HOMEWORK 4: SOLUTIONS

Problem 1. Let X be a continuous random variable with density

(
3e−3x , when x > 0,
f (x) =
0, elsewhere.

(a) Verify that f is a density function.

(b) Calculate P (−1 < X < 1).
(c) Calculate P (X < 5).
(d) Calculate P (2 < X < 4 | X < 5).
(e) Find the cumulative distribution function of X.
(f) Find the mean of X. Compute E(e2X ).
(g) Find the variance of X.

Solution to Problem 1. The numbers in bold refer to the textbook problem

number from where it has been taken.

Date: Oct 31, 2018.

1
 60 Solutions to Chapter 3

(b) The probability that X is odd is

2
1 9 3
P (X 2 {1, 3, 5}) = p(1) + p(3) + p(5) = (1 + 3 + 5) = = .
21 21 7
3.3. (a) We need to check that f is non-negative and that it integrates to 1 on R.
The non-negativity follows from the definition. For the integral we can compute
Z 1 Z 1
x=1
f (x)dx = 3e 3x dx = e 3x x=0 = lim ( e 3x ) ( e0 ) = 0 ( 1) = 1.
Solutions to Chapter 3
1 0 x!1 61
In the first step we used the formula for f (x), and the fact that it is equal to 0 for
x  0.
(b) Using
3.5. the definition
The possible values of
of the probability
a discrete randomdensity function
variable we get the values where
are exactly
the c.d.f. jumps. In this case
Z 1 these areZthe1 values 1, 4/3, 3/2 and 9/5. The corre-
x=1
spondingP ( probabilities
1 < X < 1) are
= equal = size3eof 3x
to the
f (x)dx dx = e 3x x=0
corresponding = 1 e 3.
jumps:
1 0
pX (1) = 13 0 = 13 ,
(c) Using the definition of the probability 1density function again we get
pX (4/3) = 2 13 = 16 ,
Z 5 Z 5
pX (3/2) = 34 3x 12 = 14 , 3x x=5
P (X < 5) = f (x)dx = 3e dx = e x=0
= 1 e 15 .
Solutions to Chapter 3 1 3
pX (9/5) = 1 4 = 4 .
0 1 63
(d) From
3.6. For the therandom
definition of conditional
variable in Exercise probability
3.1, we may use (3.13). For s 2 ( 1, 1),
8
From this we get that m = 4 works as P (2>
the > <
0, X <s 4but
median, <and1 anyX number
< 5)
P (2 < X < 4 | X < 5) = >
> . that is larger
or smaller than 4 is not a median. >
> 1 P (X < 5)
>
> , 1s<2
>7
For XP (2
We have from
<X Exercise
< 4 and3.3 Xwe < 5)have = P (2 <> >
< 3<, 4).
X Similar to the previous parts:
62 14 2  s3m < 3Solutions to Chapter 3
P (X  m) = FZ(s)
14 = e P (X
3m Z s)
, and 4P= (X> m) = e if m 0
P (2 < X < 4) = f (x)dx = 3e >>
>
>
3x 6 , 3  s3x<x=4
dx
14 = e 4 = e 6 e 15 .
and P (X  m) = 0, P (X m) = 1 for m< > 0. From this x=2 we get that the median
2 2 > 10
m satisfies e 3m = 1/2, which leads to m => >
> ,
ln(2)/3. 4  s < 5
> 14
3.8.
Using
(b) (a)the
We We
need have
result
P (X  q)(c): 0.9 and P (X :q)
of part
1, 0.1.
5  s. Since X is continuous, we
must have P (X  X 5 q) + P (X q) = 1 and hence P (X  q) = 0.9 and 7 P (X
1 P (21< X < 4) 3 e 62 3m e 152
q) =For
0.1.the
E[X]random
Using the<variable
P=(2 kp< X (k)
calculations
X 4|X in
=< ·5) =
1Exercise
from 2 ·3.3,
+part wewe
(a)+ · see use
3may += 4 ·(3.15).
that e + 515·=For.0.1 . 0 we
=s from have
which
7 P14
(X < 5)14 17 e 7 2
that
q = ln(10)/3. k=1

