WEBVTT

00:00:00.620 --> 00:00:03.520
We are defining a random
variable as a real valued

00:00:03.520 --> 00:00:05.430
function on the sample space.

00:00:05.430 --> 00:00:07.960
So this is a good occasion to
make sure that we understand

00:00:07.960 --> 00:00:10.130
what a function is.

00:00:10.130 --> 00:00:13.670
To define a function, we
start with two sets.

00:00:13.670 --> 00:00:14.740
One set--

00:00:14.740 --> 00:00:16.160
call it A--

00:00:16.160 --> 00:00:20.420
is the domain of the function.

00:00:20.420 --> 00:00:23.520
And we have our second set.

00:00:23.520 --> 00:00:29.800
Then a function is a rule that
for any element of A

00:00:29.800 --> 00:00:36.940
associates an element of B. And
we use a notation of this

00:00:36.940 --> 00:00:41.700
kind to indicate that we are
dealing with a function f that

00:00:41.700 --> 00:00:47.470
maps elements of A into
elements of B.

00:00:47.470 --> 00:00:52.050
Now, two elements of A may be
mapped to the same element of

00:00:52.050 --> 00:00:55.640
B. This is allowed.

00:00:55.640 --> 00:01:00.320
What is important, however, is
that every element of A is

00:01:00.320 --> 00:01:04.810
mapped to exactly one element
of B, not more.

00:01:04.810 --> 00:01:08.590
But it is also possible that
we have some elements of B

00:01:08.590 --> 00:01:13.050
that do not correspond to any
of the elements of A.

00:01:13.050 --> 00:01:16.770
Now, I said that a function is
a rule that assigns points of

00:01:16.770 --> 00:01:22.240
A to points in B. But what
exactly do we mean by a rule?

00:01:22.240 --> 00:01:25.410
If we want to be more precise,
a function would

00:01:25.410 --> 00:01:27.270
be defined as follows.

00:01:27.270 --> 00:01:32.690
It would be defined as a
set of pairs of values.

00:01:32.690 --> 00:01:39.560
It would be a set of pairs of
the form x, y such that x is

00:01:39.560 --> 00:01:45.890
always an element of A, y is
always an element of B, and

00:01:45.890 --> 00:01:47.740
also-- most important--

00:01:47.740 --> 00:01:54.259
each x in A appears in
exactly one pair.

00:02:00.030 --> 00:02:02.890
So this would be a formal
definition of

00:02:02.890 --> 00:02:04.330
what a function is.

00:02:04.330 --> 00:02:09.000
It is collection of ordered
pairs of this kind.

00:02:09.000 --> 00:02:14.590
As a concrete example, let us
start with the set consisting

00:02:14.590 --> 00:02:18.880
of these elements here.

00:02:18.880 --> 00:02:24.079
And let B be the set
of real numbers.

00:02:24.079 --> 00:02:29.100
And consider the function that
corresponds to what we usually

00:02:29.100 --> 00:02:30.610
call the square.

00:02:30.610 --> 00:02:34.320
So it's a function that
squares its argument.

00:02:34.320 --> 00:02:38.030
Then this function would be
represented by the following

00:02:38.030 --> 00:02:41.710
collection of pairs.

00:02:41.710 --> 00:02:45.690
So this is the value of x.

00:02:45.690 --> 00:02:48.820
And this is the corresponding
value of y.

00:02:48.820 --> 00:02:54.130
Any particular x shows
up just once in this

00:02:54.130 --> 00:02:56.890
collection of pairs.

00:02:56.890 --> 00:02:58.250
But a certain y--

00:02:58.250 --> 00:03:00.210
for example, y equal to 1--

00:03:00.210 --> 00:03:04.960
shows up twice, because minus
1 and plus 1 both map to the

00:03:04.960 --> 00:03:08.360
same element of B.

00:03:08.360 --> 00:03:13.030
Now, this is a representation
in terms of ordered pairs.

00:03:13.030 --> 00:03:16.070
But we could also think of
the function as being

00:03:16.070 --> 00:03:18.770
described by a table.

00:03:18.770 --> 00:03:25.340
We could, for instance, put
this information here in a

00:03:25.340 --> 00:03:30.320
form of a table of this kind
and say that this table

00:03:30.320 --> 00:03:31.560
describes the function.

00:03:31.560 --> 00:03:33.800
For any element x, it
tells us what the

00:03:33.800 --> 00:03:36.710
corresponding element y is.

00:03:36.710 --> 00:03:40.450
However, when the set A is an
infinite set it is not clear

00:03:40.450 --> 00:03:44.640
what we might mean by saying
a table, an infinite table,

00:03:44.640 --> 00:03:46.480
whereas this definition
in terms of

00:03:46.480 --> 00:03:49.310
ordered pairs still applies.

00:03:49.310 --> 00:03:53.470
For example, if you're
interested in the function

00:03:53.470 --> 00:03:56.490
which is, again, the square
function from the real

00:03:56.490 --> 00:04:00.790
numbers, the way you would
specify that function

00:04:00.790 --> 00:04:03.540
abstractly would
be as follows.

00:04:03.540 --> 00:04:13.480
You could write, it's the set of
all pairs of this form such

00:04:13.480 --> 00:04:18.420
that x is a real number.

00:04:18.420 --> 00:04:23.530
And now such pairs, of course,
belong to the two dimensional

00:04:23.530 --> 00:04:26.350
plane because it's a
pair of numbers.

