very strong shock waves. Depending on the wave length, laser light which leads to ablation may penetrate very deep into the sample. Connection conditions are presented to consistently couple atomistics and continuum description of the sample even at high temperatures together with a discussion of the temperature definition.

cond-mat.mtrl-sci

Laser Ablation of Al-Ni Alloys and Al-Ni Layer Systems simulated with Molecular Dynamics and the Two-Temperature Model

Laser ablation of Al-Ni alloys and Al films on Ni substrates has been studied by molecular dynamics simulations (MD). The MD method was combined with a two-temperature model to describe the interaction between the laser beam, the electrons and the atoms. The challenge for alloys and mixtures is to find the electronic parameters: electron heat conductivity, electron heat capacity and electron-phonon coupling parameter. The challenge for layered systems is to run simulations of an inhomogeneous system which requires modification of the simulation code. Ablation and laser-induced melting was studied for several Al-Ni compounds. At low fluences above the threshold ordinary ablation behavior occurred while at high fluences the ablation mechanism changed in Al$_3$Ni and AlNi$_3$ from phase explosion to vaporization. Al films of various thicknesses on a Ni substrate have also been simulated. Above threshold, 8 nm Al films are ablated as a whole while 24 nm Al films are only partially removed. Below threshold, alloying with a mixture gradient has been observed in the thin layer system.

cond-mat.mtrl-sci

Classical interaction potentials for diverse materials from ab initio data: a review of potfit

Force matching is an established technique to generate effective potentials for molecular dynamics simulations from first-principles data. This method has been implemented in the open source code potfit. Here, we present a review of the method and describe the main features of the code. Particular emphasis is placed on the features added since the initial release: interactions represented by analytical functions, differential evolution as optimization method, and a greatly extended set of interaction models. Beyond the initially present pair and embedded-atom method potentials, potfit can now also optimize angular dependent potentials, charge and dipolar interactions, and electron-temperature-dependent potentials. We demonstrate the functionality of these interaction models using three example systems: phonons in type I clathrates, fracture of α-alumina, and laser-irradiated silicon.

cond-mat.mtrl-sci

Simulation of Laser Ablation in Aluminum: The Effectivity of Double Pulses

Lasers are becoming a more and more important tool in cutting and shaping materials. Improving precision and effectivity is an ongoing demand in science and industry. One possibility are double pulses. Here we study laser ablation of aluminum by the two-temperature model. There the laser is modeled as a source in a continuum heat conduction equation for the electrons, whose temperature then is transferred to a molecular dynamics particle model by an electron-phonon coupling term. The melting and ablation effectivity is investigated depending on the relative intensity and the time delay between two Gaussian shaped laser pulses. It turns out that at least for aluminum the optimal pulse shapes are standard Gaussian pulses. For double pulses with delay times up to 200 ps we find a behavior as observed in experiment: the ablation depth decreases beyond a delay of 10 ps even if one does not account for the weakening at the second pulse due to laser-plasma interaction.

cond-mat.mtrl-sci

Statistical investigations on nitrogen-vacancy center creation

Experiments show that shallow nitrogen implantations (<10keV) result in a negatively charged nitrogen-vacancy center (NV-) yield of 0.01-0.1%. The most succesful technique for introducing NV- centers in the carbon matrix is ion implantation followed by annealing at 1100K. We investigated the influence of channeling effects during shallow implantation and statistical diffusion using molecular dynamics (MD) and Monte Carlo (MC) approaches. Energy barriers for the diffusion process were calculated using the density functional theory (DFT). Our simulations show a significant difference in the NV yield compared to the experiment. Statistically, 25% of the implanted nitrogens form a NV center after annealing.

cond-mat.mtrl-sci

Influence of polarizability on metal oxide properties studied by molecular dynamics simulations

We have studied the dependence of metal oxide properties in molecular dynamics (MD) simulations on the polarizability of oxygen ions. We present studies of both liquid and crystalline structures of silica (SiO2), magnesia (MgO) and alumina (Al2O3). For each of the three oxides, two separately optimized sets of force fields were used: (i) Long-range Coulomb interactions between oxide and metal ions combined with a short-range pair potential. (ii) Extension of force field (i) by adding polarizability to the oxygen ions. We show that while an effective potential of type (i) without polarizable oxygen ions can describe radial distributions and lattice constants reasonably well, potentials of type (ii) are required to obtain correct values for bond angles and the equation of state. The importance of polarizability for metal oxide properties decreases with increasing temperature.

