Physics-Informed Neural Networks (PINNs) have emerged as a highly active research topic across multiple disciplines in science and engineering, including computational geomechanics. PINNs offer a promising approach in different applications where faster, near real-time or real-time numerical prediction is required. Examples of such areas in geomechanics include geotechnical design optimization, digital twins of geo-structures and stability prediction of monitored slopes. But there remain challenges in training of PINNs, especially for problems with high spatial and temporal complexity. In this paper, we study how the training of PINNs can be improved by using an ideal-ized poroelasticity problem as a demonstration example. A curriculum training strat-egy is employed where the PINN model is trained gradually by dividing the training data into intervals along the temporal dimension. We find that the PINN model with curriculum training takes nearly half the time required for training compared to con-ventional training over the whole solution domain. For the particular example here, the quality of the predicted solution was found to be good in both training approach-es, but it is anticipated that the curriculum training approach has the potential to offer a better prediction capability for more complex problems, a subject for further research.
相關內容
Networking：IFIP International Conferences on Networking。
Explanation：國際(ji)網絡會議。
Publisher：IFIP。
SIT：
Recurrent neural networks (RNNs) notoriously struggle to learn long-term memories, primarily due to vanishing and exploding gradients. The recent success of state-space models (SSMs), a subclass of RNNs, to overcome such difficulties challenges our theoretical understanding. In this paper, we delve into the optimization challenges of RNNs and discover that, as the memory of a network increases, changes in its parameters result in increasingly large output variations, making gradient-based learning highly sensitive, even without exploding gradients. Our analysis further reveals the importance of the element-wise recurrence design pattern combined with careful parametrizations in mitigating this effect. This feature is present in SSMs, as well as in other architectures, such as LSTMs. Overall, our insights provide a new explanation for some of the difficulties in gradient-based learning of RNNs and why some architectures perform better than others.
This paper presents a novel approach for constructing graph neural networks equivariant to 2D rotations and translations and leveraging them as PDE surrogates on non-gridded domains. We show that aligning the representations with the principal axis allows us to sidestep many constraints while preserving SE(2) equivariance. By applying our model as a surrogate for fluid flow simulations and conducting thorough benchmarks against non-equivariant models, we demonstrate significant gains in terms of both data efficiency and accuracy.
Robotic exploration has long captivated researchers aiming to map complex environments efficiently. Techniques such as potential fields and frontier exploration have traditionally been employed in this pursuit, primarily focusing on solitary agents. Recent advancements have shifted towards optimizing exploration efficiency through multiagent systems. However, many existing approaches overlook critical real-world factors, such as broadcast range limitations, communication costs, and coverage overlap. This paper addresses these gaps by proposing a distributed maze exploration strategy (CU-LVP) that assumes constrained broadcast ranges and utilizes Voronoi diagrams for better area partitioning. By adapting traditional multiagent methods to distributed environments with limited broadcast ranges, this study evaluates their performance across diverse maze topologies, demonstrating the efficacy and practical applicability of the proposed method. The code and experimental results supporting this study are available in the following repository: //github.com/manouslinard/multiagent-exploration/.
Predictive posterior densities (PPDs) are of interest in approximate Bayesian inference. Typically, these are estimated by simple Monte Carlo (MC) averages using samples from the approximate posterior. We observe that the signal-to-noise ratio (SNR) of such estimators can be extremely low. An analysis for exact inference reveals SNR decays exponentially as there is an increase in (a) the mismatch between training and test data, (b) the dimensionality of the latent space, or (c) the size of the test data relative to the training data. Further analysis extends these results to approximate inference. To remedy the low SNR problem, we propose replacing simple MC sampling with importance sampling using a proposal distribution optimized at test time on a variational proxy for the SNR and demonstrate that this yields greatly improved estimates.
It is increasingly recognized that participation bias can pose problems for genetic studies. Recently, to overcome the challenge that genetic information of non-participants is unavailable, it is shown that by comparing the IBD (identity by descent) shared and not-shared segments among the participants, one can estimate the genetic component underlying participation. That, however, does not directly address how to adjust estimates of heritability and genetic correlation for phenotypes correlated with participation. Here, for phenotypes whose mean differences between population and sample are known, we demonstrate a way to do so by adopting a statistical framework that separates out the genetic and non-genetic correlations between participation and these phenotypes. Crucially, our method avoids making the assumption that the effect of the genetic component underlying participation is manifested entirely through these other phenotypes. Applying the method to 12 UK Biobank phenotypes, we found 8 have significant genetic correlations with participation, including body mass index, educational attainment, and smoking status. For most of these phenotypes, without adjustments, estimates of heritability and the absolute value of genetic correlation would have underestimation biases.
