WEBVTT

00:00:00.890 --> 00:00:03.020
In this problem, we'll be
looking at an ambulance that

00:00:03.020 --> 00:00:06.790
is traveling back and forth
in interval of size l.

00:00:06.790 --> 00:00:09.030
Say from 0 to l.

00:00:09.030 --> 00:00:11.900
At some point in time, there's
an accident occurring, let's

00:00:11.900 --> 00:00:14.070
say at location x.

00:00:14.070 --> 00:00:16.940
And we'll assume the accident
occurs in a random location so

00:00:16.940 --> 00:00:21.540
that x is uniformly distributed
between 0 and l.

00:00:21.540 --> 00:00:24.280
Now, at this point in time,
let's say the ambulance turns

00:00:24.280 --> 00:00:26.640
out to be at location y.

00:00:26.640 --> 00:00:29.860
Again, we'll assume that y is
a uniform random variable

00:00:29.860 --> 00:00:34.110
between 0 and l, and also that
x and y are independently

00:00:34.110 --> 00:00:35.570
distributed.

00:00:35.570 --> 00:00:39.100
The question we're interested
in answering is how long it

00:00:39.100 --> 00:00:42.760
would take an ambulance to
respond to travel from point y

00:00:42.760 --> 00:00:44.500
to point x.

00:00:44.500 --> 00:00:48.240
Let's call this time T. And in
particular, we want to know

00:00:48.240 --> 00:00:52.170
everything about distribution of
T. For example, what is the

00:00:52.170 --> 00:00:56.330
CDF of T given by the
probability of big T, less

00:00:56.330 --> 00:01:00.710
than or equal to little t, or
the PDF, which is done by

00:01:00.710 --> 00:01:03.680
differentiating the CDF
once we have it.

00:01:03.680 --> 00:01:08.850
Now, to start, we'll express T
you as a function of X and Y.

00:01:08.850 --> 00:01:10.450
Since we know that
the ambulance

00:01:10.450 --> 00:01:13.301
travels at a speed V--

00:01:13.301 --> 00:01:16.990
V meters or V units of
distance per second--

00:01:16.990 --> 00:01:21.420
then we can write that big T is
simply equal to Y minus X,

00:01:21.420 --> 00:01:25.160
absolute value the distance
between X and Y, divided by

00:01:25.160 --> 00:01:27.620
the speed at which the ambulance
is traveling at, V.

00:01:27.620 --> 00:01:32.910
So now if we look at the
probability of T less than or

00:01:32.910 --> 00:01:36.950
equal to little t, this is then
equal to the probability

00:01:36.950 --> 00:01:40.720
that Y minus X divided
by V less than or

00:01:40.720 --> 00:01:43.300
equal to little t.

00:01:43.300 --> 00:01:46.620
We now take off the absolute
value by writing the

00:01:46.620 --> 00:01:52.690
expression as negative vt less
equal to Y minus X less equal

00:01:52.690 --> 00:01:55.060
to positive vt.

00:01:55.060 --> 00:01:59.350
Here we multiply v on the other
side of t, and then took

00:01:59.350 --> 00:02:02.210
out the absolute value sign.

00:02:02.210 --> 00:02:06.280
As a final step, we'll also move
X to the other side of

00:02:06.280 --> 00:02:12.080
inequalities by writing this as
X minus vt less equal to y

00:02:12.080 --> 00:02:16.560
less equal to x plus vt.

00:02:16.560 --> 00:02:20.960
To compute this quantity, we'll
define a set A as a set

00:02:20.960 --> 00:02:24.870
of all points that satisfies
this condition right here.

00:02:24.870 --> 00:02:30.890
In particular, it's a pair of
all X and Y such that X minus

00:02:30.890 --> 00:02:37.580
vt less equal to little y less
equal to X plus vt, and also

00:02:37.580 --> 00:02:45.570
that X is within 0 and
l, and so is Y.

00:02:45.570 --> 00:02:48.420
So the set A will be the
set of values we'll

00:02:48.420 --> 00:02:50.040
be integrating over.

00:02:50.040 --> 00:02:53.620
Now that we have A, we can
express the above probability

00:02:53.620 --> 00:03:00.870
as the integral of all X and Y,
this pair within the set A,

00:03:00.870 --> 00:03:06.840
integrating the PDF of f of
X, Y, little x, little y.

00:03:12.100 --> 00:03:15.130
Let's now evaluate this
expression right here in a

00:03:15.130 --> 00:03:16.460
graphical way.

