Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userScalable and modular material point method for large-scale simulations
236
0
0.0
(
0
)
Ask ChatGPT about the research
No Arabic abstract
In this paper, we describe a new scalable and modular material point method (MPM) code developed for solving large-scale problems in continuum mechanics. The MPM is a hybrid Eulerian-Lagrangian approach, which uses both moving material points and computational nodes on a background mesh. The MPM has been successfully applied to solve large-deformation problems such as landslides, failure of slopes, concrete flows, etc. Solving these large-deformation problems result in the material points actively moving through the mesh. Developing an efficient parallelisation scheme for the MPM code requires dynamic load-balancing techniques for both the material points and the background mesh. This paper describes the data structures and algorithms employed to improve the performance and portability of the MPM code. An object-oriented programming paradigm is adopted to modularise the MPM code. The Unified Modelling Language (UML) diagram of the MPM code structure is shown in Figure 1.
rate research
Read More
Visualizing regional-scale landslides is the key to conveying the threat of natural hazards to stakeholders and policymakers. Traditional visualization techniques are restricted to post-processing a limited subset of simulation data and are not scalable to rendering exascale models with billions of particles. In-situ visualization is a technique of rendering simulation data in real-time, i.e., rendering visuals in tandem while the simulation is running. In this study, we develop a scalable N:M interface architecture to visualize regional-scale landslides. We demonstrate the scalability of the architecture by simulating the long runout of the 2014 Oso landslide using the Material Point Method coupled with the Galaxy ray tracing engine rendering 4.2 million material points as spheres. In-situ visualization has an amortized runtime increase of 2% compared to non-visualized simulations. The developed approach can achieve in-situ visualization of regional-scale landslides with billions of particles with minimal impact on the simulation process.
Full-batch training on Graph Neural Networks (GNN) to learn the structure of large graphs is a critical problem that needs to scale to hundreds of compute nodes to be feasible. It is challenging due to large memory capacity and bandwidth requirements on a single compute node and high communication volumes across multiple nodes. In this paper, we present DistGNN that optimizes the well-known Deep Graph Library (DGL) for full-batch training on CPU clusters via an efficient shared memory implementation, communication reduction using a minimum vertex-cut graph partitioning algorithm and communication avoidance using a family of delayed-update algorithms. Our results on four common GNN benchmark datasets: Reddit, OGB-Products, OGB-Papers and Proteins, show up to 3.7x speed-up using a single CPU socket and up to 97x speed-up using 128 CPU sockets, respectively, over baseline DGL implementations running on a single CPU socket
A new thixotropic model is developed integrating the Papanastasiou-Bingham model with thixotropy equations to simulate the flow behaviour of Tremie Concrete in the Material Point Method framework. The effect of thixotropy on the rheological behaviour of fresh concrete is investigated by comparing field measurements with numerical simulations. The comparison yields new insights into a critical and often overlooked behaviour of concrete. A parametric study is performed to understand the effect of model parameters and rest-time on the shear stress response of fresh concrete. The Material Point Method with the Papanastasiou-Bingham model reproduces slump-flow measurements observed in the field. The novel model revealed a decline in concrete workability during the Slump-flow test after a period of rest due to thixotropy, which the physical version of the test fails to capture. This reduction in workability significantly affects the flow behaviour and the effective use of fresh concrete in construction operation.
We present an Extended Kalman Filter framework for system identification and control of a stochastic high-dimensional epidemic model. The scale and severity of the COVID-19 emergency have highlighted the need for accurate forecasts of the state of the pandemic at a high resolution. Mechanistic compartmental models are widely used to produce such forecasts and assist in the design of control and relief policies. Unfortunately, the scale and stochastic nature of many of these models often makes the estimation of their parameters difficult. With the goal of calibrating a high dimensional COVID-19 model using low-level mobility data, we introduce a method for tractable maximum likelihood estimation that combines tools from Bayesian inference with scalable optimization techniques from machine learning. The proposed approach uses automatic backward-differentiation to directly compute the gradient of the likelihood of COVID-19 incidence and death data. The likelihood of the observations is estimated recursively using an Extended Kalman Filter and can be easily optimized using gradient-based methods to compute maximum likelihood estimators. Our compartmental model is trained using GPS mobility data that measures the mobility patterns of millions of mobile phones across the United States. We show that, after calibrating against incidence and deaths data from the city of Philadelphia, our model is able to produce an accurate 30-day forecast of the evolution of the pandemic.
In the field of database deduplication, the goal is to find approximately matching records within a database. Blocking is a typical stage in this process that involves cheaply finding candidate pairs of records that are potential matches for further processing. We present here Hashed Dynamic Blocking, a new approach to blocking designed to address datasets larger than those studied in most prior work. Hashed Dynamic Blocking (HDB) extends Dynamic Blocking, which leverages the insight that rare matching values and rare intersections of values are predictive of a matching relationship. We also present a novel use of Locality Sensitive Hashing (LSH) to build blocking key values for huge databases with a convenient configuration to control the trade-off between precision and recall. HDB achieves massive scale by minimizing data movement, using compact block representation, and greedily pruning ineffective candidate blocks using a Count-min Sketch approximate counting data structure. We benchmark the algorithm by focusing on real-world datasets in excess of one million rows, demonstrating that the algorithm displays linear time complexity scaling in this range. Furthermore, we execute HDB on a 530 million row industrial dataset, detecting 68 billion candidate pairs in less than three hours at a cost of $307 on a major cloud service.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