00:04:26.350 --> 00:04:29.880
So this set here can be viewed
as a formal definition or a

00:04:29.880 --> 00:04:32.960
specification of the
squaring function.

00:04:32.960 --> 00:04:36.050
Now, what this set is
is something that we

00:04:36.050 --> 00:04:37.540
can actually plot.

00:04:37.540 --> 00:04:40.670
If we go in the two dimensional
plane, the points

00:04:40.670 --> 00:04:44.980
of this form are exactly the
points that belong to the

00:04:44.980 --> 00:04:47.810
graph of the square function.

00:04:47.810 --> 00:04:51.070
So this abstract definition,
really all that it says is

00:04:51.070 --> 00:04:53.310
that a function is the
same thing as the

00:04:53.310 --> 00:04:55.550
plot of that function.

00:04:55.550 --> 00:04:58.810
But it's important here
to make a distinction.

00:04:58.810 --> 00:05:02.720
The function is the
entire plot--

00:05:02.720 --> 00:05:06.270
so this set here is
the function f--

00:05:06.270 --> 00:05:10.400
whereas if I tell you a
specific number x, the

00:05:10.400 --> 00:05:14.240
corresponding value here
would be f of x.

00:05:14.240 --> 00:05:19.370
So here x is a number and
f of x is also a number.

00:05:19.370 --> 00:05:23.510
And those two values, x and f
of x, define this particular

00:05:23.510 --> 00:05:25.490
point on this plot.

00:05:25.490 --> 00:05:30.141
But the function itself
is the entire plot.

00:05:30.141 --> 00:05:34.159
Let us also take this occasion
to talk a little bit about the

00:05:34.159 --> 00:05:38.430
notation and the proper way of
talking about functions.

00:05:38.430 --> 00:05:41.510
Here is the most common
way that one

00:05:41.510 --> 00:05:43.890
would describe a function.

00:05:43.890 --> 00:05:46.320
And this is an appropriate
way.

00:05:46.320 --> 00:05:48.520
We've described the domain.

00:05:48.520 --> 00:05:50.780
We've described the
set on which the

00:05:50.780 --> 00:05:52.480
function takes values.

00:05:52.480 --> 00:05:56.710
And I'm telling you for any x in
that set what the value of

00:05:56.710 --> 00:05:58.596
the function is.

00:05:58.596 --> 00:06:02.060
On the other hand, sometimes
people use a more loose

00:06:02.060 --> 00:06:05.190
language, such as for example,
they would say,

00:06:05.190 --> 00:06:07.250
the function x squared.

00:06:07.250 --> 00:06:08.420
What does that mean?

00:06:08.420 --> 00:06:13.700
Well, what this means is
exactly this statement.

00:06:13.700 --> 00:06:16.310
Now let us consider
this function.

00:06:16.310 --> 00:06:18.850
The function f--

00:06:18.850 --> 00:06:20.660
again, from the reals
to the reals--

00:06:20.660 --> 00:06:24.440
that's defined by f of
z equal to z squared.

00:06:24.440 --> 00:06:27.530
Is this a different function
or is it the same function?

00:06:27.530 --> 00:06:31.870
It's actually the same function,
because these two

00:06:31.870 --> 00:06:33.710
involve the same sets.

00:06:33.710 --> 00:06:37.680
And they produce their outputs,
the values of f,

00:06:37.680 --> 00:06:40.010
using exactly the same rule.

00:06:40.010 --> 00:06:43.480
They take an argument and they
square that argument.

00:06:43.480 --> 00:06:46.860
Now, if you were to use informal
notation, you would

00:06:46.860 --> 00:06:49.430
be referring to that
second function as

00:06:49.430 --> 00:06:52.120
the function z squared.

00:06:52.120 --> 00:06:55.420
And now, if you use informal
language, it's less clear that

00:06:55.420 --> 00:06:58.750
the function x squared and the
function z squared are one and

00:06:58.750 --> 00:07:02.740
the same thing, whereas with
this terminology here, it

00:07:02.740 --> 00:07:05.110
would be pretty clear
that we're talking

00:07:05.110 --> 00:07:07.460
about the same function.

00:07:07.460 --> 00:07:10.960
Finally, suppose that we have
already defined a function.

00:07:10.960 --> 00:07:14.870
How should we refer
to it in general?

00:07:14.870 --> 00:07:18.620
Should we call it the function
f, or should we say the

00:07:18.620 --> 00:07:20.780
function f of x?

00:07:20.780 --> 00:07:24.840
Well, when x is a number,
f of x is also a number.

00:07:24.840 --> 00:07:27.760
So f of x is not really
a function.

00:07:27.760 --> 00:07:30.270
The appropriate language
is this one.

00:07:30.270 --> 00:07:34.400
We talk about the function f,
although quite often, people

00:07:34.400 --> 00:07:37.590
will abuse language and they
will use this terminology.

00:07:37.590 --> 00:07:41.800
But it's important to keep in
mind what we really mean.

00:07:41.800 --> 00:07:45.970
The idea is that we need to
think of a function as some

00:07:45.970 --> 00:07:51.520
kind of box or even a computer
program, if you wish, that

00:07:51.520 --> 00:07:55.640
takes inputs and produces
outputs.

00:07:55.640 --> 00:07:59.770
And there's a distinction
between f, which is the box,

00:07:59.770 --> 00:08:04.310
from the value f of x that the
function takes if we feed it

00:08:04.310 --> 00:08:05.560
with a specific argument.