1
00:00:05,140 --> 00:00:08,210
In this second of the videos on
matrix multiplication, we're

2
00:00:08,210 --> 00:00:11,280
going to delve a little bit more
deeply into matrix

3
00:00:11,280 --> 00:00:14,350
multiplication and look at some
of the properties and the

4
00:00:14,350 --> 00:00:17,113
conditions under which different
sorts of multiplication can be

5
00:00:17,113 --> 00:00:20,490
carried out. Let's start by
looking at looking at a specific

6
00:00:20,490 --> 00:00:23,560
example in this example. Here
I've written down to matrices

7
00:00:23,560 --> 00:00:28,791
M&N. And let's look at the sizes
of these matrices. The first

8
00:00:28,791 --> 00:00:33,392
matrix M. Is a three row
three column matrix, so

9
00:00:33,392 --> 00:00:34,848
it's three by three.

10
00:00:37,710 --> 00:00:42,070
And the second matrix N is 3
rows, two columns.

11
00:00:42,900 --> 00:00:44,250
So it's a 3 by 2.

12
00:00:46,920 --> 00:00:49,152
And we notice that these
numbers are the same.

13
00:00:50,920 --> 00:00:54,261
The number of columns in the
first is the same as the number

14
00:00:54,261 --> 00:00:57,088
of rows in the second, so we can
perform this matrix

15
00:00:57,088 --> 00:01:00,172
multiplication and the size of
the answer will be a three by

16
00:01:00,172 --> 00:01:03,513
two matrix. So right at the
start we know the size of the

17
00:01:03,513 --> 00:01:06,597
answer. It's going to have three
rows and two columns just like

18
00:01:06,597 --> 00:01:09,788
this one had. So the shape
of the answer is.

19
00:01:10,910 --> 00:01:13,396
Like we have here and we're
looking for these 6 numbers.

20
00:01:13,980 --> 00:01:14,760
In the product.

21
00:01:16,510 --> 00:01:17,818
Let's try and work it out.

22
00:01:18,780 --> 00:01:23,239
To find the number that's in the
first row, first column. We work

23
00:01:23,239 --> 00:01:27,698
with the first row of the first
matrix and the first column of

24
00:01:27,698 --> 00:01:28,727
the Second Matrix.

25
00:01:29,510 --> 00:01:33,570
What we want is 3
* 1 which is 3.

26
00:01:34,580 --> 00:01:41,116
2 * -- 2 which is minus 4 and 1
* 3, which is 3. So we've got 3

27
00:01:41,116 --> 00:01:42,148
ones or three.

28
00:01:43,730 --> 00:01:50,322
2 * -- 2 is minus 4 and 1
* 3 is 3. We multiply the

29
00:01:50,322 --> 00:01:53,206
paired elements together and
add the result.

30
00:01:56,590 --> 00:02:00,776
When we come to the first row,
second column, we work with the

31
00:02:00,776 --> 00:02:02,708
first row here and the second

32
00:02:02,708 --> 00:02:08,880
column here. And again, pairing
off 3 * -- 2 is minus 6.

33
00:02:10,400 --> 00:02:12,188
2 * 3.

34
00:02:13,030 --> 00:02:13,960
Is 6.

35
00:02:15,470 --> 00:02:18,270
1 * -- 4 is minus 4.

36
00:02:19,030 --> 00:02:21,613
So in each case, we're
multiplying the paired elements

37
00:02:21,613 --> 00:02:23,048
together and adding the results.

38
00:02:24,750 --> 00:02:28,936
When we want the element that's
going in here, which is in the

39
00:02:28,936 --> 00:02:32,800
2nd row first column of the
answer, we work with the 2nd

40
00:02:32,800 --> 00:02:36,664
row, first column of the given
matrices 4 * 1 is 4.

41
00:02:38,220 --> 00:02:41,706
Minus 3 * -- 2 is +6.

42
00:02:43,760 --> 00:02:45,510
2 * 3 is 6.

43
00:02:47,600 --> 00:02:52,033
And continuing in the same way,
the answer that goes in the 2nd

44
00:02:52,033 --> 00:02:55,443
row, second column comes from
taking the 2nd row, second

45
00:02:55,443 --> 00:02:58,846
column. 4 * -- 2 is minus 8.

46
00:03:00,090 --> 00:03:03,498
Minus 3 * + 3 is minus 9.

