Gai E.V.
A. I. Leypunsky Institute for Physics and Power Engineering, Obninsk, Russia
Abstract
Analysis of experimental data by mathematical statistics allows to estimate underlying physical regularities or, at least, to construct smooth approximant of these data. Generally, such analysis uses declared statistic and systematic uncertainties of the experimental data. Conformance verification of these uncertainties with the observable data spread in classical mathematical statistics is accomplished by χ2-criterion. In this work it is shown that in the case of the large number of experimental points, uncertainty of the estimate is defined by the systematic uncertainties of the experiment, but χ2-criterion magnitude does not depend on them.
Keywords
criterion χ2, error, OLS, likelihood function, variance of the experimental data, systematic errors
Article Text (in Russian, PDF)
1. Koks D., Hinkli D. Teoreticheskaja statistika. M., Mir, 1978. [in Russian]
2. Vinogradov V.N., Gaj E.V., Rabotnov N.S. Analiticheskaja approksimacija dannyh v jadernoj i nejtronnoj fizike. – M., Jenergoatomizdat, 1987. [in Russian]
3. Kramer G. Matematicheskie metody statistiki. M., Mir, 1975.
4. Koroljuk V.S. i dr. Spravochnik po teorii verojatnostej i matematicheskoj statistike. – M., Nauka, 1985.[in Russian]
5. Pyt'ev Ju.P., Shishmarev I.A. Kurs teorii verojatnostej i matematicheskoj statistiki dlja fizikov. – Izdatel'stvo Moskovskogo Universiteta, 1983, p. 211. [in Russian]
6. Spravochnik pol'zovatelja biblioteki ENDF, Document ENDF-102, Report BNL-NCS-4495-05-Rev, 30.1 (p.200). [in Russian]
UDC 539.17
Problems of Atomic Science and Technology. Series: Nuclear Constants, issue 1-2, 2011-2012