1
00:00:00 --> 00:00:07
OK, this lecture is like the
beginning of the second half of

2
00:00:07 --> 00:00:14
this course because up to now we
paid a lot of attention to

3
00:00:14 --> 00:00:16
rectangular matrices.

4
00:00:16 --> 00:00:23
Now, concentrating on square
matrices, so we're at two big

5
00:00:23 --> 00:00:24
topics.

6
00:00:24 --> 00:00:31
The determinant of a square
matrix, so this is the first

7
00:00:31 --> 00:00:37
lecture in that
new chapter on determinants,

8
00:00:37 --> 00:00:40.69
and the reason,
the big reason we need the

9
00:00:40.69 --> 00:00:43
determinants is for the Eigen
values.

10
00:00:43 --> 00:00:47.9
So this is really determinants
and Eigen values,

11
00:00:47.9 --> 00:00:50.68
the next big,
big chunk of 18.06.

12
00:00:50.68 --> 00:00:55
OK, so the determinant is a
number associated with every

13
00:00:55 --> 00:01:01
square matrix,
so every square matrix has this

14
00:01:01 --> 00:01:06
number associated with called
the, its determinant.

15
00:01:06 --> 00:01:12
I'll often write it as D E T A
or often also I'll write it as,

16
00:01:12 --> 00:01:17
A with vertical bars,
so that's going to mean the

17
00:01:17 --> 00:01:20
determinant of the matrix.

18
00:01:20 --> 00:01:26
That's going to mean this one,
like, magic number.

19
00:01:26 --> 00:01:32
Well, one number can't tell you
what the whole matrix was.

20
00:01:32 --> 00:01:35
But this one number,
just packs in as much

21
00:01:35 --> 00:01:39
information as possible into a
single number,

22
00:01:39 --> 00:01:45
and of course the one fact that
you've seen before and we have

23
00:01:45 --> 00:01:51
to see it again is the matrix is
invertible when the determinant

24
00:01:51 --> 00:01:53
is not zero.

25
00:01:53 --> 00:01:57
The matrix is singular when the
determinant is zero.

26
00:01:57 --> 00:02:01
So the determinant will be a
test for invertibility,

27
00:02:01 --> 00:02:06
but the determinant's got a lot
more to it than that,

28
00:02:06 --> 00:02:07
so let me start.

29
00:02:07 --> 00:02:10
OK, now the question is how to
start.

30
00:02:10 --> 00:02:14
Do I give you a big formula for
the determinant,

31
00:02:14 --> 00:02:17
all in one
gulp?

32
00:02:17 --> 00:02:18.7
I don't think so!

33
00:02:18.7 --> 00:02:23
That big formula has got too
much packed in it.

34
00:02:23 --> 00:02:28
I would rather start with three
properties of the determinant,

35
00:02:28 --> 00:02:31
three properties that it has.

36
00:02:31 --> 00:02:34
And let me tell you property
one.

37
00:02:34 --> 00:02:38
The determinant of the identity
is one.

38
00:02:38 --> 00:02:38
OK.

39
00:02:38 --> 00:02:39.13
I...

40
00:02:39.13 --> 00:02:45
I wish the other two
properties were as easy to tell

41
00:02:45 --> 00:02:46
you as that.

42
00:02:46 --> 00:02:51
But actually the second
property is pretty

43
00:02:51 --> 00:02:56
straightforward too,
and then once we get the third

44
00:02:56 --> 00:03:00.62
we will actually have the
determinant.

45
00:03:00.62 --> 00:03:06
Those three properties define
the determinant and we can --

46
00:03:06 --> 00:03:13
then we can figure out,
well, what is the determinant?

47
00:03:13 --> 00:03:17
What's a formula for it?

48
00:03:17 --> 00:03:26
Now, the second property is
what happens if you exchange two

49
00:03:26 --> 00:03:28
rows of a matrix.

50
00:03:28 --> 00:03:33
What happens to the
determinant?

51
00:03:33 --> 00:03:42
So, property two is exchange
rows, reverse the sign of the

52
00:03:42 --> 00:03:44
determinant.

53
00:03:44 --> 00:03:48
A lot of plus and minus signs.

54
00:03:48 --> 00:03:55
I even wrote here,
"plus and minus

55
00:03:55 --> 00:03:59
signs," because this is,
like, that's what you have to

56
00:03:59 --> 00:04:04
pay attention to in the formulas
and properties of determinants.

57
00:04:04 --> 00:04:08
So that -- you see what I mean
by a property here?

58
00:04:08 --> 00:04:12
I haven't yet told you what the
determinant is,

59
00:04:12 --> 00:04:17
but whatever it is,
if I exchange two rows to get a

60
00:04:17 --> 00:04:21
different matrix that reverses
the sign of the determinant.

61
00:04:21 --> 00:04:25
And so now, actually,
what matrices do we now know

62
00:04:25 --> 00:04:27
the determinant of?

63
00:04:27 --> 00:04:30
From one and two,
I now know the determinant.

64
00:04:30 --> 00:04:35
Well, I certainly know the
determinant of the identity

65
00:04:35 --> 00:04:39
matrix and now I know the
determinant of

66
00:04:39 --> 00:04:44
every other matrix that comes
from row exchanges from the

67
00:04:44 --> 00:04:45
identities still.

68
00:04:45 --> 00:04:49
So what matrices have I gotten
at this point?

69
00:04:49 --> 00:04:51
The permutations,
right.

70
00:04:51 --> 00:04:55
At this point I know every
permutation matrix,

71
00:04:55 --> 00:05:00
so now I'm saying the
determinant of a permutation

72
00:05:00 --> 00:05:04
matrix is one or minus one.

73
00:05:04 --> 00:05:09
One or minus one,
depending whether the number of

74
00:05:09 --> 00:05:14
exchanges was even or the number
of exchanges was odd.

75
00:05:14 --> 00:05:19
So this is the determinant of a
permutation.

76
00:05:19 --> 00:05:23
Now, P is back to standing for
permutation.

77
00:05:23 --> 00:05:28
OK.
if I could carry on this board,

78
00:05:28 --> 00:05:31
I could, like,
do the two-by-two's.

