Physics

In the past couple of years, there is a proliferation in the use of machine learning approaches to represent subgrid scale processes in geophysical flows with an aim to improve the forecasting capability and to accelerate numerical simulations of these flows. Despite its success for different types of flow, the online deployment of a data-driven closure model can cause instabilities and biases in modeling the overall effect of subgrid scale processes, which in turn leads to inaccurate prediction. To tackle this issue, we exploit the data assimilation technique to correct the physics-based model coupled with the neural network as a surrogate for unresolved flow dynamics in multiscale systems. In particular, we use a set of neural network architectures to learn the correlation between resolved flow variables and the parameterizations of unresolved flow dynamics and formulate a data assimilation approach to correct the hybrid model during their online deployment. We illustrate our framework in a set of applications of the multiscale Lorenz 96 system for which the parameterization model for unresolved scales is exactly known, and the two-dimensional Kraichnan turbulence system for which the parameterization model for unresolved scales is not known a priori. Our analysis, therefore, comprises a predictive dynamical core empowered by (i) a data-driven closure model for subgrid scale processes, (ii) a data assimilation approach for forecast error correction, and (iii) both data-driven closure and data assimilation procedures. We show significant improvement in the long-term prediction of the underlying chaotic dynamics with our framework compared to using only neural network parameterizations for future prediction.

We present a method for accelerating discrete ordinates radiative transfer calculations for radiative transfer. Our method works with nonlinear positivity fixes, in contrast to most acceleration schemes. The method is based on the dynamic mode decomposition (DMD) and using a sequence of rank-one updates to compute the singular value decomposition needed for DMD. Using a sequential method allows us to automatically determine the number of solution vectors to include in the DMD acceleration. We present results for slab geometry discrete ordinates calculations with the standard temperature linearization. Compared with positive source iteration, our results demonstrate that our acceleration method reduces the number of transport sweeps required to solve the problem by a factor of about 3 on a standard diffusive Marshak wave problem, a factor of several thousand on a cooling problem where the effective scattering ratio approaches unity, and a factor of 20 improvement in a realistic, multimaterial radiating shock problem.

This work presents a data-driven magnetostatic finite-element solver that is specifically well-suited to cope with strongly nonlinear material responses. The data-driven computing framework is essentially a multiobjective optimization procedure matching the material operation points as closely as possible to given material data while obeying Maxwell's equations. Here, the framework is extended with heterogeneous (local) weighting factors - one per finite element - equilibrating the goal function locally according to the material behavior. This modification allows the data-driven solver to cope with unbalanced measurement data sets, i.e. data sets suffering from unbalanced space filling. This occurs particularly in the case of strongly nonlinear materials, which constitute problematic cases that hinder the efficiency and accuracy of standard data-driven solvers with a homogeneous (global) weighting factor. The local weighting factors are embedded in the distance-minimizing data-driven algorithm used for noiseless data, likewise for the maximum entropy data-driven algorithm used for noisy data. Numerical experiments based on a quadrupole magnet model with a soft magnetic material show that the proposed modification results in major improvements in terms of solution accuracy and solver efficiency. For the case of noiseless data, local weighting factors improve the convergence of the data-driven solver by orders of magnitude. When noisy data are considered, the convergence rate of the data-driven solver is doubled.

Industrial dynamical systems often exhibit multi-scale response due to material heterogeneities, operation conditions and complex environmental loadings. In such problems, it is the case that the smallest length-scale of the systems dynamics controls the numerical resolution required to effectively resolve the embedded physics. In practice however, high numerical resolutions is only required in a confined region of the system where fast dynamics or localized material variability are exhibited, whereas a coarser discretization can be sufficient in the rest majority of the system. To this end, a unified computational scheme with uniform spatio-temporal resolutions for uncertainty quantification can be very computationally demanding. Partitioning the complex dynamical system into smaller easier-to-solve problems based of the localized dynamics and material variability can reduce the overall computational cost. However, identifying the region of interest for high-resolution and intensive uncertainty quantification can be a problem dependent. The region of interest can be specified based on the localization features of the solution, user interest, and correlation length of the random material properties. For problems where a region of interest is not evident, Bayesian inference can provide a feasible solution. In this work, we employ a Bayesian framework to update our prior knowledge on the localized region of interest using measurements and system response. To address the computational cost of the Bayesian inference, we construct a Gaussian process surrogate for the forward model. Once, the localized region of interest is identified, we use polynomial chaos expansion to propagate the localization uncertainty. We demonstrate our framework through numerical experiments on a three-dimensional elastodynamic problem.