(b)
3.4. We
3.14. (a) have
The The density
mean of the X is 16 on
of random P[4,(X
variable 10]X s)from
and =zero
0,Exercise
otherwise. 3.1 Hence,
is
whereas for s > 0X we
55 have
X 1Z s 1 1 1 63 43 21 2 2 72 25
E[|X E[X] 2|] == |kkp
P (X <X
X 2|p
(k) 6) =
=(k)1 ·P=(41+< · 2X·+< 0+
3x ·6)3=· + 1 + · 4=·3s+ 2.+· 5 · + 3=· . = .
k=1
P (X  s) =7 73e 14 dx 14=14 1 6 e14 73. 7 7 27 14
k=1
0
(b)
3.9. (a) Since
The (a)
3.7. second
If (aXisXcontinuous,
P moment is
 b) = 1 then we can compute
F (y) =pZ
its ymean
0 for <a as
p and F (y) = 1 for y b.
5 Z 1 1
From P (|X X
the definition
7| > 1) = ofP F(Xwe 7see that
1 + Pa2 (X =1 27 and 3b = 3x3
(Xgives
2 > 8)the + 2Psmallest
< 6) such
2> 1) < 1) =P (X197
E[X 2 ] =
interval. k 2 pE[X]
X (k) = 1 · xf 2 · =+ 32 ·x · 3e
+(x)dx + 42 dx.
· + 52 · = .
10 8 117 2 14 0 14 7 7 14
(b) Since X is k=1 =
continuous, P +
(X ==1.6) . = 0. We can also see this directly from F :
Using integration by parts 6 3 3 1
Therefore, the variance is we can evaluate the last integral to get E[X] = 3 .
(c) eFor 4isaP (X6 = we1.6) of = X,F (1.6) islim F (x) = F (1.6)
we◆2canF compute
(1.6 ). E[e2X ] as
2X
(b) tfunction
 have and X x!1.6 continuous, so✓
follows: 2 2 197 7 51
Var(X)
Z 1 = atPE[X ] (actually,
(E[X])
Z< =it is continuous = .
Since F (x) is continuous x(X=< 1.6 t, X 1 6) P14(X < t)2 Z 1 everywhere),
t28 4 t 4 we have
P
F (1.6 ) E[e(X <2X t
= F (1.6) | X < 6)
] = and this = 2x
e fgives(x)dx P =(X = 2x
1.6)
e =· =
3e0 3x
again.
dx = = 3 · 3e x =
dx = 3. .
P (X < 6) 1/3 6 2
Now let X be the random variable from Exercise 3.3. The mean is
(c) Because X is continuous, Z 1we have P (1Z1X  3/2) = P (1 < X  3/2). We
1 0 0

also have
3.10. (a) The random variable |X| takes values 0 and3x1 with 1probabilities
E[X] = xf (x)dx = x · 3e dx = ,
= 13P1(1and 3
P (|X| =P0) (1 = PX(X = 3/2)
0) = <XP (|X|
3/2)0==1)P= (XP (X3/2)
= 1) +PP(X (X=1) 1) = 23 .
which follows from an application F (3/2)of integration
(1) Z= (( 32by )2 parts.
2) The0 = second
9
2 =moment
1 is
Then the definition of= Zexpectation Fgives 4 4.
1 1
p 2 p 22
We used 1 < 2 E[X 3/2] 
E[|X|] = = 30 ·whenx2(|X|
P f (x)we= dx 0)=+ 1 · Px(|X|
evaluated
2
3e=3x1)dx=
F·(3/2) F= 3 .9
(1). ,
1 0
(d)
(b) Since
p Applying F is continuous,
p formula (3.24): and it is di↵erentiable apart from finitely many points
(where
2 and the integral
3),X we can is calculated using two
just di↵erentiate it rounds
to get the of integration
probabilitybydensity parts. function:
Thus, the
variance is
E[|X|] = |k| P (X = k) = 1 · P (X = 1)p ( + 0 · P (X p = 0) + 1 · P (X = 1)
2x if 2 <✓x <◆2 3
k
f (x) = F 0 (x) 2 = 2 2 1 1
= 12 + 16Var(X)
= 23 . = E[X ] (E[X]) 0 =
otherwise.
9 3
= .
9
p p
We
3.11.
3.15.chose
(a)0(3.25)
By for have
We the wevalue
haveof f at 2 and 3, but the actual values are not important.
Z 1 Z 2
2 E[3X + 2] = 3E[X] 2 + 2 = 3 · 3 + 2 = 211.2 7
 3.14. The mean of the random variable X from Exercise 3.1 is
5
X 1 1 3 2 2 7
E[X] = kpX (k) = 1 · +2· +3· +4· +5· = .
7 14 14 7 7 2
k=1