79
00:05:31 --> 00:05:35
So, property one tells me that
this two-by-two matrix.

80
00:05:35 --> 00:05:40
Oh, I better write absolute --
I mean, I'd better write

81
00:05:40 --> 00:05:42
vertical bars,
not brackets,

82
00:05:42 --> 00:05:44
for that determinant.

83
00:05:44 --> 00:05:47
Property one said,
in the two-by-two case,

84
00:05:47 --> 00:05:51
that this matrix has
determinant one.

85
00:05:51 --> 00:05:59
Property two tells me that this
matrix has determinant -- what?

86
00:05:59 --> 00:06:00
Negative one.

87
00:06:00 --> 00:06:03
This is, like,
two-by-twos.

88
00:06:03 --> 00:06:10.32
Now, I finally want to get --
well, ultimately I want to get

89
00:06:10.32 --> 00:06:14
to, the formula that we all
know.

90
00:06:14 --> 00:06:20
Let me put that way over here,
that the determinant of a

91
00:06:20 --> 00:06:24.75
general
two-by-two is ad-bc.

92
00:06:24.75 --> 00:06:25
OK.

93
00:06:25 --> 00:06:31
I'm going to leave that up,
like, as the two by two case

94
00:06:31 --> 00:06:35
that we already know,
so that every property,

95
00:06:35 --> 00:06:40
I can, like,
check that it's correct for

96
00:06:40 --> 00:06:41
two-by-twos.

97
00:06:41 --> 00:06:48
But the whole point of these
properties is that they're going

98
00:06:48 --> 00:06:52
to give me a formula for n-by-n.

99
00:06:52 --> 00:06:54
That's the whole point.

100
00:06:54 --> 00:06:59
They're going to give me this
number that's a test for

101
00:06:59 --> 00:07:05
invertibility and other great
properties for any size matrix.

102
00:07:05 --> 00:07:09
OK, now you see I'm like,
slowing down because property

103
00:07:09 --> 00:07:12
three is the key property.

104
00:07:12 --> 00:07:16
Property three is the key
property and can I somehow

105
00:07:16 --> 00:07:23
describe it --
maybe I'll separate it into 3A

106
00:07:23 --> 00:07:24
and 3B.

107
00:07:24 --> 00:07:31
Property 3A says that if I
multiply one of the rows,

108
00:07:31 --> 00:07:38
say the first row,
by a number T -- I'm going to

109
00:07:38 --> 00:07:40.23
erase that.

110
00:07:40.23 --> 00:07:46
That's, like,
what I'm headed for but I'm not

111
00:07:46 --> 00:07:49
there yet.

112
00:07:49 --> 00:07:55
It's the one we know and you'll
see that it's checked out by

113
00:07:55 --> 00:07:57
each property.

114
00:07:57 --> 00:08:00
OK, so this is for any matrix.

115
00:08:00 --> 00:08:05
For any matrix,
if I multiply one row by T and

116
00:08:05 --> 00:08:12
leave the other row or other n-1
rows alone, what happens to the

117
00:08:12 --> 00:08:13
determinant?

118
00:08:13 --> 00:08:15
The factor T comes out.

119
00:08:15 --> 00:08:20
It's T times this determinant.

120
00:08:20 --> 00:08:21
That's not hard.

121
00:08:21 --> 00:08:27
I shouldn't have made a big
deal out of property 3A,

122
00:08:27 --> 00:08:31
and 3B is that,
if is, is if I keep -- I'm

123
00:08:31 --> 00:08:37
always keeping this second row
the same, or that last n-1 rows

124
00:08:37 --> 00:08:39
are all staying the same.

125
00:08:39 --> 00:08:45.84
I'm just working -- I'm just
looking inside the first

126
00:08:45.84 --> 00:08:51
row and if I have an a+a' there
and a b+b' there -- sorry,

127
00:08:51 --> 00:08:52
I didn't.

128
00:08:52 --> 00:08:52
Ahh.

129
00:08:52 --> 00:08:57.15
Why don't -- I'll use an
eraser, do it right.

130
00:08:57.15 --> 00:08:58
b+b' there.

131
00:08:58 --> 00:09:00
You see what I'm doing?

132
00:09:00 --> 00:09:06
This property and this property
are about linear combinations,

133
00:09:06 --> 00:09:11
of the first row only,
leaving the

134
00:09:11 --> 00:09:13
other rows unchanged.

135
00:09:13 --> 00:09:16
They'll copy along.

136
00:09:16 --> 00:09:23
Then, then I get the sum --
this breaks up into the sum of

137
00:09:23 --> 00:09:27
this determinant and this one.

138
00:09:27 --> 00:09:30
I'm putting up formulas.

139
00:09:30 --> 00:09:34
Maybe I can try to say it in
words.

140
00:09:34 --> 00:09:38
The determinant is a linear
function.

141
00:09:38 --> 00:09:47
It behaves like a linear
function of first row if all the

142
00:09:47 --> 00:09:50
other rows stay the same.

143
00:09:50 --> 00:09:54
I not saying that -- let me
emphasize.

144
00:09:54 --> 00:10:00
I not saying that the
determinant of A plus B is

145
00:10:00 --> 00:10:04
determinant of A plus
determinant of B.

146
00:10:04 --> 00:10:06
I not saying that.

147
00:10:06 --> 00:10:13.53
I'd better -- can I -- how do I
get it onto tape that I'm not

148
00:10:13.53 --> 00:10:16
saying that?

149
00:10:16 --> 00:10:21
You see, this would allow all
the rows -- you know,

150
00:10:21 --> 00:10:27
A to have a bunch of rows,
B to have a bunch of rows.

151
00:10:27 --> 00:10:30
That's not the linearity I'm
after.

152
00:10:30 --> 00:10:34.45
I'm only after linearity in
each row.

153
00:10:34.45 --> 00:10:36
Linear for each row.

154
00:10:36 --> 00:10:41
Well, you may say I only talked
about the first row,

155
00:10:41 --> 00:10:46
but I claim it's also linear in
the

156
00:10:46 --> 00:10:51
second row, if I had this --
but not, I can't,

157
00:10:51 --> 00:10:56.24
I can't have a combination in
both first and second.

