Cox, Maurice; matematični kolokvij marec 2005

Iz MaFiRaWiki

Evaluating uncertainties associated with measurement results

Maurice Cox

National Physical Laboratory, Velika Britanija

17. marec 2005

Consider a model Y = f(X) of a measurement in which Y denotes the quantity of interest and X a vector quantity from which Y is derived through the function f. Information concerning the values of the components of X is available, either as best estimates and uncertainties or as probability density functions (PDFs). It is required to provide a best estimate of the value of Y and its associated uncertainty.

Approaches to treating this problem are considered, ranging from analytical to approximate analytical to numerical. Emphasis is devoted to a particular numerical method, constituting the use of a Monte Carlo procedure to approximate the PDF for the value of Y, from which a best estimate, an associated uncertainty and other relevant information can be derived.

The relevance of the content of the talk to international standards is indicated.

Examples with appropriate visualizations are used to demonstrate the approaches.

Osebna orodja