The second moment is

5
X 1 1 3 2 2 197
E[X 2 ] = k 2 pX (k) = 12 · + 22 · + 32 · + 42 · + 52 · = .
7 14 14 7 7 14
k=1
3
Therefore, the variance is
Problem 2. Suppose a random variable X has ✓ ◆2 value E(X) = 3 and
2 2 197 expected
7 51
variance Var(X) = Var(X) = E[X the
4. Compute ] following
(E[X]) =quantities. = .
14 2 28
Now +
(a) E(3X let2)X be the random variable from Exercise 3.3. The mean is
Z 1 Z 1
(b) E(X 2 ) 1
2 E[X] = xf (x)dx = x · 3e 3x dx = ,
(c) E(2X + 3) 1 0 3
(d) Var(4X − 2)
which follows from an application of integration by parts. The second moment is
Z 1 Z 1
2
E[X 2 ] = x2 f (x) dx = x2 · 3e 3x dx = ,
1 0 9
where thetointegral
Solution Problemis calculated
2. using two rounds of integration by parts. Thus, the
variance is
✓ ◆2
2 1 1
Var(X) = E[X 2 ] (E[X])2 = = .
9 3 9
3.15. (a) We have
E[3X + 2] = 3E[X] + 2 = 3 · 3 + 2 = 11.
(b) We know that Var(X) = E[X 2 ] E[X]2 . Rearranging the terms gives
E[X 2 ] = Var(X) + E[X]2 = 4 + 32 = 13.
(c) Expanding the square gives
E[(2X + 3)2 ] = E[4X 2 + 12X + 9] = 4E[X 2 ] + 12E[X] + 9 = 4 · 13 + 12 · 3 + 9 = 97,
where we also used the result of part (b).
(d) We have Var(4X 2) = 42 Var(X) = 42 · 4 = 64.

Problem 3. There are five closed boxes on the table. Three of the boxes have
prizes inside while the other two are empty. You open the boxes one at a time
randomly until you find a prize. Let X be the number of boxes you open.

(a) Find the probability mass function of X.

(b) Fix E(X).
(c) Find Var(X).
(d) Suppose the prize inside each of the three boxes is $100, but each empty box
you open costs you $100. What is you expected gain or loss in this game?
(Express this gain or loss as a function of X.)

Solution to Problem 3.
Solutions to Chapter 3 69

and X
E[X 2 ] = k 2 P (X = k) = 1 · 2
5 +4· 1
5 +9· 1
5 + 16 · 1
5 = 31
5 .
k
This leads to
4
Var(X) = E[X 2 ] (E[X])2 = 34
25 .
3.28. (a) The possible values of X are 1, 2, and 3. Since there are three boxes with
nice prizes, we have
3
P (X = 1) = .
5
Next, for X = 2, we must first choose a box that does not have a good prize
(two choices) followed by one that does (three choices). Hence,
2·3 3
P (X = 2) = = .
5·4 10
Similarly,
2·1·3 1
P (X = 3) = = .
5·4·3 10
(b) The expectation is
3 3 1 3
E[X] = 1 · + 2 · +3· = .
5 10 10 2
(c) The second moment is
3 3 1 27
E[X 2 ] = 12 · + 22 · + 32 · = .
5 10 10 10
Hence, the variance is
✓ ◆2
2 2 27 3 9
Var(X) = E[X ] (E[X]) = = .
10 2 20
(d) Let W be the gain or loss in this game. Then
8
>
<100, if X = 1
W = 100(2 X) = 200 100X = 0, if X = 2
>
:
100, if X = 3.
Thus, by Fact 3.52,
3
E[W ] = E[200 100X] = 200 100E[X] = 200
= 50. 100 ·
2
3.29. The possible values of X are the possible class sizes: 17, 21, 24, 28. We can
compute the corresponding probabilities by computing the probability of choosing
a student from that class:
17 21 7 24 4 28 14
pX (17) = 90 , pX (21) = 90 = 30 , pX (24) = 90 = 15 , pX (28) = 90 = 45 .
From this we can compute E[X]:
X
17 7 4 14 209
E[X] = kP (X = k) = 17 · 90 + 21 · 30 + 24 · 15 + 28 · 45 = 9 .
k

For the variance

Problem we need
4. Choose E[X 2uniformly
a point ]: at random in the unit square (square of
X
side length
2 one). Let
E[X ] = 2 X be the distance
2 17 of point
k P (X = k) = 17 · 90 + 212 · 307chosen2 to4 the nearest
+ 24 · 15 + 282 · 14 edge of the
45 = 555.
square. k

(a) Find the cumulative distribution function of X.

