Probability Homework 2

Michael Dickerson
13 September 2000

(1)A system is composed of 5 components, each of which is either working or failed. Consider
an experiment that consists of observing the status of each component, and let the outcome of
the experiment be given by the vector (x1  x2  x3  x4  x5 ), where xi is equal to 1 if component i
is working and is equal to 0 if component i is failed.
1. How many outcomes are in the sample space of this experiment?
2. Suppose that the system will work if components 1 and 2 are both working, or if components
3 and 4 are both working, or if components 1, 3, and 5 are all working. Let W be the event
that the system works. Specify all the outcomes in W .
3. Let A be the event that components 4 and 5 are both failed. How many outcomes are
contained in the event A?
4. Write out all the outcomes in the event AW .
First observe that there are 25 = 32 possible outcomes. If W is the event that the system works, then
the outcomes in W are:
(1,1,0,0,0) (0,0,1,1,0) (1,0,1,0,1)
(1,1,0,0,1) (0,0,1,1,1) (1,0,1,1,1)
(1,1,0,1,0) (0,1,1,1,0) (1,1,1,0,1)
(1,1,0,1,1) (0,1,1,1,1) (1,1,1,1,1)
(1,1,1,0,0) (1,0,1,1,0)
(1,1,1,1,0)
If we then nd the outcomes in A, the event that components 4 and 5 are both failed, there are 8:
(0,0,0,0,0), (0,0,1,0,0), (0,1,0,0,0), (0,1,1,0,0), (1,0,0,0,0), (1,0,1,0,0), (1,1,0,0,0), (1,1,1,0,0). By comparing
the lists, or by simple computation, we nd that the outcomes in event AW are only two:
(1 1 0 0 0)(1 1 1 0 0)

(2)The following data were given in a study of a group of 1000 subscribers to a certain magazine:
In reference to job, marital status, and education, there were 312 professionals, 470 married
persons, 525 college graduates, 42 professional college graduates, 147 married college graduates,
86 married professionals, and 25 married professional college graduates. Show that the numbers
reported in the study must be incorrect.
1

Let P be the set of professionals, M be the set of married persons, and G be the set of college graduates.
Then, by inclusion-exclusion, the sample size must be exactly

jP j + jM j + jGj ; jP \ M j ; jP \ Gj ; jM \ Gj + jP \ M \ Gj
Since we were given the size of each of those sets, we know the size of the sample is
312 + 470 + 525 ; 42 ; 147 ; 86 + 25 = 1057
If the study claimed to include 1000 people, something has to be incorrect.

(3)Two cards are randomly selected from an ordinary playing deck. What is the probability that
they form a blackjack?
We assume; that
each pair of cards in the deck will be selected with equal probability, giving a sample

space of size 522 . Of those pairs, there are 4  16 = 64 possible blackjacks (since there are 4 aces and 16
cards worth ten points). Thus the probability is
64
64 = 32
;52 =
1326 663
2

(4)Suppose that you are playing blackjack against a dealer. In a freshly shued deck, what is
the probability that neither you nor the dealer is dealt a blackjack?
Like the previous problem, assume that all hands are equally likely, and there are; 64blackjacks among
the 1326 possible hands. Thus the total number of ways
two hands
could be dealt is 1326
2 , and the number
;1326;64
;1262
of ways two non-blackjack hands could be dealt is
=
.
Therefore,
the
probability
of neither
2
2
player being dealt a blackjack is
;1262

2 
;1326
2

795691  :906
= 878475

(5)The game of craps is played as follows: A player rolls two dice. If he sum of the dice is either
a 2, 3, or 12, the player loses if the sum is either a 7 or 11, he or she wins. If the outcome is
anything else, the player continues to roll the dice until he or she rolls either the initial outcome
or a 7. If the 7 comes rst, the player loses whereas if the initial outcome reoccurs before the 7,
the player wins. Compute the probability of a player winning at craps.
Let E denote the event that the player wins with an initial roll of i, for each i 2 f2 3 : : : 12g. Further,
let E be the event that the player wins on roll n after an initial roll of i. Then we see that
i

in

E =

1


=1

ik

Because the E are all disjoint (since a player cannot win a game on two dierent rolls), we can express
the probability of E as
ik

P (E ) =

1
X

ik

=1

Then, observe that for E to occur, three things must both happen: the rst roll must be an i, the
player must neither win nor lose on rolls 2 to k ; 1, and the player must win on roll k. Thus, if we x an
i, we can nd the probability p of the player winning on the next roll, as well as the probabilty q of the
player neither winning nor losing on the next roll. Then, for the i which are not specially described (i.e.
i 26 f2 3 7 11g, we nd that the probability of winning is a sequence like the following:
ik

