Lovegrove Mathematicals

logo

"Dedicated to making Likelinesses the entity of prime interest"

Dice

What is a die?

For our purposes, a die is any element of S(6).

Although the labels ("♣","♦","♥","♠","♔","⚓") are sometimes used, we usually retain use of the labels ("1",...,"6").

The die (1/6,...,1/6) is called the fair die.

A physical die used for the purposes of gaming will be called an actual die. From the mathematical point of view, the set of actual dice is represented by a small-diameter set of dice. The meaning of small is a matter for experimental investigation. Likewise, the centre of the set is also a matter for experimentation but, in the absence of such experiments, is traditionally taken as being the fair die.

What is a roll of a die?

A 'roll' of the die f is defined algorithmically.

We use the computer's RANDOM function to select a number in ]0,1].

definition of a die

The probability of "i", given f, is the likeliness over the singleton set {f} of "i". This is f(i), so the probability of "i" is f(i).

Bias

Care needs to be taken when using the expression 'biased towards i' with a degree greater than 2, in particular with dice. There are two meanings which are not equivalent. This does not affect coins, which are of degree 2 and for which the two meanings are equivalent.

First meaning of 'biased towards i'

The expression 'f is biased towards i' is often used to mean that f(i) is above average. That is, if the degree is N then f(i)>1/N .

With this interpretation, f can be biased towards up to (N-1) outcomes; for example, the die (0.18,0.19,0.03,0.20,0.21,0.19) is biased towards each of 1,2,4,5 and 6.

Second meaning of 'biased towards i'

Alternatively, 'f is biased towards i' can mean that f(i) is the uniquely largest value. Allowing the possibility of non-uniqueness would result in the fair die being considered to be biased. That is, that f(i)>f(j) for all j≠i.

If a distribution is biased towards i according to the second definition then it is also biased towards i according to the first, but the converse does not apply: the set of distributions which are biased towards i according to the second definition is a subset of those which are biased towards i according to the first.

Regardless of which concept of bias is used, a die is biased iff it is not the fair die.

Go to Coins or to Q&A about Coins and Dice