158
00:10:56.24 --> 00:11:01
If I had a combination in the
second row, then I could use

159
00:11:01 --> 00:11:07
rule two to put it up in the
first row, use my property and

160
00:11:07 --> 00:11:11
then use rule two again to put
it back,

161
00:11:11 --> 00:11:15
so each row is OK,
not only the first row,

162
00:11:15 --> 00:11:18.24
but each row separately.

163
00:11:18.24 --> 00:11:22
OK, those are the three
properties, and from those

164
00:11:22 --> 00:11:27
properties, so that's properties
one, two, three.

165
00:11:27 --> 00:11:33
From those, I want to get all
-- I'm going to learn a lot more

166
00:11:33 --> 00:11:35
about the determinant.

167
00:11:35 --> 00:11:38
Let me take an example.

168
00:11:38 --> 00:11:42.44
What would I like to learn?

169
00:11:42.44 --> 00:11:49
I would like to learn that --
so here's our property four.

170
00:11:49 --> 00:11:54
Let me use the same numbering
as here.

171
00:11:54 --> 00:12:02
Property four is if two rows
are equal, the determinant is

172
00:12:02 --> 00:12:03
zero.

173
00:12:03 --> 00:12:05
OK, so property four.

174
00:12:05 --> 00:12:11.95
Two equal rows lead to
determinant equals zero.

175
00:12:11.95 --> 00:12:12
Right.

176
00:12:12 --> 00:12:20
Now, of course I can --
in the two-by-two case I can

177
00:12:20 --> 00:12:25
check, sure, the determinant of
ab ab comes out zero.

178
00:12:25 --> 00:12:29
But I want to see why it's true
for n-by-n.

179
00:12:29 --> 00:12:34
Suppose row one equals row
three for a seven-by-seven

180
00:12:34 --> 00:12:34
matrix.

181
00:12:34 --> 00:12:38.79
So two rows are the same in a
big matrix.

182
00:12:38.79 --> 00:12:43
And all I have to work with is
these properties.

183
00:12:43 --> 00:12:48
The exchange property,
which flips the sign,

184
00:12:48 --> 00:12:52
and the linearity property
which works in each row

185
00:12:52 --> 00:12:53
separately.

186
00:12:53 --> 00:12:55
OK, can you see the reason?

187
00:12:55 --> 00:12:58
How do I get this one out of
properties one,

188
00:12:58 --> 00:12:59
two, three?

189
00:12:59 --> 00:13:03.19
Because -- that's all I have to
work with.

190
00:13:03.19 --> 00:13:06
Everything has to come from
properties one,

191
00:13:06 --> 00:13:08
two, three.

192
00:13:08 --> 00:13:14
OK, so suppose I have a matrix,
and two rows are even.

193
00:13:14 --> 00:13:20
How do I see that its
determinant has to be zero from

194
00:13:20 --> 00:13:22
these properties?

195
00:13:22 --> 00:13:24
I do an exchange.

196
00:13:24 --> 00:13:28
Property two is just set up for
this.

197
00:13:28 --> 00:13:30
Use property two.

198
00:13:30 --> 00:13:33.9
Use exchange -- exchange rows.

199
00:13:33.9 --> 00:13:40
Exchange those rows,
and I get the same matrix.

200
00:13:40 --> 00:13:44.22
Of course, because those rows
were equal.

201
00:13:44.22 --> 00:13:47
So the determinant didn't
change.

202
00:13:47 --> 00:13:52
But on the other hand,
property two says that the sign

203
00:13:52 --> 00:13:53
did change.

204
00:13:53 --> 00:13:57.35
So the -- so I,
I have a determinant whose sign

205
00:13:57.35 --> 00:14:02
doesn't change and does change,
and the only possibility then

206
00:14:02 --> 00:14:06
is that the determinant is zero.

207
00:14:06 --> 00:14:09
You see the reasoning there?

208
00:14:09 --> 00:14:10
Straightforward.

209
00:14:10 --> 00:14:15
Property two just told us,
hey, if we've got two equal

210
00:14:15 --> .
rows we.

211
. --> 00:14:15


212
00:14:15 --> 00:14:18
we've got a zero determinant.

213
00:14:18 --> 00:14:23
And of course in our minds,
that matches the fact that if I

214
00:14:23 --> 00:14:27
have two equal rows the matrix
isn't invertible.

215
00:14:27 --> 00:14:33
If I have two equal rows,
I know that the rank is less

216
00:14:33 --> 00:14:34
than n.

217
00:14:34 --> 00:14:38
OK, ready for property five.

218
00:14:38 --> 00:14:44
Now, property five you'll
recognize as P.

219
00:14:44 --> 00:14:51
It says that the elimination
step that I'm always doing,

220
00:14:51 --> 00:14:58
subtract a multiple,
subtract some multiple l times

221
00:14:58 --> 00:15:04
row one from another row,
row k, let's say.

222
00:15:04 --> 00:15:08
You remember why I did that.

223
00:15:08 --> 00:15:14
In elimination I'm always
choosing

224
00:15:14 --> 00:15:19
this multiplier so as to
produce zero in that position.

225
00:15:19 --> 00:15:24
Or row I from row k,
maybe I should just make very

226
00:15:24 --> 00:15:29
clear that there's nothing
special about row one here.

227
00:15:29 --> 00:15:32
OK, so that,
you can see why I want that

228
00:15:32 --> 00:15:38.23
one, because that will allow me
to start with this full matrix

229
00:15:38.23 --> 00:15:43
whose
determinant I don't know,

230
00:15:43 --> 00:15:48
and I can do elimination and
clean it out.

231
00:15:48 --> 00:15:55
I can get zeroes below the
diagonal by these elimination

232
00:15:55 --> 00:16:03
steps and the point is that the
determinant, the determinant

233
00:16:03 --> 00:16:05
doesn't change.

234
00:16:05 --> 00:16:12
So all those steps of
elimination are OK not changing

235
00:16:12 --> 00:16:15
the determinant.