47
00:03:04,810 --> 00:03:07,400
2 * -- 4 is minus 8.

48
00:03:09,530 --> 00:03:13,886
And finally on the last row
to find the element in the

49
00:03:13,886 --> 00:03:17,879
first row. Sorry the 3rd row
first column will work with

50
00:03:17,879 --> 00:03:21,146
the 3rd row, First Column, 5
ones of five.

51
00:03:23,170 --> 00:03:25,375
4 * -- 2 is minus 8.

52
00:03:26,410 --> 00:03:28,120
3 * 3 is 9.

53
00:03:29,810 --> 00:03:33,562
And similarly to find the last
element, it will be 5 * -- 2,

54
00:03:33,562 --> 00:03:34,634
which is minus 10.

55
00:03:35,260 --> 00:03:40,746
4 * 3 is 12 and 3 * --
4 is minus 12.

56
00:03:42,360 --> 00:03:46,404
And if we just tidy up what
we've got, we'll have 336

57
00:03:46,404 --> 00:03:48,089
subtract 4, which is 2.

58
00:03:50,120 --> 00:03:53,216
Minus 6 + 6 zero subtract 4 is

59
00:03:53,216 --> 00:03:57,269
minus 4. Four and
six is 10 and 616.

60
00:03:58,610 --> 00:04:02,912
Minus 8 -- 9 --
8 is minus 25.

61
00:04:05,450 --> 00:04:07,810
5 subtract 8 + 9.

62
00:04:08,460 --> 00:04:09,530
6th

63
00:04:10,680 --> 00:04:13,452
and minus 10 + 12 -- 12 is minus

64
00:04:13,452 --> 00:04:17,439
10. And this is the result
of multiplying these two

65
00:04:17,439 --> 00:04:18,041
matrices together.

66
00:04:20,560 --> 00:04:23,368
What about if we try and
multiply the two

67
00:04:23,368 --> 00:04:25,240
matrices together the
opposite way round?

68
00:04:25,240 --> 00:04:27,736
Suppose we try and
workout N * M.

69
00:04:45,730 --> 00:04:50,644
Now, in this case the size of
the first matrix here is 3 rows

70
00:04:50,644 --> 00:04:55,558
and two columns, so that's a
three by two and the size of the

71
00:04:55,558 --> 00:04:59,068
second matrix is 3 by 3, three
rows, three columns.

72
00:04:59,890 --> 00:05:03,102
And what we observe now is
that these two numbers here

73
00:05:03,102 --> 00:05:06,606
are not the same, they are not
equal. That means that we

74
00:05:06,606 --> 00:05:08,942
cannot do the matrix
multiplication in the order

75
00:05:08,942 --> 00:05:11,570
that I've written it down
here. That matrix product

76
00:05:11,570 --> 00:05:15,074
doesn't exist. So this is the
first point. I'd like to make

77
00:05:15,074 --> 00:05:18,286
that even when you can find a
matrix product by multiplying

78
00:05:18,286 --> 00:05:20,914
two matrices together, it
matters very much. The order

79
00:05:20,914 --> 00:05:24,418
in which you write them down.
It may be possible to workout

80
00:05:24,418 --> 00:05:27,630
a product one way, but not
another way. Let's look at

81
00:05:27,630 --> 00:05:28,506
some more examples.

82
00:05:29,570 --> 00:05:33,123
Suppose we've got two matrices
C&D as I've written them down

83
00:05:33,123 --> 00:05:37,322
here, I'm going to try to work
out the product C * D.

84
00:05:38,120 --> 00:05:41,396
And I'll also try and workout
the product D times. See if

85
00:05:41,396 --> 00:05:42,488
either of these exist.

86
00:05:43,970 --> 00:05:48,390
But in the first case, we've got
a two row three column matrix.

87
00:05:49,280 --> 00:05:52,210
And in the second example
here, within the Second matrix

88
00:05:52,210 --> 00:05:56,019
here we've got three rows into
two columns, so we can in fact

89
00:05:56,019 --> 00:05:59,242
work this product out because
these numbers are the same and

90
00:05:59,242 --> 00:06:03,051
the result will be a two by
two matrix. So the shape of

91
00:06:03,051 --> 00:06:05,688
the answer will be 2 rows and
two columns.

92
00:06:08,020 --> 00:06:12,610
If we try and do this the other
way round, D * C, The first

93
00:06:12,610 --> 00:06:16,588
matrix Now has got three rows
and two columns. It's a three by

94
00:06:16,588 --> 00:06:20,260
two matrix and the second one's
got two rows and three columns.

