Single-Precision Calculation of Iterative Refinement of Eigenpairs of a Real Symmetric-Definite Generalized Eigenproblem by Using a Filter Composed of a Single Resolvent

2021 ◽  
Author(s):  
Hiroshi Murakami
Keyword(s):  
Iterative Refinement ◽  
Single Precision ◽  
Generalized Eigenproblem

Double Precision Is Not Needed for Many-Body Calculations: New Conventional Wisdom

2018 ◽  
Author(s):  
Pavel Pokhilko ◽  
Evgeny Epifanovsky ◽  
Anna I. Krylov
Keyword(s):  
Large Scale ◽  
Computation Time ◽  
Coupled Cluster ◽  
Double Precision ◽  
Many Body ◽  
Single Precision ◽  
Parallel Performance ◽  
Point Representation ◽  
Electron Repulsion Integrals ◽  
Cluster Methods

Using single precision floating point representation reduces the size of data and computation time by a factor of two relative to double precision conventionally used in electronic structure programs. For large-scale calculations, such as those encountered in many-body theories, reduced memory footprint alleviates memory and input/output bottlenecks. Reduced size of data can lead to additional gains due to improved parallel performance on CPUs and various accelerators. However, using single precision can potentially reduce the accuracy of computed observables. Here we report an implementation of coupled-cluster and equation-of-motion coupled-cluster methods with single and double excitations in single precision. We consider both standard implementation and one using Cholesky decomposition or resolution-of-the-identity of electron-repulsion integrals. Numerical tests illustrate that when single precision is used in correlated calculations, the loss of accuracy is insignificant and pure single-precision implementation can be used for computing energies, analytic gradients, excited states, and molecular properties. In addition to pure single-precision calculations, our implementation allows one to follow a single-precision calculation by clean-up iterations, fully recovering double-precision results while retaining significant savings.

scholarly journals RefineLoc: Iterative Refinement for Weakly-Supervised Action Localization

2021 ◽  
Author(s):  
Alejandro Pardo ◽  
Humam Alwassel ◽  
Fabian Caba Heilbron ◽  
Ali Thabet ◽  
Bernard Ghanem
Keyword(s):  
Iterative Refinement ◽  
Action Localization ◽  
Weakly Supervised

The VEOS Project

1994 ◽  
Vol 3 (2) ◽  
pp. 111-129 ◽  
Author(s):  
William Bricken ◽  
Geoffrey Coco
Keyword(s):  
Rapid Prototyping ◽  
Virtual Environments ◽  
Virtual Environment ◽  
Object Oriented ◽  
Organization Structure ◽  
Object Oriented Programming ◽  
Iterative Refinement ◽  
Software Infrastructure ◽  
Human Interface ◽  
Heterogeneous Distributed Computing

The Virtual Environment Operating Shell (veos) was developed at University of Washington's Human Interface Technology Laboratory as software infrastructure for the lab's research in virtual environments. veos was designed from scratch to provide a comprehensive and unified management facility to support generation of, interaction with, and maintenance of virtual environments. VEOS emphasizes rapid prototyping, heterogeneous distributed computing, and portability. We discuss the design, philosophy and implementation of veos in depth. Within the Kernel, the shared database transformations are pattern-directed, communications are asynchronous, and the programmer's interface is LISP. An entity-based metaphor extends object-oriented programming to systems-oriented programming. Entities provide first-class environments and biological programming constructs such as perceive, react, and persist. The organization, structure, and programming of entities are discussed in detail. The article concludes with a description of the applications that have contributed to the iterative refinement of the VEOS software.

Sign in / Sign up

Export Citation Format

Share Document