236
00:16:15 --> 00:16:20
In our language in the early
chapter the determinant of A is

237
00:16:20 --> 00:16:24
going to be the same as the
determinant of U,

238
00:16:24 --> 00:16:26
the upper triangular one.

239
00:16:26 --> 00:16:29
It just has the pivots on the
diagonal.

240
00:16:29 --> 00:16:31
That's why we'll want this
property.

241
00:16:31 --> 00:16:35
OK, do you see where that
property's coming from?

242
00:16:35 --> 00:16:39
Let me do the two-by-two case.

243
00:16:39 --> 00:16:42
Let me do the two-by-two case
here.

244
00:16:42 --> 00:16:47.1
So, I'll keep property five
going along.

245
00:16:47.1 --> 00:16:48
So what I doing?

246
00:16:48 --> 00:16:53
I'm going to keep -- I'm going
to have ab cd,

247
00:16:53 --> 00:16:59
but I'm going to subtract l
times the first row from the

248
00:16:59 --> 00:17:00
second row.

249
00:17:00 --> 00:17:05
And that's the determinant and
of

250
00:17:05 --> 00:17:09
course I can multiply that out
and figure out,

251
00:17:09 --> 00:17:13
sure enough,
ad-bc is there and this minus

252
00:17:13 --> 00:17:17
ALB plus ALB cancels out,
but I just cheated,

253
00:17:17 --> 00:17:17
right?

254
00:17:17 --> 00:17:20
I've got to use the properties.

255
00:17:20 --> 00:17:22
So what property?

256
00:17:22 --> 00:17:25
Well, of course,
this is a com -- I'm keeping

257
00:17:25 --> 00:17:30
the first row the same and the
second

258
00:17:30 --> 00:17:34
row has a c and a d,
and then there's the

259
00:17:34 --> 00:17:39
determinant of the A and the B,
and the minus LA,

260
00:17:39 --> 00:17:41
and the minus LB.

261
00:17:41 --> 00:17:44
So what property was that?
3B.

262
00:17:44 --> 00:17:50
I kept one row the same and I
had a combination in the second,

263
00:17:50 --> 00:17:56
in the other row,
and I just separated it out.

264
00:17:56 --> 00:17:59
OK, so that's property 3.

265
00:17:59 --> 00:18:02
That's by number 3,
3B if you like.

266
00:18:02 --> 00:18:03
OK, now use 3A.

267
00:18:03 --> 00:18:08
How do you use 3A,
which says I can factor out an

268
00:18:08 --> 00:18:12
l, I can factor out a minus l
here.

269
00:18:12 --> 00:18:17
I can factor a minus l out from
this row, no problem.

270
00:18:17 --> 00:18:18.52
That was 3A.

271
00:18:18.52 --> 00:18:26
So now I've used property three
and now I'm ready for the kill.

272
00:18:26 --> 00:18:30
Property four says that this
determinant is zero,

273
00:18:30 --> 00:18:31
has two equal rows.

274
00:18:31 --> 00:18:34
You see how that would always
work?

275
00:18:34 --> 00:18:38
I subtract a multiple of one
row from another one.

276
00:18:38 --> 00:18:44
It gives me a combination in
row k of the old row and l times

277
00:18:44 --> 00:18:50
this copy of the higher row,
and then if -- since I have two

278
00:18:50 --> 00:18:56
equal rows, that's zero,
so the determinant after

279
00:18:56 --> 00:19:00
elimination is the same as
before.

280
00:19:00 --> 00:19:01
Good.

281
00:19:01 --> 00:19:01
OK.

282
00:19:01 --> 00:19:09.22
Now, let's see -- if I rescue
my glasses, I can see what's

283
00:19:09.22 --> 00:19:10
property six.

284
00:19:10 --> 00:19:15
Oh, six is easy,
plenty of space.

285
00:19:15 --> 00:19:23
Row of zeroes leads to
determinant of A equals zero.

286
00:19:23 --> 00:19:25
A complete row of zeroes.

287
00:19:25 --> 00:19:27
So I'm again,
this is like,

288
00:19:27 --> 00:19:31
something I'll use in the
singular case.

289
00:19:31 --> 00:19:36
Actually, you can look ahead to
why I need these properties.

290
00:19:36 --> 00:19:40
So I'm going to use property
five, the elimination,

291
00:19:40 --> 00:19:45
use this stuff to say that this
determinant is the same as that

292
00:19:45 --> 00:19:50
determinant and I'll
produce a zero there.

293
00:19:50 --> 00:19:52
But what if I also produce a
zero there?

294
00:19:52 --> 00:19:56
What if elimination gives a row
of zeroes?

295
00:19:56 --> 00:19:59
That, that used to be my test
for, mmm, singular,

296
00:19:59 --> 00:20:03
not invertible,
rank two -- rank less than N,

297
00:20:03 --> 00:20:06
and now I'm seeing it's also
gives determinant zero.

298
00:20:06 --> 00:20:11
How do I get that one from the
previous properties?

299
00:20:11 --> 00:20:17
'Cause I -- this is not a new
law, this has got to come from

300
00:20:17 --> 00:20:18
the old ones.

301
00:20:18 --> 00:20:20
So what shall I do?

302
00:20:20 --> 00:20:23
Yeah, use -- that's brilliant.

303
00:20:23 --> 00:20:26
If you use 3A with T equals
zero.

304
00:20:26 --> 00:20:27
Right.

305
00:20:27 --> 00:20:32
So I have this zero zero cd,
and I'm trying to show that

306
00:20:32 --> 00:20:35
that
determinant is zero.

307
00:20:35 --> 00:20:42
OK, so the zero is the same is
-- five, can I take T equals

308
00:20:42 --> 00:20:46
five, just to,
like, pin it down?

309
00:20:46 --> 00:20:49
I'll multiply this row by five.

310
00:20:49 --> 00:20:54
Five, well then,
five of this should -- if I,

311
00:20:54 --> 00:21:01
if there's a factor five in
that, you see what -- so this is

312
00:21:01 --> 00:21:06.52
property 3A, with taking T as
five.

313
00:21:06.52 --> 00:21:11
If I multiply a row by five,
out comes a five.

314
00:21:11 --> 00:21:13
So why I doing this?