High-fidelity modeling of blood flow is crucial for enhancing our understanding of cardiovascular disease. Despite significant advances in computational and experimental characterization of blood flow, the knowledge that we can acquire from such investigations remains limited by the presence of uncertainty in parameters, low spatiotemporal resolution, and measurement noise. Additionally, extracting useful information from these datasets is challenging. Data-driven modeling techniques have the potential to overcome these challenges and transform cardiovascular flow modeling. In this paper, we review several data-driven modeling techniques, highlight the common ideas and principles that emerge across numerous such techniques, and provide illustrative examples of how they could be used in the context of cardiovascular fluid mechanics. In particular, we discuss principal component analysis (PCA), robust PCA, compressed sensing, the Kalman filter for data assimilation, low-rank data recovery, and several additional methods for reduced-order modeling for cardiovascular flows, including the dynamic mode decomposition (DMD), and the sparse identification of nonlinear dynamics (SINDy). All of these techniques are presented in the context of cardiovascular flows with simple examples. These data-driven modeling techniques have the potential to transform computational and experimental cardiovascular flow research, and we discuss challenges and opportunities in applying these techniques in the field, looking ultimately towards data-driven patient-specific blood flow modeling.

In this paper, we propose a sensitivity-free and multi-objective structural design methodology called data-driven topology design. It is schemed to obtain high-performance material distributions from initially given material distributions in a given design domain. Its basic idea is to iterate the following processes: (i) selecting material distributions from a dataset of material distributions according to eliteness, (ii) generating new material distributions using a deep generative model trained with the selected elite material distributions, and (iii) merging the generated material distributions with the dataset. Because of the nature of a deep generative model, the generated material distributions are diverse and inherit features of the training data, that is, the elite material distributions. Therefore, it is expected that some of the generated material distributions are superior to the current elite material distributions, and by merging the generated material distributions with the dataset, the performances of the newly selected elite material distributions are improved. The performances are further improved by iterating the above processes. The usefulness of data-driven topology design is demonstrated through numerical examples.

We present two new developments for computing excited state energies within the GW approximation. First, calculations of the Green's function and the screened Coulomb interaction are decomposed into two parts: one is deterministic while the other relies on stochastic sampling. Second, this separation allows constructing a subspace self-energy, which contains dynamic correlation from only a particular (spatial or energetic) region of interest. The methodology is exemplified on large-scale simulations of nitrogen-vacancy states in a periodic hBN monolayer and hBN-graphene heterostructure. We demonstrate that the deterministic embedding of strongly localized states significantly reduces statistical errors, and the computational cost decreases by more than an order of magnitude. The computed subspace self-energy unveils how interfacial couplings affect electronic correlations and identifies contributions to excited-state lifetimes. While the embedding is necessary for the proper treatment of impurity states, the decomposition yields new physical insight into quantum phenomena in heterogeneous systems.

We propose a general machine learning-based framework for building an accurate and widely-applicable energy functional within the framework of generalized Kohn-Sham density functional theory. To this end, we develop a way of training self-consistent models that are capable of taking large datasets from different systems and different kinds of labels. We demonstrate that the functional that results from this training procedure gives chemically accurate predictions on energy, force, dipole, and electron density for a large class of molecules. It can be continuously improved when more and more data are available.

Near-term prediction of the structured spatio-temporal processes driving our climate is of profound importance to the safety and well-being of millions, but the prounced nonlinear convection of these processes make a complete mechanistic description even of the short-term dynamics challenging. However, convective transport provides not only a principled physical description of the problem, but is also indicative of the transport in time of informative features which has lead to the recent successful development of ``physics free'' approaches to the now-casting problem. In this work we demonstrate that their remains an important role to be played by physically informed models, which can successfully leverage deep learning (DL) to project the process onto a lower dimensional space on which a minimal dynamical description holds. Our approach synthesises the feature extraction capabilities of DL with physically motivated dynamics to outperform existing model free approaches, as well as state of the art hybrid approaches, on complex real world datasets including sea surface temperature and precipitation.

It has been a challenge to accurately simulate Li-ion diffusion processes in battery materials at room temperature using {\it ab initio} molecular dynamics (AIMD) due to its high computational cost. This situation has changed drastically in recent years due to the advances in machine learning-based interatomic potentials. Here we implement the Deep Potential Generator scheme to \textit{automatically} generate interatomic potentials for LiGePS-type solid-state electrolyte materials. This increases our ability to simulate such materials by several orders of magnitude without sacrificing {\it ab initio} accuracy. Important technical aspects like the statistical error and size effects are carefully investigated. We further establish a reliable protocol for accurate computation of Li-ion diffusion processes at experimental conditions, by investigating important technical aspects like the statistical error and size effects. Such a protocol and the automated workflow allow us to screen materials for their relevant properties with much-improved efficiency. By using the protocol and automated workflow developed here, we obtain the diffusivity data and activation energies of Li-ion diffusion that agree well with the experiment. Our work paves the way for future investigation of Li-ion diffusion mechanisms and optimization of Li-ion conductivity of solid-state electrolyte materials.