Risk glossary

Quantization is a method of representing continuous random variables using discrete ones, in order to make model simulations less computationally intensive. That means replacing variables that can take any value – for example 1, 2, 3, 4, 5 and any fraction in between – with variables that can take only a limited number of values, such as rounded-up numbers. Using an arbitrary set of discrete numbers to represent the original continuous variables will produce errors in calculations. Therefore, methods have been developed to choose the discrete values carefully so as to minimise these errors.

Click here for articles on quantization.