(b) Find the probability density function of X.

Solution to Problem 4.
(b) Both cumulative distribution functions found in part (a) are continuous ev-
erywhere, and di↵erentiable everywhere apart from maybe a couple of points.
Thus we
point canthe
is in find fX andsector
circular fY bycorresponding
di↵erentiatingto FXthe
and FY : ⇡/2 and ↵ and
angles
(
radius 1. The area of this circular 4 sector
2x is ↵ + ⇡/2,
if 0  x < 1 while the area of the half
fX (x) = 3 3 ,
disk is ⇡. Thus
0, otherwise.
↵ + ⇡/2 1 arctan(s)
FS (s) = P (S81 s) = = + .
> 3 (4 2y) ,⇡
< if2 1  y <⇡2
(b) The c.d.f. found in fYpart
(y) =(a) is23 ,di↵erentiable everywhere,
if 0  y <hence
1 the p.d.f. is equal
>
: 5
to its derivative: 0, otherwise.
✓ ◆0
3.43. If (a, b) is a point in =
1 arctan(s)
the square 2 1
fS (s) + [0, 1] then=the distances . from the four sides
2 ⇡
are a, b, 1 a, 1 b and the minimal distance is the minimum ⇡(1 + s2 )of these four numbers.
Since Let
3.45. min(a,
(X,1Y ) a)be  the1/2, this minimal
uniformly chosen distance is atS most
point, then Y 1/2 (which can be
= X . We can disregard
achieved at (a, b) = (1/2, 1/2)), and at least
the case X = 0, as the probability of this is 0. 0. Thus the possible values of X are
from
(a) Wethe interval
need [0, 1/2].F (s) = P (S  s) for all s.
to compute
(a) We would like to compute F (x) = P (X  x) for all x. Because 0  X  1/2,
The slope S can be any nonnegative number, but it cannot be negative. Thus
we have F (x) = 0 for x < 0 and F (x) = 1 for x > 1/2.
FS (s) = P (S  s) = 0 if s < 0.
Denote the coordinates of the randomly chosen point by A and B. If 0  x 
If 0  s  1 then the points (x, y) 2 [0, 1]2 with y/x  s are exactly the points
1/2 then the set {X  x}c = {X > x} is the same as the set
in the triangle with vertices (0, 0), (1, 0), (1, s). The area of this triangle is s/2,
hence{xfor
< 0A,xs<11 we A, have
x < B, B} = {x < A < 1 x, x < B < 1 x}.
FS1(s) x=<s/2.
ThisIf is1 <
thes same as the
then the point
points (x, (A, 1]2 with
y) 2B)[0,being in the
y/x square
 s are(x, 1 x)
either in2thewhich has
triangle
2
probability
with vertices(1(0, 0),
2x)(1,. 0),
Hence, or0inthe
(1, 1)for x triangle
1/2 wewith
havevertices (0, 0), (1, 1), (1/s, 1).
1 12
The area Fof(x) the=union
P (X  ofx)
these
= 1 triangles
P (X > isx)1/2
= 1+ 2(1 (1 2x)1/s)
2 = 1
= 4x 4x 2s ,. hence for
1
1 < s we have FS (s) = 1 2s .
(b) Since the cumulative distribution function F (x) that we found in part (a) is
To summarize:
continuous, and it is di↵erentiable8apart from x = 0, we can find f (x) just by
di↵erentiating F (x). This means> 0 f (x)s =
<that < 04 8x for 0  x  1/2 and 0
otherwise. F (s) = 12 s 0<s1.
>
: 1
3.44. (a) Let s be a real number. Let 1 ↵2s 1<s
= arctan(r) 2 ( ⇡/2, ⇡/2) be the angle
corresponding to the slope s, this is the number
(b) Since F (s) is continuous everywhere and it is di↵erentiable ↵ 2 ( ⇡/2, ⇡/2) with
apart from tan(↵)
s ==0,
Problem
s. The5. A stick
event that of{Slength
 s} `isis broken
the same at
as a uniformly
the event chosen
that the
we can get the probability density function f (s) just by di↵erentiating F . This random
uniformly location.
chosen
We denote the length of the smaller piece by X.
gives
(a) Find the cumulative distribution 8function of X.
>
<0 ofs < 0
(b) Find the probability density function X.
f (s) = 12 0<s1.
>
: 1
Solution to Problem 5. 1<s
2s2

