Implementation of Monte Carlo method ("MCM")
For every input measurand, M values are generated
(
by default in GUM_MC, it can be modified in
"options" menu)
following the PDF choosen for
.
So we have samples:
for
:
(
,
,...,
)
for
:
(
,
,...,
)
...
for
:
(
,
,...,
)
The model is evaluated for each of the M draws from the PDFs
for the N
input quantities:
Estimate of the output
quantity is the average
Estimate of
standard uncertainty is standard
deviation:
Values of Y are assembled into histogram form a frequency distribution
wich is an approximation of
.
Then we can approximate the
the distribution function for Y
:
(number of cells in histogram can be modified in "options" menu).
Skewness is a measure of the
degree of asymmetry of a distribution and is estimated by:
.
It's value is 0 for a symetric distribution.
Excess kurtosis is estimated by:
It's value is 0 for a gaussian distribution, -1.2 for an uniform distribution.