315
00:21:13 --> 00:21:17
Oh, because that's still zero
zero, right?

316
00:21:17 --> 00:21:22.58
So that's this determinant
equals five times this

317
00:21:22.58 --> 00:21:27
determinant, and the determinant
has to be zero.

318
00:21:27 --> 00:21:31
I think I didn't do that the
very best way.

319
00:21:31 --> 00:21:37
You were, yeah,
you had the idea better.

320
00:21:37 --> 00:21:41
Multiply, yeah,
take T equals zero.

321
00:21:41 --> 00:21:43
Was that your idea?

322
00:21:43 --> 00:21:46
Take T equals zero in rule 3B.

323
00:21:46 --> 00:21:52.49
If T is zero in rule 3B,
and I bring the camera back to

324
00:21:52.49 --> 00:21:54
rule 3B -- sorry.

325
00:21:54 --> 00:21:59
If T is zero,
then I have a zero zero there

326
00:21:59 --> 00:22:03
and the determinant is zero.

327
00:22:03 --> 00:22:08
OK, one way or another,
a row of zeroes means zero

328
00:22:08 --> 00:22:09
determinant.

329
00:22:09 --> 00:22:13
OK, now I have to get serious.

330
00:22:13 --> 00:22:19
The next properties are the
ones that we're building up to.

331
00:22:19 --> 00:22:22
OK, so I can do elimination.

332
00:22:22 --> 00:22:28
I can reduce to a triangular
matrix and now what's the

333
00:22:28 --> 00:22:33
determinant of that triangular
matrix?

334
00:22:33 --> 00:22:37
Suppose, suppose I -- all
right, rule seven.

335
00:22:37 --> 00:22:41.75
So suppose my matrix is now
triangular.

336
00:22:41.75 --> 00:22:47
So it's got -- so I even give
these the names of the pivots,

337
00:22:47 --> 00:22:51
d1, d2, to dn,
and stuff is up here,

338
00:22:51 --> 00:22:58
I don't know what that is,
but what I do know is this is

339
00:22:58 --> 00:22:59.14
all zeroes.

340
00:22:59.14 --> 00:23:03
That's all zeroes,
and I would like to know the

341
00:23:03 --> 00:23:09
determinant, because elimination
will get me to this.

342
00:23:09 --> 00:23:13.84
So once I'm here,
what's the determinant then?

343
00:23:13.84 --> 00:23:18.26
Let me use an eraser to get
those,

344
00:23:18.26 --> 00:23:23
get that vertical bar again,
so that I'm taking the

345
00:23:23 --> 00:23:29
determinant of U so that,
so, what is the determinant of

346
00:23:29 --> 00:23:32
an upper triangular matrix?

347
00:23:32 --> 00:23:35
Do you know the answer?

348
00:23:35 --> 00:23:38.97
It's just the product of the
d's.

349
00:23:38.97 --> 00:23:45
The -- these things that I
don't even put in letters for,

350
00:23:45 --> 00:23:51
because they don't matter,
the determinant is d1 times d2

351
00:23:51 --> 00:23:53
times
dn.

352
00:23:53 --> 00:24:00
If I have a triangular matrix,
then the diagonal is all I have

353
00:24:00 --> 00:24:01
to work with.

354
00:24:01 --> 00:24:05
And that's, that's telling us
then.

355
00:24:05 --> 00:24:11
We now have the way that
MATLAB, any reasonable software,

356
00:24:11 --> 00:24:14
would compute a determinant.

357
00:24:14 --> 00:24:19
If I have a matrix of size a
hundred, the way I would

358
00:24:19 --> 00:24:25
actually compute its
determinant would be

359
00:24:25 --> 00:24:32
elimination, make it triangular,
multiply the pivots together,

360
00:24:32 --> 00:24:38
the product of the pivots,
the product of pivots.

361
00:24:38 --> 00:24:44
Now, there's always in
determinants a plus or minus

362
00:24:44 --> 00:24:46
sign to remember.

363
00:24:46 --> 00:24:51
Where -- where does that come
into this rule?

364
00:24:51 --> 00:24:57
Could it be,
could the determinant be minus

365
00:24:57 --> 00:24:59
the product of the pivots?

366
00:24:59 --> 00:25:02
The determinant is d1,
d2, to dn.

367
00:25:02 --> 00:25:03.74
No doubt about that.

368
00:25:03.74 --> 00:25:08.73
But to get to this triangular
form, we may have had to do a

369
00:25:08.73 --> 00:25:12
row exchange,
so, so this -- it's the product

370
00:25:12 --> 00:25:16
of the pivots if there were no
row exchanges.

371
00:25:16 --> 00:25:20
If, if SLU code,
the simple LU code,

372
00:25:20 --> 00:25:23
the square LU went right
through.

373
00:25:23 --> 00:25:29
If we had to do some row
exchanges, then we've got to

374
00:25:29 --> 00:25:31
watch plus or minus.

375
00:25:31 --> 00:25:35
OK, but though -- this law is
simply that.

376
00:25:35 --> 00:25:37
OK, how do I prove that?

377
00:25:37 --> 00:25:42
Let's see, let me suppose that
d's are not

378
00:25:42 --> 00:25:43
zeroes.

379
00:25:43 --> 00:25:45
The pivots are not zeroes.

380
00:25:45 --> 00:25:49
And tell me,
how do I show that none of this

381
00:25:49 --> 00:25:52.7
upper stuff makes any
difference?

382
00:25:52.7 --> 00:25:55
How do I get zeroes there?

383
00:25:55 --> 00:25:56
By elimination,
right?

384
00:25:56 --> 00:26:00.9
I just multiply this row by the
right number,

385
00:26:00.9 --> 00:26:03
subtract from that row,
kills that.

386
00:26:03 --> 00:26:08
I multiply this row by the
right

387
00:26:08 --> 00:26:11
number, kills that,
by the right number,

388
00:26:11 --> 00:26:12
kills that.

389
00:26:12 --> 00:26:16
Now, you kill every one of
these off-diagonal terms,

390
00:26:16 --> 00:26:20.6
no problem and I'm just left
with the diagonal.

