Moment Generating Functions
Moment generating functions: derivatives and independent sums
The moment generating function is MX(t) = E[etX], when this expectation exists for t in an interval around zero. Its value at zero is 1. Differentiation exposes moments: M′(0) = E[X] and M″(0) = E[X2], so variance is M″(0) − [M′(0)]2.
MGFs are particularly efficient for sums. Since et(X+Y) = etXetY, independence allows MX+Y(t) = MX(t)MY(t). The word independent licenses the factorization; without it, multiplying individual MGFs can be wrong.
An MGF identifies many common distributions when it exists near zero, but existence is a mathematical condition. Do not assume every random variable has a finite MGF merely because it has a mean or variance.
A random variable X equals 1 with probability 0.7 and 3 with probability 0.3, so MX(t) = 0.7et + 0.3e3t. What is E[X]?
- 1.0
- 1.6
- 2.0
- 3.0
X and Y are independent Bernoulli variables with success probabilities 0 < p < 1 and 0 < q < 1. Which expression is MX+Y(t)?
- (1 − p + pet) + (1 − q + qet)
- 1 − (p + q) + (p + q)et
- (1 − p + pet)(1 − q + qet)
- et(p+q)