Você está na página 1de 6

Summation Notation Tutorial

Throughout this handout, a and b are real-valued constants, x, y and u are real-valued variables,
and n and m are integers.
1 Introduction
Suppose you want to nd the average (mean) of a list of numbers, for example: 3, 5, 2, 7, 8. You
would rst add the numbers together, and then divide by ve, which is the number of quantities.
One way to write this would be:
3 + 5 + 2 + 7 + 8
5
=
25
5
= 5
Now, suppose you had a very long list of numbers. It would be useful to have some notation so
that you could write an expression for the average without having to write out all of the individual
numbers. We can do that using subscripts to identify the individual numbers and convenient
mathematical symbology called summation notation.
2 Using subscripts and -notation
Continuing with the example above, lets label the number of quantities n, so that in this case
n = 5. Next, label the rst number x
1
, the second number x
2
, and so on, so that with the list of
numbers we have, x
1
= 3, x
2
= 5, x
3
= 2, x
4
= 7, and x
5
= 8. We can call the last number in the
list x
n
, since there are ve numbers in the list and n = 5. In this example, x
n
= x
5
= 8.
Subscripts give us an easy way of referring to any single quantity in the list. For instance, x
3
is
short-hand notation for the third element in the list. Now, if we used subscripts to write out the
expression for the numerator in the average, we would have x
1
+ x
2
+ x
3
+ x
4
+ x
5
, or we might
write this as x
1
+ x
2
+ + x
n
, but this is still a bit long to write out. A shorter form of notation
for this same expression using the summation sign () would be:
5

i=1
x
i
=
n

i=1
x
i
where n = 5. This can be read as the sumof all x
i
for i = 1 through i = n. In this case, it would be
the sum of the numbers from x
1
through x
n
= x
5
, which is the sum of all of the numbers in the list.
2.1 Example
Let n = 3, with x
1
= 7, x
2
= 8, and x
3
= 13.
Then,
n

i=1
x
i
=
3

i=1
x
i
= x
1
+ x
2
+ x
3
= 7 + (8) + 13 = 12.
To calculate the average of these three numbers we would take the sumof all three numbers divided
by three. If we use summation notation, we would write:
average =
1
3
3

i=1
x
i
=
x
1
+ x
2
+ x
3
3
=
7 + (8) + 13
3
= 12/3 = 4.
Stat 23400 Summation Notation (January 9, 2014, 9:44am) 1 of 6
2.2 Common statistical formulas that use summations
You probably recognize that in the examples so far, we have been working with the formula for
the sample average (the quantities come from a sample of a population), which can be written like
this using summation notation:
sample average =
1
n
n

i=1
x
i
= x = x-bar. (1)
Compare this to the formula we used in Example 2.1, where n = 3. The general formula in equa-
tion (1) will work for any list of numbers. The number n represents the number of quantities in
the list, and the terms x
i
represent the actual numbers (where i = 1, 2, 3, . . . , n).
The formula for the sample variance can also be written using summation notation:
sample variance =
1
n 1
n

i=1
(x
i
x)
2
= s
2
= s-squared. (2)
Yes, n 1, explained in class.) The sample standard deviation is s =

s
2
. Lets look at an
example.
2.2.1 Example
Suppose we wanted to calculate the sample variance for the list of numbers from Example 2.1.
Lets think about what equation (2) is telling us. We can read the summation part of the formula
as the sum of the terms (x
i
x)
2
for i = 1 to i = n. So we can think of expanding out the terms
in the numerator like this:
n

i=1
(x
i
x)
2
= (x
1
x)
2
+ (x
2
x)
2
+ + (x
n
x)
2
If we substitute this into the formula for the sample variance we get this:
s
2
=
1
n 1
n