P (E
P (E
P (E
P (E

pp
pqp
pqqp
p2 q ;1
Since the series of P (E ) is geometric with initial value p2 and ratio q < 1, we know that
i
i
i

)
2)
3)
)
1

ik

=
=
=
=

ik

1
X

E = 1 p; q = P (E )
=1
2

ik

Note that from the rules of the game, we read that P (E2 ) = 0, since the player has already lost if the
initial roll is 2. Similarly, P (E3 ) = P (E12 ) = 0. And the player automatically wins if the initial roll is 7 or
11, so P (E7 ) is exactly the probability of rolling that rst 7, which is 1=6. Likewise, we nd P (E11 ) = 1=18.
We can now use the above formula to compute all the other E , and we nd
i

P (E2 ) = 0
P (E3 ) = 0
2
P (E4 ) = 1(3;=36)
9=36  :009
36)2  :017
P (E5 ) = 1(4;=10
=36
36)2  :028
P (E6 ) = 1(5;=11
=36
6
P (E7 ) = 36  :167
36)2  :028
P (E8 ) = 1(5;=11
=36
36)2  :017
P (E9 ) = 1(4;=10
=36
2
P (E1 0) = 1(3;=36)
9=36  :009
2  :055
P (E1 1) = 36
P (E1 2) = 0
Finally, let E be the event that the player wins.S Since the initial roll must obviously be one of 2 : : : 12
no matter how the player wins, we know that E = 12=1 E , and since the E are all disjoint, the probability
P (E ) is simply the sum of the P (E ):
i

P (E ) =

12
X

P (E )  :43
i

=1

Hence the probability of winning at craps is about 0.43.

(6)A person has n keys, one of which will open his or her door.
1. If he or she tries the keys at random, discarding those that do not work, what is the probability
that he or she will open the door on the kth try?
2. What if the person does not discard previously used keys?
Let E be the event that the kth key opens the door. For E to occur, the rst k ; 1 tries must not open
the door, and the kth try must be successful. Thus, for k 2 f1 2 : : : ng,
k

2 )    ( n ; k + 1 )( 1
P (E ) = ( n ;n 1 )( nn ;
;1
n;k+2 n;k+1
k

= n1

We could also see that P (E ) = 1=n if we assume that the keys were in a random sequence, and the
person tried them in order: then clearly the chance of the winning key being in any particular one of the n
spots is 1=n. However, if the person replaces each key after trial, then the chance of success does not change
with each trial, and now
k

(n ; 1)
1
P (E ) = ( n ;
n ) (1=n) = n

;1

(7)Two players play the following game. Player A chooses one of the three spinners below, and
then player B chooses one of the remaining two spinners. Both players then spin their spinner
and the one that lands on the higher number is declared the winner. Would you rather be player
A or player B ?

First, consider the 3 potential contests between spinners: a vs. b, b vs. c, and a vs. c:

a
9
9
9
5
5
5
1
1
1

b Winner a c Winner b c Winner

8
a
9 7
a
8 7
b
4
a
9 6
a
8 6
b
3
a
9 2
a
8 2
b
8
b
5 7
c
4 7
c
4
a
5 6
c
4 6
c
3
a
5 2
a
4 2
b
8
b
1 7
c
3 7
c
4
b
1 6
c
3 6
c
3
b
1 2
c
3 2
b

By counting the outcomes we see that a wins against b with probability 5=9, a wins against c with
probability 4=9, and b wins against c with probability 5=9. Assume that player B knows this, perhaps
having taken probability in college, and also assume that player B is trying to win.
1. If player A selects spinner a, B will choose c, giving B the advantage with a 5=9 chance of winning.
2. If player A chooses spinner b, B can choose a and have, again, a 5=9 chance of winning.
3. If player A chooses spinner c, B then chooses spinner b in order to have a 5=9 chance of winning.
Thus, if player B is intelligent, he or she will always have the advantage, with a 5=9 chance of winning.
Furthermore, even if player B plays randomly, the most A can hope for is a 5=9 chance of winning. Thus an
intelligent player would prefer to be player B , regardless of the opponent.

(8)Prove that if E

F , then F c E c.

Let x 2 F . Then x 62 F , which means x 62 E , because E  F . Therefore x 2 E , and since x was

arbitrary, conclude that F  E .
c

(9)Prove the following:

1.

( Ei)F =
2.