391
00:26:20.6 --> 00:26:25
Well, strictly speaking,
I still have to figure out why

392
00:26:25 --> 00:26:29
is, for a diagonal matrix now,
why is that the right

393
00:26:29 --> 00:26:31
determinant?

394
00:26:31 --> 00:26:34
I mean, we sure hope it is,
but why?

395
00:26:34 --> 00:26:38
I have to go back to properties
one, two, three.

396
00:26:38 --> 00:26:43
Why is -- now that the matrix
is suddenly diagonal,

397
00:26:43 --> 00:26:47
how do I know that the
determinant is just a product of

398
00:26:47 --> 00:26:50
those diagonal entries?

399
00:26:50 --> 00:26:52
Well, what I going to use?

400
00:26:52 --> 00:26:56
I'm going to use property 3A,
is

401
00:26:56 --> 00:26:57.01
that right?

402
00:26:57.01 --> 00:26:59
I'll factor this,
I'll factor this,

403
00:26:59 --> 00:27:04
I'll factor that d1 out and
have one and have the first row

404
00:27:04 --> 00:27:05
will be that.

405
00:27:05 --> 00:27:10
And then I'll factor out the
d2, shall I shall I put the d2

406
00:27:10 --> 00:27:13
here, and the second row will
look like that,

407
00:27:13 --> 00:27:14
and so on.

408
00:27:14 --> 00:27:18
So I've factored out all the
d's and

409
00:27:18 --> 00:27:20
what I left with?

410
00:27:20 --> 00:27:21
The identity.

411
00:27:21 --> 00:27:24
And what rule do I finally get
to use?

412
00:27:24 --> 00:27:25
Rule one.

413
00:27:25 --> 00:27:30.59
Finally, this is the point
where rule one finally chips in

414
00:27:30.59 --> 00:27:35
and says that this determinant
is one, so it's the product of

415
00:27:35 --> 00:27:36
the d's.

416
00:27:36 --> 00:27:40
So this was rules five,
to do elimination,

417
00:27:40 --> 00:27:44.15
3A to factor out
the D's, and,

418
00:27:44.15 --> 00:27:46
and our best friend,
rule one.

419
00:27:46 --> 00:27:51
And possibly rule two,
the exchanges may have been

420
00:27:51 --> 00:27:52
needed also.

421
00:27:52 --> 00:27:53
OK.

422
00:27:53 --> 00:27:59
Now I guess I have to consider
also the case if some d is zero,

423
00:27:59 --> 00:28:04
because I was assuming I could
use the d's to clean out and

424
00:28:04 --> 00:28:09
make a diagonal,
but what if --

425
00:28:09 --> 00:28:13
what if one of those diagonal
entries is zero?

426
00:28:13 --> 00:28:19
Well, then with elimination we
know that we can get a row of

427
00:28:19 --> 00:28:24
zeroes, and for a row of zeroes
I'm using rule six,

428
00:28:24 --> 00:28:28.73
the determinant is zero,
and that's right.

429
00:28:28.73 --> 00:28:31.95
So I can check the singular
case.

430
00:28:31.95 --> 00:28:37
In fact, I can now get to the
key point that determinant of A

431
00:28:37 --> 00:28:41
is zero,
exactly when,

432
00:28:41 --> 00:28:44
exactly when A is singular.

433
00:28:44 --> 00:28:51
And otherwise is not singular,
so that the determinant is a

434
00:28:51 --> 00:28:56
fair test for invertibility or
non-invertibility.

435
00:28:56 --> 00:29:03
So, I must be close to that
because I can take any matrix

436
00:29:03 --> 00:29:04
and get there.

437
00:29:04 --> 00:29:08
Do I -- did I have anything to
say?

438
00:29:08 --> 00:29:13
The, the proofs,
it starts by saying by

439
00:29:13 --> 00:29:16
elimination go from A to U.

440
00:29:16 --> 00:29:17
Oh, yeah.

441
00:29:17 --> 00:29:22
Actually looks to me like I
don't -- haven't said anything

442
00:29:22 --> 00:29:26
brand-new here,
that, that really,

443
00:29:26 --> 00:29:30.32
I've got this,
because let's just remember the

444
00:29:30.32 --> 00:29:30
reason.

445
00:29:30 --> 00:29:35.4
By elimination,
I can go from the original A to

446
00:29:35.4 --> 00:29:35
U.

447
00:29:35 --> 00:29:40
Well, OK, now suppose the
matrix is

448
00:29:40 --> 00:29:41
singular.

449
00:29:41 --> 00:29:45
If the matrix is singular,
what happens?

450
00:29:45 --> 00:29:51
Then by elimination I get a row
of zeroes and therefore the

451
00:29:51 --> 00:29:53
determinant is zero.

452
00:29:53 --> 00:29:58
And if the matrix is not
singular, I don't get zero,

453
00:29:58 --> 00:30:03
so maybe -- do you want me to
put this, like,

454
00:30:03 --> 00:30:04.74
in two parts?

455
00:30:04.74 --> 00:30:09
The determinant of A is not
zero

456
00:30:09 --> 00:30:11
when A is invertible.

457
00:30:11 --> 00:30:17
Because I've already -- in
chapter two we figured out when

458
00:30:17 --> 00:30:19
is the matrix invertible.

459
00:30:19 --> 00:30:25.14
It's invertible when
elimination produces a full set

460
00:30:25.14 --> 00:30:29
of pivots and now,
and we now, we know the

461
00:30:29 --> 00:30:34
determinant is the product of
those non-zero numbers.

462
00:30:34 --> 00:30:38
So those are the two cases.

463
00:30:38 --> 00:30:43
If it's singular,
I go to a row of zeroes.

464
00:30:43 --> 00:30:48
If it's invertible,
I go to U and then to the

465
00:30:48 --> 00:30:53
diagonal D, and then which --
and then to d1,

466
00:30:53 --> 00:30:54
d2, up to dn.

467
00:30:54 --> 00:30:59
As the formula -- so we have a
formula now.

468
00:30:59 --> 00:31:05
We have a formula for the
determinant and it's actually a

469
00:31:05 --> 00:31:11
very much more practical formula
than the

470
00:31:11 --> 00:31:12
ad-bc formula.

