The SimLang function LOGNORMAL can take two arguments, one that is advertised as a mean,
and one as a standard deviation. But they don’t seem to be the mean and the SD of the results
of the function. For example when I write:
M StartTime = 0
M EndTime = 100
M TimeStep = 1
V Foo = LOGNORMAL(FooMean, FooSD)
V FooAVG = ACCUM(Foo) / TIME
D FooMean = 10
D FooSD = 3
FooAVG is consistently about 14.5, not 10.
Perhaps those 2 arguments are not the mean and SD of the distribution, but rather the
scale and location. But that isn’t right either. When I calculate the scale and location based
on a mean of 10 and an SD of 3, i.e. when I do this:
V Foo2 = LOGNORMAL(Foo2Location, Foo2Scale)
V Foo2AVG = ACCUM(Foo2) / TIME
D FooMean = 10
D FooSD = 3
V Component2 = 1 + FooSD * FooSD / FooMean / FooMean
V Foo2Location = LN(FooMean / SQRT(Component2))
V Foo2Scale = SQRT(LN(Component2))
I see a Foo2AVG of about 2.7.
So what do the arguments to LOGNORMAL mean?
(To be fair, I am testing this in Forio Simulation, so perhaps the same
function is defined differently in Epicenter.)
Dave