95
00:06:20,260 --> 00:06:22,096
It's a two by three matrix.

96
00:06:22,690 --> 00:06:26,092
So you can. You can see that we
can still work it out because

97
00:06:26,092 --> 00:06:29,251
these two numbers are still the
same 2 into the same, but this

98
00:06:29,251 --> 00:06:32,653
time the result is going to be a
three by three matrix, so it's

99
00:06:32,653 --> 00:06:35,569
going to be a bigger matrix with
three rows and three columns.

100
00:06:38,990 --> 00:06:43,322
We can use the process that we
evaluate that we worked on

101
00:06:43,322 --> 00:06:47,293
before to evaluate the elements
in the these matrices. So for

102
00:06:47,293 --> 00:06:53,069
example, the element that goes
in here is 1 * 3 + 2 * 5 added

103
00:06:53,069 --> 00:06:55,957
to 3 * -- 1, which is 10.

104
00:06:57,730 --> 00:07:02,295
And you can check for yourself
that the remaining elements are

105
00:07:02,295 --> 00:07:03,955
131 and minus 11.

106
00:07:04,780 --> 00:07:08,108
So it's possible to workout
C * D and the answer is a

107
00:07:08,108 --> 00:07:09,132
two by two matrix.

108
00:07:10,430 --> 00:07:14,846
When we do it the other way
round, let's take an element

109
00:07:14,846 --> 00:07:19,262
here. Let's take the elements in
the first row, first column and

110
00:07:19,262 --> 00:07:23,310
we obtain the answer by working
with the first row, first

111
00:07:23,310 --> 00:07:27,726
column. Here, that's three
times, one is 3 added to minus 7

112
00:07:27,726 --> 00:07:32,142
* 4. That's three added to minus
28, which is minus 25.

113
00:07:34,760 --> 00:07:39,713
And you can proceed in the same
way to fill out this resulting

114
00:07:39,713 --> 00:07:44,666
matrix and the numbers. You'll
get a -- 25 -- 29 -- 33.

115
00:07:46,540 --> 00:07:52,620
9. 1521
789

116
00:07:53,990 --> 00:07:58,540
The important point that I want
to make here is that when you

117
00:07:58,540 --> 00:08:00,290
multiply C * D together.

118
00:08:00,950 --> 00:08:06,350
It may be possible to also find
D * C, But the answers that you

119
00:08:06,350 --> 00:08:09,230
get may have completely
different sizes. It's certainly

120
00:08:09,230 --> 00:08:14,270
not true that CD is the same as
DC, so one of the observations

121
00:08:14,270 --> 00:08:18,950
we take away straight away is
that in general CD is not equal

122
00:08:18,950 --> 00:08:23,270
to DC. Even in situations where
both of these products do exist,

123
00:08:23,270 --> 00:08:25,790
we say that matrix
multiplication is not

124
00:08:25,790 --> 00:08:29,030
commutative. In general, it
really doesn't matter the order

125
00:08:29,030 --> 00:08:31,190
in which you carry out the

126
00:08:31,190 --> 00:08:34,950
multiplication. Now that we know
how to multiply 2 matrices

127
00:08:34,950 --> 00:08:38,050
together, I'm going to show you
an important property of

128
00:08:38,050 --> 00:08:43,818
identity matrices. Suppose we
have a two by two identity

129
00:08:43,818 --> 00:08:45,456
matrix, that's 1001.

130
00:08:46,920 --> 00:08:52,250
And suppose we have a second
matrix, two 3 -- 4 and seven.

131
00:08:52,250 --> 00:08:55,940
And suppose I want to multiply
these two together.

132
00:08:58,950 --> 00:09:02,030
The identity matrix is
certainly a two by two matrix,

133
00:09:02,030 --> 00:09:05,726
and this matrix is also a two
by two matrix. So because

134
00:09:05,726 --> 00:09:08,806
these numbers are the same, we
can actually workout the

135
00:09:08,806 --> 00:09:12,810
product and the answer is also
a two by two matrix. So the

136
00:09:12,810 --> 00:09:16,198
answer has this sort of shape
with four elements in there.

