Corpus ID: 215238534

Deep-learning enhancement of large scale numerical simulations

  title={Deep-learning enhancement of large scale numerical simulations},
  author={C. N. Leeuwen and Damian Podareanu and V. Codreanu and M. Cai and Axel Berg and S. Zwart and Robin Stoffer and M. Veerman and C. V. Heerwaarden and S. Otten and S. Caron and C. Geng and F. Ambrosetti and A. Bonvin},
Traditional simulations on High-Performance Computing (HPC) systems typically involve modeling very large domains and/or very complex equations. HPC systems allow running large models, but limits in performance increase that have become more prominent in the last 5-10 years will likely be experienced. Therefore new approaches are needed to increase application performance. Deep learning appears to be a promising way to achieve this. Recently deep learning has been employed to enhance solving… Expand


A Deep Hybrid Model for Weather Forecasting
This work studies specifically the power of making predictions via a hybrid approach that combines discriminatively trained predictive models with a deep neural network that models the joint statistics of a set of weather-related variables. Expand
Balancing Accuracy, Efficiency, and Flexibility in Radiation Calculations for Dynamical Models
The initial implementation of a new toolbox that seeks to balance accuracy, efficiency, and flexibility in radiation calculations for dynamical models is described, with a focus on accuracy, by basing the k‐distribution on state‐of‐the‐art spectroscopy and on the sometimes‐conflicting goals of flexibility and efficiency. Expand
Large eddy simulation using the general circulation model ICON
An extension of ICON is presented that enables it to perform as a large eddy simulation (LES) model, and the details of the implementation of the LES turbulence scheme in ICON are explained and test cases are performed to validate it against two standard LES models. Expand
Strategies for training large scale neural network language models
This work describes how to effectively train neural network based language models on large data sets and introduces hash-based implementation of a maximum entropy model, that can be trained as a part of the neural network model. Expand
Image-to-Image Translation with Conditional Adversarial Networks
Conditional adversarial networks are investigated as a general-purpose solution to image-to-image translation problems and it is demonstrated that this approach is effective at synthesizing photos from label maps, reconstructing objects from edge maps, and colorizing images, among other tasks. Expand
Deep Neural Networks for Acoustic Modeling in Speech Recognition: The Shared Views of Four Research Groups
This article provides an overview of progress and represents the shared views of four research groups that have had recent successes in using DNNs for acoustic modeling in speech recognition. Expand
Inference of macromolecular assemblies from crystalline state.
A new method, based on chemical thermodynamics, is developed for automatic detection of macromolecular assemblies in the Protein Data Bank (PDB) entries that are the results of X-ray diffraction experiments, as found, biological units may be recovered at 80-90% success rate, which makesX-ray crystallography an important source of experimental data on macromolescular complexes and protein-protein interactions. Expand
An analysis of oligomerization interfaces in transmembrane proteins
The first fully comprehensive set of validated transmembrane protein interfaces is compiled, finding that the putative dimer interfaces proposed for class A GPCRs do not show the usual patterns of stable biological interfaces, thus they likely correspond to crystal interfaces. Expand
Protein interface classification by evolutionary analysis
A protein-protein interface classifier that relies on evolutionary data to detect the biological character of interfaces and it is demonstrated that a more conservative selection of homolog sequences - with relatively high sequence identities to the query - is able to produce a clearer signal than previous attempts. Expand
Automated evaluation of quaternary structures from protein crystals
The two methods are complementary and could be combined to improve biological assembly assignments and achieve 85% precision on a new dataset of 1,481 biological assemblies with consensus of PDB annotations. Expand