Evaluates «x» in Mid-Mode, i.e. deterministically.
Whenever an expression or subexpression is evaluated, it is evaluated either in Mid-mode or Sample-mode, in which sample-mode carries through information about uncertainty whereas mid-mode does not. The article on Evaluation Modes explains this in detail. Mid(x) forces the evaluation of «x» to occur in Mid-mode even when the current evaluation mode is sample-mode.
The Sample function does the opposite -- forcing «x» to be evaluated in sample mode.
Distribution functions return their median value in Mid-mode, or a Monte Carlo when evaluated in Sample-mode.
x := Uniform(-1, 1)^2
Mid(x) → 0
Median(x) → 0.25
Mean(x) → 0.3333
When Mid is evaluated, the median value of
Uniform(-1, 1) is used, which is 0 and is then squared to get the mid-value of
x. As seen, this is not equivalent to the median of
x when uncertainty is properly accounted for.