471
00:31:12 --> 00:31:17
Is it correct,
maybe you should just -- let's

472
00:31:17 --> 00:31:18
just check that.

473
00:31:18 --> 00:31:19
Two-by-two.

474
00:31:19 --> 00:31:23
What are the pivots of a
two-by-two matrix?

475
00:31:23 --> 00:31:28
What does elimination do with a
two-by-two matrix?

476
00:31:28 --> 00:31:31
It -- there's the first pivot,
fine.

477
00:31:31 --> 00:31:33
What's the second pivot?

478
00:31:33 --> 00:31:39
We subtract,
so I'm putting it in this upper

479
00:31:39 --> 00:31:41
triangular form.

480
00:31:41 --> 00:31:46
What do I -- my multiplier is c
over a, right?

481
00:31:46 --> 00:31:53
I multiply that row by c over a
and I subtract to get that zero,

482
00:31:53 --> 00:31:57.63
and here I have d minus c over
a times b.

483
00:31:57.63 --> 00:32:02
That's the elimination on a
two-by-two.

484
00:32:02 --> 00:32:08
So I've finally discovered that
the determinant of this matrix

485
00:32:08 --> 00:32:13
-- I've got it from the
properties, not by knowing the

486
00:32:13 --> 00:32:17
answer from last year,
that the determinant of this

487
00:32:17 --> 00:32:23
two-by-two is the product of A
times that, and of course when I

488
00:32:23 --> 00:32:28
multiply A by that,
the product of that and that

489
00:32:28 --> 00:32:32
is ad minus,
the a is canceled,

490
00:32:32 --> 00:32:32
bc.

491
00:32:32 --> 00:32:36
So that's great,
provided a isn't zero.

492
00:32:36 --> 00:32:41
If a was zero,
that step wasn't allowed,

493
00:32:41 --> 00:32:43
zero wasn't a pivot.

494
00:32:43 --> 00:32:48
So that's what I -- I've
covered all the bases.

495
00:32:48 --> 00:32:56.58
I have to -- if a is zero,
then I have to do the exchange,

496
00:32:56.58 --> 00:33:02
and if the exchange doesn't
work, it's because a is

497
00:33:02 --> 00:33:03
singular.

498
00:33:03 --> 00:33:09
OK, those are -- those are the
direct properties of the

499
00:33:09 --> 00:33:10.59
determinant.

500
00:33:10.59 --> 00:33:14
And now, finally,
I've got two more,

501
00:33:14 --> 00:33:15.97
nine and ten.

502
00:33:15.97 --> 00:33:19
And that's -- I think you
can...

503
00:33:19 --> 00:33:25
Like, the ones
we've got here are totally

504
00:33:25 --> 00:33:31
connected with our elimination
process and whether pivots are

505
00:33:31 --> 00:33:36.46
available and whether we get a
row of zeroes.

506
00:33:36.46 --> 00:33:41
I think all that you can
swallow in one shot.

507
00:33:41 --> 00:33:45
Let me tell you properties nine
and ten.

508
00:33:45 --> 00:33:48
They're quick to write down.

509
00:33:48 --> 00:33:51
They're very,
very useful.

510
00:33:51 --> 00:33:57
So I'll just write them down
and use them.

511
00:33:57 --> 00:34:03.36
Property nine says that the
determinant of a product -- if I

512
00:34:03.36 --> 00:34:05
multiply two matrices.

513
00:34:05 --> 00:34:12
So if I multiply two matrices,
A and B, that the determinant

514
00:34:12 --> 00:34:18.2
of the product is determinant of
A times determinant of B,

515
00:34:18.2 --> 00:34:23
and for me that one is like,
that's a very valuable

516
00:34:23 --> 00:34:29
property, and it's sort of like
partly coming

517
00:34:29 --> 00:34:33
out of the blue,
because we haven't been

518
00:34:33 --> 00:34:38
multiplying matrices and here
suddenly this determinant has

519
00:34:38 --> 00:34:40
this multiplying property.

520
00:34:40 --> 00:34:44
Remember, it didn't have the
linear property,

521
00:34:44 --> 00:34:47
it didn't have the adding
property.

522
00:34:47 --> 00:34:54.67
The determinant of A plus B is
not the sum of the determinants,

523
00:34:54.67 --> 00:35:00.57
but the determinant of A times
B is the product,

524
00:35:00.57 --> 00:35:04
is the product of the
determinants.

525
00:35:04 --> 00:35:10
OK, so for example,
what's the determinant of A

526
00:35:10 --> 00:35:11
inverse?

527
00:35:11 --> 00:35:14
Using that property nine.

528
00:35:14 --> 00:35:21
Let me, let me put that under
here because the camera is

529
00:35:21 --> 00:35:26
happier if it can focus on both
at once.

530
00:35:26 --> 00:35:31
So let me put it underneath.

531
00:35:31 --> 00:35:37
The determinant of A inverse,
because property ten will come

532
00:35:37 --> 00:35:39
in that space.

533
00:35:39 --> 00:35:44
What does this tell me about A
inverse, its determinant?

534
00:35:44 --> 00:35:48
OK, well, what do I know about
A inverse?

535
00:35:48 --> 00:35:52
I know that A inverse times A
is odd.

536
00:35:52 --> 00:35:54
So what I going to do?

537
00:35:54 --> 00:35:59
I'm going to take determinants
of

538
00:35:59 --> 00:36:00
both sides.

539
00:36:00 --> 00:36:05.78
The determinant of I is one,
and what's the determinant of A

540
00:36:05.78 --> 00:36:06
inverse A?

541
00:36:06 --> 00:36:10
That's a product of two
matrices, A and B.

542
00:36:10 --> 00:36:16
So it's the product of the
determinant, so what I learning?

543
00:36:16 --> 00:36:20.85
I'm learning that the
determinant of A inverse times

544
00:36:20.85 --> 00:36:25
the determinant of A is the
determinant of

545
00:36:25 --> 00:36:28
I, that's this one.

546
00:36:28 --> 00:36:33
Again, I happily use property
one.

547
00:36:33 --> 00:36:41
OK, so that tells me that the
determinant of A inverse is one

548
00:36:41 --> 00:36:42
over.