i=1
(x
i
x)
2
=
(x
1
x)
2
+ (x
2
x)
2
+ + (x
n
x)
2
n 1
In Example 2.1 we found that the average of the numbers is x = 4, and there are n = 3 numbers in
the list. If we substitute x = 4 and n = 3 into the expanded formula for the sample variance we get:
s
2
=
(x
1
4)
2
+ (x
2
4)
2
+ (x
3
4)
2
3 1
and then we can plug in the specic values, x
1
= 7, x
2
= 8, and x
3
= 13, to get:
s
2
=
(7 4)
2
+ ((8) 4)
2
+ (13 4)
2
2
=
(3)
2
+ (12)
2
+ (9)
2
2
=
9 + 144 + 81
2
=
234
2
= 117
So the sample standard deviation of the list of numbers in Example 2.1 is s =

s
2
=

117 = 10.82.
Stat 23400 Summation Notation (January 9, 2014, 9:44am) 2 of 6
3 Useful rules for working with summations
We present four rules that are helpful for simplifying and working with summation notation.
We have given these rules informal names that we hope will help you to remember them.
3.1 A Constant Rule
n

i=1
a = n a
This rule says that if we have a summation of a constant (a number that does not change with the
index i), then the result will just be that number times n. Why is this true? Lets use a = 7 and
n = 5. Then we would get:
n

i=1
a =
5

i=1
7 = 7 + 7 + 7 + 7 + 7 = 5 7 = 35 .
3.2 A Multiplication Rule
n

i=1
(a x
i
) = a
n

i=1
x
i
Why is this true? Lets work an example with specic numbers. Suppose we have x
1
= 3, x
2
= 5,
and x
3
= 1, and we want to calculate

3
i=1
(5 x
i
). That would be the sum of the three terms
that each look like this: (5 x
i
) for i = 1 to i = 3.
3

i=1
(5 x
i
) = (5 x
1
) + (5 x
2
) + (5 x
3
) = (5 3) + (5 5) + (5 (1))
= 5 (3 + 5 + (1)) = 5 (x
1
+ x
2
+ x
3
) = 5
3

i=1
x
i
This is quite handy. Whenever the terms in a summation are multiplied by a constant, we can
pull that constant outside of the summation.
3.3 An Addition Rule
n

i=1
(x
i
+ y
i
) =
n

i=1
x
i
+
n

i=1
y
i
Whenever there is an addition or subtraction sign in the summation, we can break the equation
apart into separate summations.
Stat 23400 Summation Notation (January 9, 2014, 9:44am) 3 of 6
To see that this is true, suppose you have two lists of numbers (both must be the same length).
The rst list of numbers is x
1
= 3, x
2
= 5, x
3
= 1, and the second list is y
1
= 7, y
2
= 4, and
y
3
= 2. Then,
3

i=1
(x
i
+ y
i
) = (x
1
+ y
1
) + (x
2
+ y
2
) + (x
3
+ y
3
) = (3 + 7) + (5 + 4) + ((1) + 2).
Since we are just adding a list of numbers together, we know from basic algebra that we can change
the order of addition without changing the result (the commutative property of addition). So, we
could rewrite the last sum above as (3 + 5 + (1)) + (7 + 4 + 2), which is just the sum of all of
the numbers in the rst list plus the sum of all of the numbers in the second list. If we write this
all out carefully using summation notation we get:
3

i=1
(x
i
+ y
i
) = (3+7)+(5+4)+((1)+2) = (3+5+(1))+(7+4+2) =
3

i=1
x
i
+
3

i=1
y
i
.
The addition rule only works for additive functions. That is, the rule works for functions of the
form f(x
i
) = g(x
i
) + h(x
i
) + p(x
i
). In fact, a more general form for the addition rule is this:
n

