Volume 5, Issue 1
Uncertainty Management in Scientific Numerical Computation

Shigeo Kawata, Takashi Ishihara, Daisuke Barada, J

DOI:

Int. J. Numer. Anal. Mod. B, 5 (2014), pp. 147-155

Published online: 2014-05

Preview Full PDF 350 1000
Export citation
  • Abstract

The uncertainty in scientific computing has induced serious accidents and disasters. New proposals are presented toward an uncertainty inference or management in scientific numerical computing. The first important point is to share the knowledge on uncertainty with other researchers. Sharing known uncertainties contributes to the uncertainty management. In addition, for specific problems, one can prepare multiple programs with different rounding methods and/or with different precisions, and can compare the results among the programs generated for the specific problems. If differences appear among the results, it suggests that the specific program may have some uncertainty problems, and has a lack of the precision or a lack of the digit number. This second method provides a simple and effective inference method of the uncertainty. Uncertainty comes from various sources from physical and mathematical model errors, unknown input data, numerical model errors, insuficient numerical precision, oating point precision, programming errors, data processing errors, visualization errors, etc., as well as human errors. Another uncertainty comes from the discretization step size of Δt or Δx in numerical computations. The discretization step size of Δt or Δx must be selected appropriately in numerical computations in order to describe short waves or fast phenomena concerned to the target problems. This uncertainty would be also reduced by multiple program computations with the different size of Δt and Δx to find out the appropriate step size under keeping numerical stabilities. These uncertainty reduction mechanisms are proposed in this paper.

  • Keywords

  • AMS Subject Headings

  • Copyright

COPYRIGHT: © Global Science Press

  • Email address
  • BibTex
  • RIS
  • TXT
@Article{IJNAMB-5-147, author = {Shigeo Kawata, Takashi Ishihara, Daisuke Barada, J}, title = {Uncertainty Management in Scientific Numerical Computation}, journal = {International Journal of Numerical Analysis Modeling Series B}, year = {2014}, volume = {5}, number = {1}, pages = {147--155}, abstract = {The uncertainty in scientific computing has induced serious accidents and disasters. New proposals are presented toward an uncertainty inference or management in scientific numerical computing. The first important point is to share the knowledge on uncertainty with other researchers. Sharing known uncertainties contributes to the uncertainty management. In addition, for specific problems, one can prepare multiple programs with different rounding methods and/or with different precisions, and can compare the results among the programs generated for the specific problems. If differences appear among the results, it suggests that the specific program may have some uncertainty problems, and has a lack of the precision or a lack of the digit number. This second method provides a simple and effective inference method of the uncertainty. Uncertainty comes from various sources from physical and mathematical model errors, unknown input data, numerical model errors, insuficient numerical precision, oating point precision, programming errors, data processing errors, visualization errors, etc., as well as human errors. Another uncertainty comes from the discretization step size of Δt or Δx in numerical computations. The discretization step size of Δt or Δx must be selected appropriately in numerical computations in order to describe short waves or fast phenomena concerned to the target problems. This uncertainty would be also reduced by multiple program computations with the different size of Δt and Δx to find out the appropriate step size under keeping numerical stabilities. These uncertainty reduction mechanisms are proposed in this paper.}, issn = {}, doi = {https://doi.org/}, url = {http://global-sci.org/intro/article_detail/ijnamb/226.html} }
TY - JOUR T1 - Uncertainty Management in Scientific Numerical Computation AU - Shigeo Kawata, Takashi Ishihara, Daisuke Barada, J JO - International Journal of Numerical Analysis Modeling Series B VL - 1 SP - 147 EP - 155 PY - 2014 DA - 2014/05 SN - 5 DO - http://dor.org/ UR - https://global-sci.org/intro/article_detail/ijnamb/226.html KW - AB - The uncertainty in scientific computing has induced serious accidents and disasters. New proposals are presented toward an uncertainty inference or management in scientific numerical computing. The first important point is to share the knowledge on uncertainty with other researchers. Sharing known uncertainties contributes to the uncertainty management. In addition, for specific problems, one can prepare multiple programs with different rounding methods and/or with different precisions, and can compare the results among the programs generated for the specific problems. If differences appear among the results, it suggests that the specific program may have some uncertainty problems, and has a lack of the precision or a lack of the digit number. This second method provides a simple and effective inference method of the uncertainty. Uncertainty comes from various sources from physical and mathematical model errors, unknown input data, numerical model errors, insuficient numerical precision, oating point precision, programming errors, data processing errors, visualization errors, etc., as well as human errors. Another uncertainty comes from the discretization step size of Δt or Δx in numerical computations. The discretization step size of Δt or Δx must be selected appropriately in numerical computations in order to describe short waves or fast phenomena concerned to the target problems. This uncertainty would be also reduced by multiple program computations with the different size of Δt and Δx to find out the appropriate step size under keeping numerical stabilities. These uncertainty reduction mechanisms are proposed in this paper.
Shigeo Kawata, Takashi Ishihara, Daisuke Barada, J. (1970). Uncertainty Management in Scientific Numerical Computation. International Journal of Numerical Analysis Modeling Series B. 5 (1). 147-155. doi:
Copy to clipboard
The citation has been copied to your clipboard