M5: Monte Carlo simulation

Question 1

What is monte carlo simulation suitable for?

Answer 1

Is suitable for:
o Any distribution function of the random variables
o Large uncertainties
o Complex equations

Question 2

List the 5 steps in the monte carlo procedure

Answer 2

1. Specify probability distributions for all input parameters in place of
   constant values
2. Draw a random sample of all input parameters
3. Compute the result for the selected set of random samples
4. Repeat 2 and 3 for many trials (hundreds or thousands of iterations)
   
   • -> Law of large numbers says convergence
5. Analyze the resulting model outputs using statistics
   
   • -> Histogram, mean, standard deviation…

Question 3

Mention the two ways uncertainty propagation can be done.

Answer 3

• Uncertainty propagation can be done analytically or numerically
–> numerical (MC) approach is simpler (for complex problems) and more flexible (not limited to normal distributions) but requires a computer