Ei F

( Ei)  F = (Ei  F )
Suppose E is a set for any i 2 f1 2 : : :gS, and that x 2 ( 1=1 E )F . Then
x 2 F , and x 2 E for some
there must exist
k. Therefore, x 2 E F , which means x 2 1=1 E F . Conversely, if y 2 S1=1 E F , then
S
someSk such that y 2 E F . Thus y 2 E , and y 2 F . But y 2 E means that y 2 1=1 E , so therefore,
y 2 ( 1=1 E )F . Conclude that
S

1


( E )F =

1


EF
i

1 E . If x 2 F , then x 2 E F
Next, suppose that x 2 ( 1
1 E ) F . Then either x 2 F , or x 2
1
T1
regardless of theT E . The same is true if x was not in F , but was in all the E instead. So
=1 (E F ).
T1x 2
1
And if y 2 =1 (E F ), then for each
i
,
y
2
E

F
.
If
y
2
F
,
then
clearly
y
2
(
E
)

F . But if
=1
T
T
y 62 F , then y 2 E 8i, which means y 2 1=1 E , so again y 2 ( 1=1 E ) F . Conclude that
T

1
\

1
\

( E ) F =

(E F )
i

(10)For any sequence of events E1  E2  : : :, dene a new sequence F1  F2  : : : of disjoint events

such that for all n  1,
n
n

1

Fi = Ei
1

Let E1  E2  : : : be any arbitrary sequence of events. Dene F1 = E1 . Then let F2 = E2 n E1 , that is

F2 = E2 \ E1 . Similarly, let
c

= E n E ;1 n E ;2 n    n E1
= E \ E ;1 \ E ;2 \    \ E1
k

c
k

c
k

Then the F are all disjoint, because for arbitrary j > k,

i

F \ F = E \ E ;1 \    \ E1 \ E
 E \E
= 
S
But for any n, the union of the rst nF is equal to the union of the rst nE , because if x 2 =1 E ,
then there exists some smallest k such that x 2 E , and x 2 F for that k.
j

c
j

c
k

(11)Find the simplest expression for the following events:

1. (E  F )(E  F c)
2. (E  F )(E c  F )(E  F c)
3. (E  F )(F  G)
(E F )(E F ) = E FF
= E 
= E
c

(E F )(E F )(E F ) =
=
=
=
=
c

((E F )(E F ))(E F )

E (E F )

EE EF
 EF
EF
c

(E F )(F G) = F EG

(12)Let S be a given set.

If, for some k > 0, S1  S2  : : :  Sk are mutually exclusive nonempty
S
subsets of S such that ki=1 Si = S , then we call the set fS1  S2  : : :  Sk g a partition of S . Let
Tn denote the number of dierent partitions of f1 2 : : :  ng.
1. Show, by computing all partitions, that T3 = 5, T4 = 15
2. Show that
 !
n

Tn+1 = 1 +

and use this to compute T1 0.

k =1

n T
k k

We can partition f1 2 3g into ff1 2 3gg, ff1g f2 3gg, ff1 2g f3gg, ff1 3g f2gg, or ff1g f2g f3gg, so
we know T3 = 5.
The dierent partitions of f1 2 3 4g are

ff1 2 3 4gg
ff3g f1 2 4gg
ff1 3g f2 4gg
ff1 3g f2g f4gg
ff2 4g f1g f3gg

ff1g f2 3 4gg ff2g f1 3 4gg

ff4g f1 2 3gg ff1 2g f3 4gg
ff1 4g f2 3gg ff1 2g f3g f4gg
ff1 4g f2g f3gg ff2 3g f1g f4gg
ff3 4g f1g f2gg ff1g f2g f3g f4gg
so T4 = 15. More generally, suppose that we know T for all k n, and we want to nd T +1 . Take
your favorite ball from the n + ;1, and decide how many more balls you want to put in a set with it. Call
that number k. You then have
choices for the set containing your favorite ball, and n ; k balls left to
partition. We can partition those n ; k balls in any of T ; ways, so if we remember that you could have
picked anything from 0 to n for k, we see that
X n
T +1 = 1 +
k T
=1
k

n
k

(13)Suppose that an experiment is performed n times. For any event E of the sample space, let
n(E ) denote the number of times that event E occurs, and dene f (E ) = n(E )=n. Show that
f () satises Axioms 1, 2 and 3.
7

Event E could not have occurred less than 0 times, and since we only did teh experiment n times, E
could not have occurred more than n times. Therefore, 0 n(E ) n, which means 0 n(E )=n 1, and
f () satises Axiom 1.
P
Each of then n repetitions of the experiment resulted in exactly one event E , so
2 n(E ) = n.
Therefore, f (S ) = n=n = 1, and f () satises Axiom 2.
Finally, if E1  E2  : : : is a sequence of disjoint events, we know that n(E1 E2   ) = n(E1 )+ n(E2 )+   .
Thus
E

1


P1

n(E )
n
1 n(E )
X
=
( n )
=1

f( E ) =
i

=1

=1

1
X

f (E )
i

=1

which is to say that f () satises Axiom 3.