137
00:09:18,340 --> 00:09:22,904
To get the first element in the
answer, we want to pair 10 with

138
00:09:22,904 --> 00:09:26,816
2 -- 4, multiply the paired
elements together and add so we

139
00:09:26,816 --> 00:09:30,402
get 1 * 2 is 2 added to 0 * --

140
00:09:30,402 --> 00:09:33,184
4. Which is just 1 * 2 is 2.

141
00:09:35,720 --> 00:09:41,080
To get this element here, we
want 1 * 3 which is 3 added to 0

142
00:09:41,080 --> 00:09:43,090
* 7, which is just three.

143
00:09:45,430 --> 00:09:49,630
To get the element in here, we
want to pair 01 with two and

144
00:09:49,630 --> 00:09:54,730
minus four, so it's 0 * 2, which
is nothing 1 * -- 4 is minus 4,

145
00:09:54,730 --> 00:09:56,530
so we just get minus 4.

146
00:09:57,630 --> 00:09:59,436
And finally, the last element is

147
00:09:59,436 --> 00:10:03,854
0 times. Three, which is nothing
1 * 7 is 7, so that's our

148
00:10:03,854 --> 00:10:07,910
answer. And if you look at the
answer you'll see the answer is

149
00:10:07,910 --> 00:10:11,030
identical to the matrix we
started with here. In other

150
00:10:11,030 --> 00:10:14,150
words, multiplying a matrix by
an identity matrix when this

151
00:10:14,150 --> 00:10:16,646
multiplication is possible
leaves an answer which is

152
00:10:16,646 --> 00:10:20,078
identical to the matrix you
started with, and that's a very

153
00:10:20,078 --> 00:10:21,326
important property of identity

154
00:10:21,326 --> 00:10:25,940
matrices. The same result occurs
if we do the multiplication the

155
00:10:25,940 --> 00:10:30,860
other way round. If we take two
3 -- 4 seven and we multiply it

156
00:10:30,860 --> 00:10:34,140
by the identity matrix, one
nought nought one will find.

157
00:10:34,140 --> 00:10:38,404
It's also possible, and if you
go through the operation 2 * 1

158
00:10:38,404 --> 00:10:42,340
is 2 three times. Nothing is
nothing. The result there is 2.

159
00:10:43,150 --> 00:10:46,000
Two times nothing
is nothing 313.

160
00:10:47,440 --> 00:10:51,604
Minus 4 * 1 added to 7 times
nought is minus 4.

161
00:10:52,270 --> 00:10:56,131
And minus four times North,
which is nothing added to 717

162
00:10:56,131 --> 00:11:00,694
and you'll see again this answer
here is the same as this matrix

163
00:11:00,694 --> 00:11:04,204
here. So that's very important
property to remember when you

164
00:11:04,204 --> 00:11:08,065
multiply a matrix by an identity
matrix, it leaves the original

165
00:11:08,065 --> 00:11:10,873
matrix unaltered, identical to
what it was before.

166
00:11:11,480 --> 00:11:14,135
The same works even if we
haven't got square

167
00:11:14,135 --> 00:11:16,200
matrices. Suppose we have
this identity matrix.

168
00:11:19,650 --> 00:11:23,178
And we multiply, for example by
the Matrix 78.

169
00:11:24,810 --> 00:11:28,702
Well, this has got one row and
two columns. It's a one by two

170
00:11:28,702 --> 00:11:32,038
matrix. This is got two rows,
two columns, so we can perform

171
00:11:32,038 --> 00:11:35,096
the matrix multiplication and
the result is going to be a

172
00:11:35,096 --> 00:11:38,432
one by two matrix that's the
same shape as the one we

173
00:11:38,432 --> 00:11:38,988
started with.

174
00:11:40,360 --> 00:11:43,708
And if we carry out the
operations, it's 7 * 1, which

175
00:11:43,708 --> 00:11:46,777
is 7 added to 8 times
nothing, which is nothing. So

176
00:11:46,777 --> 00:11:48,172
the result is just 7th.

177
00:11:49,200 --> 00:11:54,015
7 times and nothing is nothing
and 8 * 1 is 8, so it's just

178
00:11:54,015 --> 00:11:58,188
eight. And again, this answer 7
eight is the same as the matrix

179
00:11:58,188 --> 00:12:01,719
we started with over here. So
that's just to reinforce the

180
00:12:01,719 --> 00:12:04,608
message that multiplying by an
identity matrix leaves the

181
00:12:04,608 --> 00:12:05,571
original matrix unaltered.