Professional Documents
Culture Documents
(a) Permutations (number of ways to linearly arrange k objects out of n distinct objects, when the
order of the k objects matters):
n!
P (n, k) =
(n − k)!
(b) Combinations (number of ways to choose k objects out of n distinct objects, when the order of the
k objects does not matter):
n! n
= = C(n, k)
k!(n − k)! k
(d) Multinomial coefficients: Suppose there are n objects, but only k are distinct, with k ≤ n. (For
example, “godoggy” has n = 7 objects (characters) but only k = 4 are distinct: (g, o, d, y)). Let ni be
the number of times object i appears, for i ∈ {1, 2, . . . , k}. (For example, (3, 2, 1, 1), continuing the
“godoggy” example.) The number of distinct ways to arrange the n objects is:
n! n
=
n1 !n2 ! · · · nk !
n1 , n 2 , . . . , n k
(g) Pigeonhole Principle: If there are n pigeons with k holes and n > k, then at least one hole contains at
least 2 (or to be precise, d nk e) pigeons.
(h) Complementary Counting (Complementing): If asked to find the number of ways to do X, you can:
find the total number of ways and then subtract the number of ways to not do X.
(a) Sample Space: The set of all possible outcomes of an experiment, denoted Ω or S
(b) Event: Some subset of the sample space, usually a capital letter such as E ⊆ Ω
(c) Union: The union of two events E and F is denoted E ∪ F
(d) Intersection: The intersection of two events E and F is denoted E ∩ F or EF
(e) Mutually Exclusive: Events E and F are mutually exclusive iff E ∩ F = ∅
(f) Complement: The complement of an event E is denoted E C or E or ¬E, and is equal to Ω \ E
1
(g) DeMorgan’s Laws: (E ∪ F )C = E C ∩ F C and (E ∩ F )C = E C ∪ F C
(h) Probability of an event E: denoted Pr(E) or Pr(E) or P (E)
(i) Partition: Nonempty events E1 , . . . , En partition the sample space Ω iff
• E1 , . . . , En are exhaustive: E1 ∪ E2 ∪ · · · ∪ En = ni=1 Ei = Ω, and
S
(k) Equally Likely Outcomes: If every outcome in a finite sample space Ω is equally likely, and E is an
|E|
event, then Pr(E) = .
|Ω|
• Make sure to be consistent when counting |E| and |Ω|. Either order matters in both, or order doesn’t
matter in both.
Pr(A ∩ B)
(l) Conditional Probability: Pr(A | B) =
Pr(B)
(m) Independence: Events E and F are independent iff Pr(E ∩ F ) = Pr(E) Pr(F ), or equivalently Pr(F ) =
Pr(F | E), or equivalently Pr(E) = Pr(E | F )
Pr(B | A) Pr(A)
(n) Bayes Theorem: Pr(A | B) =
Pr(B)
(o) Partition: Nonempty events E1 , . . . , En partition the sample space Ω iff
(p) Law of Total Probability (LTP): Suppose A1 , . . . , An partition Ω and let B be any event. Then
Pr(B) = ni=1 Pr(B ∩ Ai ) = ni=1 Pr(B | Ai ) Pr(Ai )
P P
(q) Bayes Theorem with LTP: Suppose A1 , . . . , An partition Ω and let B be any event. Then Pr(A1 | B) =
Pr(B | A1 ) Pr(A1 ) Pr(B | A) Pr(A)
Pn . In particular, Pr(A | B) =
i=1 Pr(B | Ai ) Pr(Ai ) Pr(B | A) Pr(A) + Pr(B | AC ) Pr(AC )
(r) Chain Rule: Suppose A1 , ..., An are events. Then,