SOME REMARKS ON THE USE OF U-SHAPE PROBABILITY DISTRIBUTION FUNCTIONS IN MONTE CARLO SIMULATION |
|---|
| A. S. Ribeiro, J. A. Sousa, M. P. Castro |
- Abstract:
- Although Monte Carlo simulation (MCS) has been applied to the calculation of measurement uncertainties in metrology problems, its use has generally been restricted to cases where the distributions assigned to the pdf’s of the input quantities are the more common Gaussian, rectangular and triangular distributions. However, cases exist where the distributions of the input quantities are more complex, such as U-shape (e.g. electrical metrology), and combinations of this and the above and other distributions. This study will establish procedures for sampling from and using such distributions and develop appropriate algorithms.
In particular, this paper will concentrate on approaches that are free from the assumptions inherent in the GUM-based procedures that are conventionally used. Those procedures are at the moment “taken as read” by many laboratories when, in fact, the conditions for their application do not necessarily apply (e.g. normality of distributions involved, input quantities uncorrelated). Thus, simulations will be carried out based on a U-shape probability distribution function, to investigate the validity of the GUM approach using the MCS technique as the validation tool. - Keywords:
- Measurement uncertainty, Monte Carlo simulation, probability distribution function
- Download:
- IMEKO-TC7-2004-034.pdf
- DOI:
- -
- Event details
- IMEKO TC:
- TC7
- Event name:
- TC7 Symposium 2004
- Title:
10th Symposium on Advances of Measurement Science
- Place:
- St. Petersburg, RUSSIA
- Time:
- 30 June 2004 - 02 July 2004