3.46. (a) The smaller piece cannot be larger than `/2, hence 0  X  `/2. Thus
FX (x) = 0 for x < 0 and FX (x) = 1 for x `/2.
For 0  x < `/2 the event {X  x} is the same as the event that the chosen
point where we break the stick in two is within x of one of the end points. The
set of possible locations is thus the union of two intervals of length x, hence
the probability of the uniformly chosen point to be in this set is 2·x
` . Hence for
0  x < `/2 we have FX (x) = 2x ` .
To summarize
8
>
<1 for x `/2
Solutions to Chapter 3 77
2x
FX (x) = ` for 0  x < `/2
>
:
0 for x < 0.
(b) The c.d.f. found in part (a) is continuous everywhere, and di↵erentiable apart
from x = `/2. Hence we can find the p.d.f. by di↵erentiating it, which gives
(
2
for 0  x < `/2
fX (x) = `
0 otherwise.

3.47. (a) We need to find F (x) = P (X  x) for all x. The X coordinate of a point
in the triangle must be between 0 and 30, so F (x) = 0 for x < 0 and F (x) = 1 for
x 30. 6. Let Z ∼ N (0, 1) and X ∼ N (µ, σ 2 ).
Problem
(a) Calculate < 304 ).
For 0  x E(Z then the set of points in the triangle with X  x is the triangle
with vertices (0, 0), (x, 0) and (x, 23 x). The area of this triangle is 13 x2 , while the
area of the original triangle is 20·30
2 = 300. This means that if 0  x < 30 then
1 2
3x x2
F (x) = 300 = 900 . Thus
8
>
<0 2 x<0
x
F (x) = 0  x < 30 .
> 900
:
1 x 30
 Z 1 Z 1
1 x2
E[Z 3 ] = x3 '(x)dx = x3 p e 2 dx.
1 1 2⇡
x2
Note that the function g(x) = x3 p12⇡ e 2 is odd: g(x) = g( x). Thus if the
integral is finite then it must be equal to 0, as the values on the positive and
negative half lines cancel each other out. The fact that the integral is finite follows
x2
from the fact that x3 grows a lot slower than e 2 . (Or you can evaluate the integral
on the positive and negative half lines separately by integration by parts.)
6
(b) We can express X as X = Y + µ where Y ⇠ N (0, 1). Then
3 4
E[X
(b) Calculate ] =) E[(
E(X Y + µ)3 ] =XE[as 3aY function
by expressing
3
+ 3 2 µY 2
+ 3 µ2 Y + µ 3 ]
of Z.
3
= E[Y 3 ] + 3 2
µE[Y 2 ] + 3 µ2 E[Y ] + µ3 .

We have E[Y
Solution ] = E[Y 3 ] =
to Problem
2
6.0 and E[Y ] = 1. Thus
E[X 3 ] = 3
E[Y 3 ] + 3 2
µE[Y 2 ] + 3 µ2 E[Y ] + µ3 = 3 2
µ + µ3
x2
3.68. (a) Since the p.d.f. of Z is '(x) = p12⇡ e 2 , we have
88 Z 1 Z 1
4 4 1 Solutions
x2
to Chapter 3
E[Z ] = '(x)x dx = p e 2 x4 dx.
1 1 2⇡
x2
We can evaluate the integral using integration by parts noting that e 2 x =
x2
( e 2 )0 :
Z 1 Z 1
1 x2 1 x2
p e 2 x4 dx = p e 2 x · x3 dx
1 2⇡ 1 2⇡
Z 1
1 x2 x=1 1 x2
= p ( e 2 ) · x3 x= 1 p ( e 2 ) · 3x2
2⇡ 1 2⇡
Z 1
1 x2
2
=3 p e 2 x dx = 3.
1 2⇡
x2 R1 x2
We used that lim e 2 x3 = 0 (and the same for x ! 1), and that 1 p12⇡ e 2 x2 dx =
x!1
E[Z 2 ] = 1.
Hence E[Z 4 ] = 3.
(b) We can express X as X = Y + µ where Y ⇠ N (0, 1). Then
E[X 4 ] = E[( Y + µ)4 ]
4
= E[ Y4+4 3
µY 3 + 6 2 2
µ Y 2 + 4 Y µ 3 + µ4 ]
4
= E[Y 4 ] + 4 3
µE[Y 3 ] + 6 2 2
µ E[Y 2 ] + 4 µ3 E[Y ] + µ4 .
We know that E[Y ] = 0, E[Y 2 ] = 1. By part (a) we have E[Y 4 ] = 3 and by
the previous problem we have E[Y 3 ] = 0. Substituting these in the previous
expression we get
E[X 4 ] = 3 4 + 6 2 µ2 + µ4 .
3.69. Denote the nth moment E[Z n ] by mn . It can be computed as
Z 1 Z 1
1 x2
mn = xn '(x)dx = xn p e 2 dx
1 1 2⇡
We have seen that m1 = E[Z] = 0 and m2 = E[Z 2 ] = 1.
Suppose first that n = 2k + 1 is an odd number. Then the function x2k+1 is
Problem 7. Let
odd and hence T be a random
the function x2k+1 '(x)variable
is odd aswith
well.density
If the
R integral is finite then 1
the contribution of the positive and negative half lines in 1 x2k+1 '(x)dx cancel