This paper aims to investigate the diffusion behavior of particles moving in stochastic flows under a structure-preserving scheme. We compute the effective diffusivity for normal diffusive random flows and establish the power law between spatial and temporal variables for cases with anomalous diffusion phenomena. From a Lagrangian approach, we separate the corresponding stochastic differential equations (SDEs) into sub-problems and construct a one-step structure-preserving method to solve them. Then by modified equation systems, the convergence analysis in calculating the effective diffusivity is provided and compared between the structure-preserving scheme and the Euler-Maruyama scheme. Also, we provide the error estimate for the structure-preserving scheme in calculating the power law for a series of super-diffusive random flows. Finally, we calculate the effective diffusivity and anomalous diffusion phenomena for a series of 2D and 3D random fields.
Reinforcement learning is commonly concerned with problems of maximizing accumulated rewards in Markov decision processes. Oftentimes, a certain goal state or a subset of the state space attain maximal reward. In such a case, the environment may be considered solved when the goal is reached. Whereas numerous techniques, learning or non-learning based, exist for solving environments, doing so optimally is the biggest challenge. Say, one may choose a reward rate which penalizes the action effort. Reinforcement learning is currently among the most actively developed frameworks for solving environments optimally by virtue of maximizing accumulated reward, in other words, returns. Yet, tuning agents is a notoriously hard task as reported in a series of works. Our aim here is to help the agent learn a near-optimal policy efficiently while ensuring a goal reaching property of some basis policy that merely solves the environment. We suggest an algorithm, which is fairly flexible, and can be used to augment practically any agent as long as it comprises of a critic. A formal proof of a goal reaching property is provided. Simulation experiments on six problems under five agents, including the benchmarked one, provided an empirical evidence that the learning can indeed be boosted while ensuring goal reaching property.
This work is related to developing entropy-stable positivity-preserving Discontinuous Galerkin (DG) methods as a computational scheme for Boltzmann-Poisson systems modeling the probability density of collisional electronic transport along semiconductor energy bands. In momentum coordinates representing spherical / energy-angular variables, we pose the respective Vlasov-Boltzmann equation with a linear collision operator and a singular measure, modeling scatterings as functions of the band structure appropriately for hot electron nanoscale transport. We show stability results of semi-discrete DG schemes under an entropy norm for 1D-position (2D-momentum) and 2D-position (3D-momentum), using dissipative properties of the collisional operator given its entropy inequality. The latter depends on an exponential of the Hamiltonian rather than the Maxwellian associated with only kinetic energy. For the 1D problem, knowing the analytic solution to the Poisson equation and convergence to a constant current is crucial to obtaining full stability (weighted entropy norm decreasing over time). For the 2D problem, specular reflection boundary conditions and periodicity are considered in estimating stability under an entropy norm. Regarding the positivity-preservation proofs in the DG scheme for the 1D problem, inspired by \cite{ZhangShu1}, \cite{ZhangShu2}, and \cite{CGP}, \cite{EECHXM-JCP}, we treat collisions as a source and find convex combinations of the transport and collision terms which guarantee positivity of the cell average of our numerical probability density at the next time. The positivity of the numerical solution to the probability density in the domain is guaranteed by applying the limiters in \cite{ZhangShu1} and \cite{ZhangShu2} that preserve the cell average modifying the slope of the piecewise linear solutions to make the function non-negative.
We consider the problem of sampling a high dimensional multimodal target probability measure. We assume that a good proposal kernel to move only a subset of the degrees of freedoms (also known as collective variables) is known a priori. This proposal kernel can for example be built using normalizing flows. We show how to extend the move from the collective variable space to the full space and how to implement an accept-reject step in order to get a reversible chain with respect to a target probability measure. The accept-reject step does not require to know the marginal of the original measure in the collective variable (namely to know the free energy). The obtained algorithm admits several variants, some of them being very close to methods which have been proposed previously in the literature. We show how the obtained acceptance ratio can be expressed in terms of the work which appears in the Jarzynski-Crooks equality, at least for some variants. Numerical illustrations demonstrate the efficiency of the approach on various simple test cases, and allow us to compare the variants of the algorithm.
Graph representation learning for hypergraphs can be used to extract patterns among higher-order interactions that are critically important in many real world problems. Current approaches designed for hypergraphs, however, are unable to handle different types of hypergraphs and are typically not generic for various learning tasks. Indeed, models that can predict variable-sized heterogeneous hyperedges have not been available. Here we develop a new self-attention based graph neural network called Hyper-SAGNN applicable to homogeneous and heterogeneous hypergraphs with variable hyperedge sizes. We perform extensive evaluations on multiple datasets, including four benchmark network datasets and two single-cell Hi-C datasets in genomics. We demonstrate that Hyper-SAGNN significantly outperforms the state-of-the-art methods on traditional tasks while also achieving great performance on a new task called outsider identification. Hyper-SAGNN will be useful for graph representation learning to uncover complex higher-order interactions in different applications.
小貼士
登錄享
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191