cond-mat.mtrl-sci

Simulation of crack propagation in alumina with ab-initio based polarizable force field

We present an effective atomic interaction potential for crystalline alpha-Al2O3 generated by the program potfit. The Wolf direct, pairwise summation method with spherical truncation is used for electrostatic interactions. The polarizability of oxygen atoms is included by use of the Tangney-Scandolo interatomic force field approach. The potential is optimized to reproduce the forces, energies and stresses in relaxed and strained configurations as well as {0001}, {10-10} and {11-20} surfaces of Al2O3. Details of the force field generation are given, and its validation is demonstrated. We apply the developed potential to investigate crack propagation in alpha-Al2O3 single crystals.

cond-mat.mtrl-sci

Ab initio based polarizable force field generation and application to liquid silica and magnesia

We extend the program potfit, which generates effective atomic interaction potentials from ab initio data, to electrostatic interactions and induced dipoles. The potential parametrization algorithm uses the Wolf direct, pairwise summation method with spherical truncation. The polarizability of oxygen atoms is modeled with the Tangney-Scandolo interatomic force field approach. Due to the Wolf summation, the computational effort in simulation scales linearly in the number of particles, despite the presence of electrostatic interactions. Thus, this model allows to perform large-scale molecular dynamics simulations of metal oxides with realistic potentials. Details of the implementation are given, and the generation of potentials for SiO2 and MgO is demonstrated. The approach is validated by simulations of microstructural, thermodynamic and vibrational properties of liquid silica and magnesia.

cond-mat.mtrl-sci

Direct Wolf summation of a polarizable force field for silica

We extend the Wolf direct, pairwise r^(-1) summation method with spherical truncation to dipolar interactions in silica. The Tangney-Scandolo interatomic force field for silica takes regard of polarizable oxygen atoms whose dipole moments are determined by iteration to a self-consistent solution. With Wolf summation, the computational effort scales linearly in the system size and can easily be distributed among many processors, thus making large-scale simulations of dipoles possible. The details of the implementation are explained. The approach is validated by estimations of the error term and simulations of microstructural and thermodynamic properties of silica.

cond-mat.mtrl-sci

Archimedean-like colloidal tilings on substrates with decagonal and tetradecagonal symmetry

Two-dimensional colloidal suspensions subject to laser interference patterns with decagonal symmetry can form an Archimedean-like tiling phase where rows of squares and triangles order aperiodically along one direction [J. Mikhael et al., Nature 454, 501 (2008)]. In experiments as well as in Monte-Carlo and Brownian dynamics simulations, we identify a similar phase when the laser field possesses tetradecagonal symmetry. We characterize the structure of both Archimedean-like tilings in detail and point out how the tilings differ from each other. Furthermore, we also estimate specific particle densities where the Archimedean-like tiling phases occur. Finally, using Brownian dynamics simulations we demonstrate how phasonic distortions of the decagonal laser field influence the Archimedean-like tiling. In particular, the domain size of the tiling can be enlarged by phasonic drifts and constant gradients in the phasonic displacement. We demonstrate that the latter occurs when the interfering laser beams are not adjusted properly.

cond-mat.soft

Planar coincidences for N-fold symmetry

The coincidence problem for planar patterns with $N$-fold symmetry is considered. For the N-fold symmetric module with $N<46$, all isometries of the plane are classified that result in coincidences of finite index. This is done by reformulating the problem in terms of algebraic number fields and using prime factorization. The more complicated case $N \ge 46$ is briefly discussed and N=46 is described explicitly. The results of the coincidence problem also solve the problem of colour lattices in two dimensions and its natural generalization to colour modules.

math.MG

Quasicrystalline Order in Binary Dipolar Systems

Motivated by recent experimental findings, we investigate the possible occurrence and characteristics of quasicrystalline order in two-dimensional mixtures of point dipoles with two sorts of dipole moments. Despite the fact that the dipolar interaction potential does not exhibit an intrinsic length scale and cannot be tuned a priori to support the formation of quasicrystalline order, we find that configurations with long--range quasicrystallinity yield minima in the potential energy surface of the many particle system. These configurations emanate from an ideal or perturbed ideal decoration of a binary tiling by steepest descent relaxation. Ground state energy calculations of alternative ordered states and parallel tempering Monte-Carlo simulations reveal that the quasicrystalline configurations do not correspond to a thermodynamically stable state. On the other hand, steepest descent relaxations and conventional Monte-Carlo simulations suggest that they are rather robust against fluctuations. Local quasicrystalline order in the disordered equilibrium states can be strong.

cond-mat.other