each other out and thus m2k+1 = 0. The fact
 1 e−that
x
3,
the integral is finite follows from
x ≥ 0, x2
n
the fact that for any fixed n xf (x)
grows
= 3a lot slower than e2.
For n = 2k 2 we see that xn '(x)
is0,
even, andelsewhere.
thus (if the integrals are finite)
we have Z Z
1 1
m2k = x2k '(x)dx = 2 x2k '(x)dx
(1) Find P (T > 3). 1 0
(2) Find P (1 ≤ T < 8). ⇣ x2
⌘0
Using integration by parts with the functions x 2k 1
and x'(x) = p1 e 2 =
(3) Find P (T > 4 | T > 1). 2⇡
( '(x))0 we get
Z 1to Problem 7.
Solution x=1 Z 1
1 x2
x2k '(x)dx = x2k 1
p e 2 + (2k 1)x2k 2
'(x)dx
0 2⇡ x=0 0
Z 1
= (2k 1) x2k 2 '(x)dx.
0
 x3 e x
dx = x3 e x
3x2 ( e x
)dx = 3x2 e x
dx.
0 x=0 0 0

x x=1
R1
Note that x3 e x=0
= 0 because lim x3 e x
= 0. To evaluate 0
3x2 e x
dx
x!1
we can integrate by parts twice more, or we can quote equation (4.18) from the
text to get
Z 1 Z
3 1 2 3 2 6
3x2 e x dx = x e x dx = · 2 = 3 .
0 0
3 6 7
Thus E[T ] = 3 .
1
4.13. The probability density function of T is fT (x) = 13 e 3x for x 0, and zero
1
otherwise. The cumulative distribution function is FT (x) = 1 e 3x for x 0,
and zero otherwise. From this we can compute
1
P (T > 3) = 1 FT (3) = e ,
1/3 8/3
P (1  T < 8) = FT (8) FT (1) = e e ,
P (T > 4 and T > 1) P (T > 4)
P (T > 4 | T > 1) = =
P (T > 1) P (T > 1)
4/3
1 FT (4) e 1
= = 1/3
=e .
1 FT (1) e
P (T > 4 | T > 1) can also be computed using the memoryless property of the
exponential:
1
P (T > 4 | T > 1) = P (T > 3) = 1 FT (3) = e .
4.14. (a) Denote the lifetime of the lightbulb by T . Since T is exponentially dis-
1
tributed with expected value 1000 we have T ⇠ Exp( ) with = 1000 . The
t
cumulative distribution function of T is then FT (t) = 1 e for t > 0 and 0
otherwise. Hence
2000· 2
P (T > 2000) = 1 P (T  2000) = 1 FT (2000) = e =e .
(b) We need to compute P (T > 2000|T > 500) where we used the notation of part
(a). By the memoryless property P (T > 2000|T > 500) = P (T > 1500). Using
the steps in part (a) we get
3
1500·
P (T > 1500) = 1 FT (1500) = e =e 2.

4.15. Let N be the Poisson process of arrival times of meteors. Let 11 PM corre-
spond to the origin on the time line.