i=1
[g(x
i
) + h(x
i
) + p(y
i
)] =
n

i=1
g(x
i
) +
n

i=1
h(x
i
) +
n

i=1
p(y
i
)
Caution:
n

i=1
(a + x
i
)
2
=
n

i=1
a
2
+
n

i=1
x
2
i
.
However, we can still use the addition rule here as follows:
n

i=1
(a + x
i
)
2
=
n

i=1
(a
2
+ 2ax
i
+ x
2
i
) =
n

i=1
a
2
+
n

i=1
2ax
i
+
n

i=1
x
2
i
=
n

i=1
a
2
+
n

i=1
x
2
i
.
3.4 A Derivative Rule
d
du
[
n

i=1
f
i
(u)
]
=
n

i=1
[
d
du
f
i
(u)
]
(when n < or

i=1
f
i
(u) < )
In this course, we will study linear models and the theory of least squares estimates. It will be
important to know how to take the derivative of a function of this form (with a summation). As
long as there are a nite number of terms in the summation or the innite sum is nite (for this
course, one of these two situations will always be the case), we can move a derivative operation
into the summation.
Stat 23400 Summation Notation (January 9, 2014, 9:44am) 4 of 6
Here is an example: Suppose you were asked to show that
d
du
[
n

i=1
(ux
2
i
+ u
2
x
i
)
]
= =
n

i=1
x
2
i
+ 2u
n

i=1
x
i
.
Using the derivative rule along with this fact from calculus:
d
du
(ux
2
i
+ u
2
x
i
) = (x
2
i
+ 2ux
i
),
d
du
[
n

i=1
(ux
2
i
+ u
2
x
i
)
]
=
n

i=1
d
du
(
ux
2
i
+ u
2
x
i
)
=
n

i=1
(x
2
i
+ 2ux
i
)
and by the addition and multiplication rules, we have that:
n

i=1
(x
2
i
+ 2ux
i
) =
n

i=1
x
2
i
+
n

i=1
2ux
i
=
n

i=1
x
2
i
+ 2u
n

i=1
x
i
,
and this seems to be all of the algebraic simplication that we can perform.
Of course, this is a statistics course and not a mathematics course, so if x
1
, x
2
, . . . , x
n
are a sample
of observations from a population, we can show that
n

i=1
x
i
= nx and
n

i=1
x
2
i
= (n 1)s
2
+ nx
2
(3)
by using equations (1) and (2) and a bit of algebra using summation notation (try this on your own).
From this, we can show that
d
du
[
n

i=1
(ux
2
i
+ u
2
x
i
)
]
=
n

i=1
x
2
i
+2u
n

i=1
x
i
= (n1)s
2
+nx
2
+2unx = (n1)s
2
+nx(x+2u) .
This seems to be all of the statistical simplication that we can perform.
4 Homework Exercises
On the next page are homework exercises that utilize the rules presented above. These exercises
are intended to be representative of the diculty level of the mathematics involving summations
that you will encounter in this course. You should try to master the steps in these problems in
order to avoid having diculty understanding proofs that make use of summation notation
throughout the course.
Just for this homework be sure to identify which rule(s) you are using at each step, if any. (Its
not necessary to justify algebra steps for the rest of the course.)
Solutions will be posted on the course Chalk site after any assignments based on these exercises
have been completed.
Stat 23400 Summation Notation (January 9, 2014, 9:44am) 5 of 6
1. Show that
n

i=1
(ax
i
b)
2
= a
2
(n 1)s
2
+ anx(ax 2b) + nb
2
by expanding the sum on the left-hand side into its individual terms, then rearranging and
collecting terms together appropriately to obtain the sum on the right hand side.
2. Provide a counterexample to show that, generally
n

i=1
x
2
i
=
[
n

i=1
x
i
]
2
.
That is, for something like n = 3, choose some actual numbers for x
1
, x
2
and x
3
such that the
two sides of the equation are not equal.
3. Find the derivative of f(a) =
n

i=1
(ax
i
b)
2
with respect to a.
4. Minimize f(a) =
n

i=1
(ax
i
b)
2
with respect to a.
Be sure to conrm that the extrema that you nd is indeed a minimum (not a maximum).
5. Show that
n

i=1
m

j=1
(x
i
a)(y
j
b) =
[
n

i=1
x
i
][
m

j=1
y
j
]
mb
n

i=1
x
i
na
m

j=1
y
j
+ (mn)(ab).
If this seems complicated, rst use n = 3 and m = 2 to limit the number of individual terms
that you need to write down in order to get comfortable working with double sums, then go
back and work through the general case for any n and m.
Stat 23400 Summation Notation (January 9, 2014, 9:44am) 6 of 6

Você também pode gostar