skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: A new and inexpensive non-bit-for-bit solution reproducibility test based on time step convergence (TSC1.0)

Abstract

A test procedure is proposed for identifying numerically significant solution changes in evolution equations used in atmospheric models. The test issues a fail signal when any code modifications or computing environment changes lead to solution differences that exceed the known time step sensitivity of the reference model. Initial evidence is provided using the Community Atmosphere Model (CAM) version 5.3 that the proposed procedure can be used to distinguish rounding-level solution changes from impacts of compiler optimization or parameter perturbation, which are known to cause substantial differences in the simulated climate. The test is not exhaustive since it does not detect issues associated with diagnostic calculations that do not feedback to the model state variables. Nevertheless, it provides a practical and objective way to assess the significance of solution changes. The short simulation length implies low computational cost. The independence between ensemble members allows for parallel execution of all simulations, thus facilitating fast turnaround. The new method is simple to implement since it does not require any code modifications. We expect that the same methodology can be used for any geophysical model to which the concept of time step  convergence is applicable.

Authors:
 [1];  [1];  [1];  [1];  [1];  [2]
  1. Pacific Northwest National Lab. (PNNL), Richland, WA (United States). Atmospheric Sciences and Global Change Division
  2. National Center for Atmospheric Research, Boulder, CO (United States). Climate and Global Dynamics Laboratory
Publication Date:
Research Org.:
Pacific Northwest National Lab. (PNNL), Richland, WA (United States)
Sponsoring Org.:
USDOE Office of Science (SC), Biological and Environmental Research (BER) (SC-23)
OSTI Identifier:
1342455
Alternate Identifier(s):
OSTI ID: 1344047
Report Number(s):
PNNL-SA-118076
Journal ID: ISSN 1991-9603; KP1703020
Grant/Contract Number:
AC05-76RL01830; AC05-00OR22725
Resource Type:
Journal Article: Published Article
Journal Name:
Geoscientific Model Development (Online)
Additional Journal Information:
Journal Name: Geoscientific Model Development (Online); Journal Volume: 10; Journal Issue: 2; Journal ID: ISSN 1991-9603
Publisher:
European Geosciences Union
Country of Publication:
United States
Language:
English
Subject:
54 ENVIRONMENTAL SCIENCES; 58 GEOSCIENCES

Citation Formats

Wan, Hui, Zhang, Kai, Rasch, Philip J., Singh, Balwinder, Chen, Xingyuan, and Edwards, Jim. A new and inexpensive non-bit-for-bit solution reproducibility test based on time step convergence (TSC1.0). United States: N. p., 2017. Web. doi:10.5194/gmd-10-537-2017.
Wan, Hui, Zhang, Kai, Rasch, Philip J., Singh, Balwinder, Chen, Xingyuan, & Edwards, Jim. A new and inexpensive non-bit-for-bit solution reproducibility test based on time step convergence (TSC1.0). United States. doi:10.5194/gmd-10-537-2017.
Wan, Hui, Zhang, Kai, Rasch, Philip J., Singh, Balwinder, Chen, Xingyuan, and Edwards, Jim. Fri . "A new and inexpensive non-bit-for-bit solution reproducibility test based on time step convergence (TSC1.0)". United States. doi:10.5194/gmd-10-537-2017.
@article{osti_1342455,
title = {A new and inexpensive non-bit-for-bit solution reproducibility test based on time step convergence (TSC1.0)},
author = {Wan, Hui and Zhang, Kai and Rasch, Philip J. and Singh, Balwinder and Chen, Xingyuan and Edwards, Jim},
abstractNote = {A test procedure is proposed for identifying numerically significant solution changes in evolution equations used in atmospheric models. The test issues a fail signal when any code modifications or computing environment changes lead to solution differences that exceed the known time step sensitivity of the reference model. Initial evidence is provided using the Community Atmosphere Model (CAM) version 5.3 that the proposed procedure can be used to distinguish rounding-level solution changes from impacts of compiler optimization or parameter perturbation, which are known to cause substantial differences in the simulated climate. The test is not exhaustive since it does not detect issues associated with diagnostic calculations that do not feedback to the model state variables. Nevertheless, it provides a practical and objective way to assess the significance of solution changes. The short simulation length implies low computational cost. The independence between ensemble members allows for parallel execution of all simulations, thus facilitating fast turnaround. The new method is simple to implement since it does not require any code modifications. We expect that the same methodology can be used for any geophysical model to which the concept of time step  convergence is applicable.},
doi = {10.5194/gmd-10-537-2017},
journal = {Geoscientific Model Development (Online)},
number = 2,
volume = 10,
place = {United States},
year = {Fri Feb 03 00:00:00 EST 2017},
month = {Fri Feb 03 00:00:00 EST 2017}
}

