Ten new checks to assess the statistical quality of Monte Carlo solutions in MCNP
Conference
·
OSTI ID:10120110
- Los Alamos National Lab., NM (United States)
- Georgia Inst. of Tech., Atlanta, GA (United States)
The central limit theorem can be applied to a Monte Carlo solution if: The random variable x has a finite mean and a finite variance; and the number N of independent observations grows large. When these two conditions are satisfied, a confidence interval based on the normal distribution with a specified coverage probability can be formed. The first requirement is generally satisfied by the knowledge of the type of Monte Carlo tally being used. The Monte Carlo practitioner has only a limited number of marginally quantifiable methods to assess the fulfillment of the second requirement. Ten new statistical checks have been created and added to MCNP4A to assist with this assessment. The checks examine the mean, relative error, figure of merit, and two new quantities: The relative variance of the variance; the empirical history score probability density function f(x). The two new quantities are described. For the first time, the underlying f(x) for Monte Carlo tallies is calculated for routine inspection and automated analysis. The ten statistical checks are defined, followed by the results from a statistical study on analytic Monte Carlo and other realistic f(x)s to validate their values and uses in MCNP. Passing all 10 checks is a reasonable indicator that f(x) has been adequately sampled, N has become large, and valid confidence intervals can be formed. Additional experience with these checks is required to determine their effectiveness in assessing the fulfillment of the central limit theorem requirements for a wide variety of MCNP Monte Carlo solutions. Passing all ten checks does NOT guarantee a valid confidence interval because there is no guarantee that the entire f(x) has been sampled.
- Research Organization:
- Los Alamos National Lab., NM (United States)
- Sponsoring Organization:
- USDOE, Washington, DC (United States)
- DOE Contract Number:
- W-7405-ENG-36
- OSTI ID:
- 10120110
- Report Number(s):
- LA-UR--94-227; CONF-940424--8; ON: DE94006137
- Country of Publication:
- United States
- Language:
- English
Similar Records
Two proposed convergence criteria for Monte Carlo solutions
A new method to assess Monte Carlo convergence
A new method to assess Monte Carlo convergence
Conference
·
Tue Dec 31 23:00:00 EST 1991
· Transactions of the American Nuclear Society; (United States)
·
OSTI ID:6708394
A new method to assess Monte Carlo convergence
Conference
·
Sat May 01 00:00:00 EDT 1993
·
OSTI ID:10155061
A new method to assess Monte Carlo convergence
Conference
·
Thu Dec 31 23:00:00 EST 1992
·
OSTI ID:6522167