00:03:16.460 --> 00:03:19.710
On the right, we're plotting out
what we just illustrated

00:03:19.710 --> 00:03:23.860
here, where the shaded region is
precisely the set A. As we

00:03:23.860 --> 00:03:27.930
can see, this is a set of values
of X and Y where Y is

00:03:27.930 --> 00:03:33.150
sandwiched between two lines,
the upper one being X plus vt

00:03:33.150 --> 00:03:39.190
right here, and the lower line
being X minus vt, right here.

00:03:39.190 --> 00:03:45.310
So these are the values that
correspond to the set A.

00:03:45.310 --> 00:03:48.250
Now that we have A, let's
look f of x, y.

00:03:48.250 --> 00:03:51.840
We know that both x and y are
uniform random variables

00:03:51.840 --> 00:03:55.700
between 0 and l, and therefore,
since they're

00:03:55.700 --> 00:04:00.390
independent, the probability
density of x and y being at

00:04:00.390 --> 00:04:06.000
any point between 0 and l is
precisely 1 over l squared,

00:04:06.000 --> 00:04:11.810
where l squared is the size of
this square box right here.

00:04:11.810 --> 00:04:16.730
So given this picture, all we
need to do is to multiply by 1

00:04:16.730 --> 00:04:25.940
over l squared the area of the
region A. And depending on the

00:04:25.940 --> 00:04:27.640
value of T, we'll get different

00:04:27.640 --> 00:04:30.010
answers as right here.

00:04:30.010 --> 00:04:33.440
If T is less than 0, obviously,
the area of A

00:04:33.440 --> 00:04:35.880
diminishes to nothing,
so we get 0.

00:04:35.880 --> 00:04:40.180
If T is greater than l over V,
the area of A fills up the

00:04:40.180 --> 00:04:42.830
entire square, and we get 1.

00:04:42.830 --> 00:04:46.570
Now, if T is somewhere in
between 0 and l over v, we

00:04:46.570 --> 00:04:51.540
will have 1 over l squared,
multiply by the area looking

00:04:51.540 --> 00:04:55.440
like something like that right
here-- the shaded region.

00:04:55.440 --> 00:04:58.020
Now, if you wonder how we
arrive at exactly this

00:04:58.020 --> 00:05:01.170
expression right here, here is
a simple way to calculate it.

00:05:01.170 --> 00:05:05.820
What we want is 1 over l squared
times the area A. Now,

00:05:05.820 --> 00:05:12.320
area A can be viewed as the
entire square, l squared,

00:05:12.320 --> 00:05:16.810
minus whatever's not in area
A, which is these two

00:05:16.810 --> 00:05:19.990
triangles right here.

00:05:19.990 --> 00:05:27.090
Now, each triangle has area
1/2, l minus vt squared.

00:05:27.090 --> 00:05:30.830
This multiply 2, and this, after
some algebra, will give

00:05:30.830 --> 00:05:33.270
the answer right here.

00:05:33.270 --> 00:05:35.660
At this point, we have obtained
the probability of

00:05:35.660 --> 00:05:37.470
big T less equal to little t.

00:05:37.470 --> 00:05:43.425
Namely, we have gotten the CDF
for T. And as a final step, we

00:05:43.425 --> 00:05:46.430
can also compute the probability
density function

00:05:46.430 --> 00:05:50.720
for T. We'll call it
little f of t.

00:05:50.720 --> 00:05:54.580
And we do so by simply
differentiating the CDF in

00:05:54.580 --> 00:05:57.270
different regions of T.

00:05:57.270 --> 00:06:01.670
To begin, we'll look at t
between 0 and l over v right

00:06:01.670 --> 00:06:06.210
here at differentiating the
expression right here with

00:06:06.210 --> 00:06:07.410
respect to t.

00:06:07.410 --> 00:06:12.660
And doing so will give
us 2v over l minus 2v

00:06:12.660 --> 00:06:16.350
squared t over L squared.

00:06:16.350 --> 00:06:22.350
And this applies to t greater or
equal to 0, less than l/v.

00:06:22.350 --> 00:06:26.680
Now, any other region, either
t less than 0 or t greater

00:06:26.680 --> 00:06:32.860
than l/v, we have a constant
for the CDF, and hence its

00:06:32.860 --> 00:06:34.770
derivative will be 0.

00:06:34.770 --> 00:06:37.246
So this is for any other t.

00:06:37.246 --> 00:06:39.430
We call it otherwise.

00:06:39.430 --> 00:06:42.620
Now, this completely
characterized the PDF of big

00:06:42.620 --> 00:06:45.270
T, and hence, we've also
finished a problem.