Probability Problems and Solutions
Probability Problems and Solutions
Probability Problems and Solutions
06-02-2018
1. In the sequence of outcomes a when fair coin is tossed n times, let A be the event “The sequence
contains at most one H”, and B the event “The sequence contains both H and T”. Find the value
of n such that A and B are independent.
Solution. A is (equivalent to) the event “The sequence contains all Ts or exactly one H”.
1 n n+1
P(A) = n
+ n =
2 2 2n
B is the complement of the event “The sequence is either all H or all T”.
1 1 1
P(B) = 1 − − = 1 −
2n 2n 2n−1
For A and B to be independent, P(A ∩ B) must be equal to P(A)P(B). A ∩ B is the event “The
sequence contains exactly one H” (since at most one H and at least one H implies exactly one
H).
n
P(A ∩ B) = n
2
Thus, P(A ∩ B) = P(A)P(B) =⇒
n n+1 1
= 1 − n−1 =⇒
2n 2n 2
n+1
n = n + 1 − n−1 =⇒
2
2 = n + 1.
n−1
2. Hilbert opens a hotel with infinitely many rooms 1, 2, . . . and decides to give each arrival the
1
nth room with probability n . If the room is occupied, the visitor is sent back without a room.
2
What is the probability that the second person to arrive at the hotel gets a room? Bonus: If the
first two arrivals get rooms, what is the probability that the third person to arrive does not get
one?
1
VM
1
Solution. The first person gets the nth room with probability . The second person does not
2n
get a room if he is (randomly) allocated the same nth room as the first person. The probability
1
of this is again n . Summing over all possibilities, the probability that the second person does
2
not get a room is
∞ ∞
1 2 X 1 1
X
n
= n
= .
n=1
2 n=1
4 3
2
The probability that the second person gets a room is, therefore, .
3
Bonus part: If the first two persons get rooms, assume they are in rooms m and n, 1 ≤ m < n.
In order for the third person to not get a room, he must be allocated room m or room n, and
1 1
these mutually exclusive events happen with respective probabilities m and n . Thus, the
2 2
probability that he does not get a room is
∞ X ∞ ∞ ∞ ∞
1 1 1 1 X 1 1 1
X X X
+ n =2 + m
2 m+n m
2m n 2n
m=1 n=m+1
2 2 2 m=1
2 n=m+1
2 2 n=m+1
2
∞ ∞ ∞
X 1 1 X 1 1 1 X 1
=2 + m× m
×
4m 2m 2 n 2 4 4 n
m=1 n=1 n=1
∞
1 1
X
=2 1+
m=1
8m 3
4 1
=2× ×
3 7
8
=
21
3. A person claims to have psychic abilities and to be able to predict the outcome a fair coin
toss with 80% accuracy. Let p be the probability that the person is psychic, and q = 1 − p the
probability that he is not, and guesses randomly. What is the probability that he is psychic,
given that he correctly predicts
2
VM
(a) Let B be the event that he predicts 8 out of 10 outcomes correctly. Then
10 4 8 1 2
! !
10 48
P(B | A) = =
8 5 5 8 510
!
10 1
P(B | Ā) =
8 210
10 413 + 510
! ! !
10 48 1
P(B) = + = .
8 510 210 8 1010
413 p 1 1
P(A | B) = = ≈ q,
413 p + 510 q 510 q 1 + 0.145
1 + 13 × p
4 p
4120 p 1 1
P(A | C) = = ≈ q,
4120 p + 5100 q 5100 q 1 + 0.005
1 + 120 × p
4 p
(assuming p , 0).
Observe that P(A | C) ≥ P(A | B) (with equality only when p = 0). Thus, we are more certain of
our conclusion (that he is psychic) in the second case, where his predictions were correct only
70% of the time, than in the first case where his predictions were correct 80% of the time. This is
purely due to the larger number of trials (i.e., the sample size) on the basis of which we reached
the conclusion in the second case. In fact, in the second case, if p ≥ 33%, then our certainty is
more than 99%. On the other hand, in the first case, we need p ≥ 93% to achieve the same level
of certainty.
Moral of the story: Don’t trust psychics/astrologers/palmists/tarot card readers unless they can
predict at least 70 out of 100 tosses of a fair coin (and even then, make sure they did not just
make it look like they did).