(14)If P (E ) = :9 and P (F ) = :8, show that P (EF )  :7. In general, prove Bonferroni's
inequality, which is

P (EF )  P (E ) + P (F ) ; 1

Note that P (EF ) + P (EF ) = 1, and since P (EF ) = P (E F ), we have

c

P (EF ) + P (E F ) = 1
P (EF ) + P (E ) + P (F ) ; P (E F ) = 1
P (EF ) + (1 ; P (E )) + (1 ; P (F )) ; P (E F ) = 1
P (EF ) + 0:1 + 0:2 ; P (E F ) = 1
P (EF ) ; P (E F ) = 0:7
P (EF ) = 0:7 + P (E F )
and since P (E F ) is nonnegative, we know P (EF )  0:7. In general, if we don't know P (E ) and P (F ),
c

we nd that

P (EF ) + P (E F )
P (EF ) + P (E ) + P (F ) ; P (E F )
P (EF ) + (1 ; P (E )) + (1 ; P (F )) ; P (E F )
P (EF ) + 1 ; P (E ) + 1 ; P (F ) ; P (E F )
P (EF )
P (EF )
c

=
=
=
=
=

1
1
1
1

P (E ) + P (F ) ; 1 + P (E F )
 P (E ) + P (F ) ; 1
c

(15)Use induction to generalize Bonferroni's inequality to n events. Namely, show that

P (E1E2    En )  P (E1) + P (E2) +    + P (En) ; (n ; 1)

8

In the previous problem we showed that P (EF )  P (E ) + P (F ) ; 1, in the case with only two events.
Suppose that we know

P (E1 E2    E )  P (E1 ) + P (E2 ) +    + P (E ) ; (n ; 1)

What can we say about the probability of (E1 E2    E ) together with some new event, E
applying Bonferroni's equality to the events (E1    E ) and E +1 , we see that
n

P ((E1    E )E
n

+1

+1

? By

)  P (E1    E ) + P (E
n

+1

);1

And by applying the inductive hypothesis, we nd that

P (E1 cdotsE E
n

+1

 P (E1 ) +    + P (E ) ; (n ; 1) + P (E +1 ) ; 1
 P (E1 ) +    + P (E ) + P (E +1 ) ; (n)
n
n

(16)Let fn denote the number of ways of tossing a coin n times such that successive heads never
appear. Argue that
fn = fn;1 + fn;2wheref0 = 1 f1 = 2
If Pn denotes the probability that successive heads never appear when a coin is tossed n times,
nd Pn when all possible outcomes of the n tosses are assumed equally likely.
Suppose we want to nd f , the number of ways of tossing a coin n times with no consecutive heads
appearing. Each sequence must begin with a head or a tail. If the rst coin toss is a tail, we can follow it
with any of the f ;1 sequences of length n ; 1 in which no consecutive heads appear. However, if the rst
coin toss is a head, we must have the second coin toss a tail in order to avoid consecutive heads. We can
then follow with any of the f ;2 possible sequences of length n ; 2 in which no consecutive heads appear.
Therefore, f = f ;1 + f ;2 .
We can use this formula to quickly nd values of f .
n

n 0 1 2 3 4 5 6 7 8 9 10
f 1 2 3 5 8 13 21 34 55 89 144
Since we also know the total number of sequences of n coin tosses is 2 , we can easily nd values for
P (n) = f =2 , the probability of no consecutive heads appearing when a coin is tossed n times. For example,
144
P (10) = 1024
= 649 .
n

(17)Consider an experiment whose sample space consists of a countably innite number of points.
Show that not all points can be equally likely. Can all points have positive probability of occurring?
Recall that P (S ) = 1, so there must exist some event whose probability is nonzero. That event must
contain some point a with probability  > 0 (otherwise the event would
have probabilityP0). Suppose all
P
points x 2 S had probability . There are innitely many points, so 2 P (x) = P (S ) = 2  diverges.
But that contradicts the requirement that P (S ) = 1. Therefore, all points x 2 S cannot have the same
probability.
x

However, it is possible for all points to have positive probability of occurring. Consider this experiment:
Toss a coin repeatedly until the rst head appears. The probability of this happening on the rst toss is
1=2, the probability of it happening on the second toss is 1=4, etc. The sample space is the set of all positive
integers f1 2 3 : : :g, which is certainly countably innite, and each outcome has a positive probability of
occurring.

10