Journal Article:
Free Publicly Available Full Text
Publisher's Version of Record at 10.5194/gmd-10-537-2017

Citation Metrics:
Cited by: 2works
Citation information provided by
Web of Science

Save / Share:
  • A test procedure is proposed for identifying numerically significant solution changes in evolution equations used in atmospheric models. The test issues a fail signal when any code modifications or computing environment changes lead to solution differences that exceed the known time step sensitivity of the reference model. Initial evidence is provided using the Community Atmosphere Model (CAM) version 5.3 that the proposed procedure can be used to distinguish rounding-level solution changes from impacts of compiler optimization or parameter perturbation, which are known to cause substantial differences in the simulated climate. The test is not exhaustive since it does not detect issues associatedmore » with diagnostic calculations that do not feedback to the model state variables. Nevertheless, it provides a practical and objective way to assess the significance of solution changes. The short simulation length implies low computational cost. The independence between ensemble members allows for parallel execution of all simulations, thus facilitating fast turnaround. The new method is simple to implement since it does not require any code modifications. We expect that the same methodology can be used for any geophysical model to which the concept of time step  convergence is applicable.« less
  • An improvement in reproducibility of the critical current density (J{sub c}) has been obtained in Ag-sheathed Bi2223 tapes made by the powder-in-tube method using a non-linear temperature-time sintering process. In this process, the sintering temperature was slowly raised from the temperature of emergence of a liquid phase (about 800{degrees}C) to the temperature of sufficient crystal growth of 2223 phase (830{degrees}C), so that the rate of liquid formation could be controlled and rapid growth of the non-superconducting phases could be prevented. A reduction of about 30 % per pass during rolling, one intermediate pressing and total sintering time less than 120more » hours were employed in this process. The texture of the tape made by this process is better than that of the tape made by isothermal sintering. An average J{sub c} of 3.18 x 10{sup 4} A/cm{sup 2} at 77 K in self-field with a deviation 17 % was measured in tapes made by this method, compared with an average J{sub c} of 2.09 x 10{sup 4} A/cm{sup 2} at 77 K with a deviation up to 80 % in tapes made by a normal isothermal sintering. Doping pure silver powder into incompletely reacted powder is used to control transformation from 2212 to 2223 and enhance the texture. The J{sub c}-field dependence was improved by increasing number of pressing. The fine dispersed phase of Ca-Sr-Pb-O was observed by EDX analysis in tapes made by the non-linear sintering, while the Sr-Ca-Cu-O phase in tapes made by isothermal sintering was commonly found.« less
  • A facile new method for the synthesis of nanostructured Cu 2O–ZnO and Cu 2O–ZnO–MgO catalysts has been described.
  • Constant step size Magnus propagators are used to integrate the Secrest--Johnson vibrational excitation problem with parameters given by Stechel, Walker, and Light. Calculations are done for a six channel basis for E = 6 and 8 and for a 30 channel basis for E = 60. For all the calculations the error scales as the fourth power of the step size for small steps. This indicates a significant cancellation between the errors in the Magnus propagation and the assumed diagonalization of the potential over each interval. The error grows rapidly when the step size is such that at least onemore » channel is propagated by near half a wavelength. The errors in the E = 6 calculation are compared with those for the log derivative, renormalized Numerov, R matrix propagation methods, and a quadratic approximate potential method. The constant step size Magnus method is superior to the other methods for this vibrational excitation problem. A perturbation analysis is presented to show why accurate calculations are possible with large nonclassical steps, since we have used steps much larger than those usually recommended for use with the Magnus propagator. Finally, a perturbative calculation of the transform that diagonlizes the potential matrix is described. The perturbative transform is rapidly calculated, and gives excellent results for diagonalizations at large distances.« less
  • Lasers are widely used in industry due to their precision of operation and attainment of considerable end product quality. In order to improve the laser machining process, investigation into laser heating mechanism becomes essential. The present study covers the analytical formulation of laser pulse heating process for time exponential decaying and intensity step input pulses. The convergence of both solution is introduced in relation to pulse parameters. The analytical expression for the convergence is omitted due to complex nature of the mathematics involved, however, computational method is attempted to identify the pulse parameter for which both results converge. It ismore » found that the surface temperature profiles obtained for both pulses merge when {beta} approaches to zero.« less