549
00:36:42 --> 00:36:49
Here's my -- here's my
conclusion -- is one over the

550
00:36:49 --> 00:36:51
determinant of A.

551
00:36:51 --> 00:36:58
I guess that that --
I, I always try to think,

552
00:36:58 --> 00:37:01
well, do we know some cases of
that?

553
00:37:01 --> 00:37:06
Of course, we know it's right
already if A is diagonal.

554
00:37:06 --> 00:37:11
If A is a diagonal matrix,
then its determinant is just a

555
00:37:11 --> 00:37:13
product of those numbers.

556
00:37:13 --> 00:37:19
So if A is, for example,
two-three, then we know that

557
00:37:19 --> 00:37:24
A-inverse is one-half
one-third, and sure enough,

558
00:37:24 --> 00:37:28
that has determinant six,
and that has determinant

559
00:37:28 --> 00:37:29
one-sixth.

560
00:37:29 --> 00:37:31.84
And our rule checks.

561
00:37:31.84 --> 00:37:37
So somehow this proof,
this property has to -- somehow

562
00:37:37 --> 00:37:43
the proof of that property -- if
we can boil it down to diagonal

563
00:37:43 --> 00:37:47
matrices then we can read it
off,

564
00:37:47 --> 00:37:52
whether it's A and A-inverse,
or two different diagonal

565
00:37:52 --> 00:37:53
matrices A and B.

566
00:37:53 --> 00:37:56
For diagonal -- so what I
saying?

567
00:37:56 --> 00:37:59
I'm saying for a diagonal
matrices, check.

568
00:37:59 --> 00:38:04
But we'd have to do elimination
steps, we'd have to patiently do

569
00:38:04 --> 00:38:10
the, the, argument if we want to
use these previous properties to

570
00:38:10 --> 00:38:13
get it
for other matrices.

571
00:38:13 --> 00:38:18
And it also tells me -- what,
just let's, see what else it's

572
00:38:18 --> 00:38:19
telling me.

573
00:38:19 --> 00:38:23
What's the determinant of,
of A-squared?

574
00:38:23 --> 00:38:25
If I take a matrix and square
it?

575
00:38:25 --> 00:38:29
Then the determinant just got
squared.

576
00:38:29 --> 00:38:29
Right?

577
00:38:29 --> 00:38:33
The determinant of A-squared is
the

578
00:38:33 --> 00:38:37
determinant of A times the
determinant of A.

579
00:38:37 --> 00:38:42
So if I square the matrix,
I square the determinant.

580
00:38:42 --> 00:38:46
If I double the matrix,
what do I do to the

581
00:38:46 --> 00:38:47
determinant?

582
00:38:47 --> 00:38:49
Think about that one.

583
00:38:49 --> 00:38:54
If I double the matrix,
what -- so the determinant of

584
00:38:54 --> 00:38:59
A, since I'm writing down,
like, facts

585
00:38:59 --> 00:39:04
that follow,
the determinant of A-squared is

586
00:39:04 --> 00:39:08
the determinant of A,
all squared.

587
00:39:08 --> 00:39:11
The determinant of 2A is what?

588
00:39:11 --> 00:39:13
That's A plus A.

589
00:39:13 --> 00:39:18
But wait, er,
I don't want the answer to

590
00:39:18 --> 00:39:20.83
determinant of A here.

591
00:39:20.83 --> 00:39:22
That's wrong.

592
00:39:22 --> 00:39:28
It's not two determinant of A,
What is it?

593
00:39:28 --> 00:39:33.63
what's the number that I have
to multiply determinant of A by

594
00:39:33.63 --> 00:39:38
if I double the whole matrix,
if I double every entry in the

595
00:39:38 --> 00:39:39
matrix?

596
00:39:39 --> 00:39:42
What happens to the
determinant?

597
00:39:42 --> 00:39:45
Supposed it's an n-by-n matrix.

598
00:39:45 --> 00:39:46
Two to the n,
right.

599
00:39:46 --> 00:39:48
Two to the nth.

600
00:39:48 --> 00:39:53
Now, why is it two to the nth,
and not just two?

601
00:39:53 --> 00:39:56
So why is it two to the nth?

602
00:39:56 --> 00:39:59
Because I'm factoring out two
from every row.

603
00:39:59 --> 00:40:04
There's a factor -- this has a
factor two in every row,

604
00:40:04 --> 00:40:07
so I can factor two out of the
first row.

605
00:40:07 --> 00:40:12
I factor a different two out of
the second row,

606
00:40:12 --> 00:40:17.99
a different two out of the nth
row, so I've got all those twos

607
00:40:17.99 --> 00:40:19
coming out.

608
00:40:19 --> 00:40:23
So it's like volume,
really, and that's one of the

609
00:40:23 --> 00:40:26
great applications of
determinants.

610
00:40:26 --> 00:40:31
If I -- if I have a box and I
double all the sides,

611
00:40:31 --> 00:40:36
I multiply the volume by two to
the nth.

612
00:40:36 --> 00:40:42
If it's a box in three
dimensions, I multiply the

613
00:40:42 --> 00:40:43
volume by 8.

614
00:40:43 --> 00:40:46
So this is rule 3A here.

615
00:40:46 --> 00:40:48
This is rule nine.

616
00:40:48 --> 00:40:55
And notice the way this rule
sort of checks out with the

617
00:40:55 --> 00:41:01
singular/non-singular stuff,
that if A is invertible,

618
00:41:01 --> 00:41:06
what does that mean about its
determinant?

619
00:41:06 --> 00:41:11
It's not zero,
and therefore this makes sense.

620
00:41:11 --> 00:41:17
The case when determinant of A
is zero, that's the case where

621
00:41:17 --> 00:41:19
my formula doesn't work anymore.

622
00:41:19 --> 00:41:23
If determinant of A is zero,
this is ridiculous,

623
00:41:23 --> 00:41:25
and that's ridiculous.

624
00:41:25 --> 00:41:30
A-inverse doesn't exist,
and one over zero doesn't make

625
00:41:30 --> 00:41:31.13
sense.

626
00:41:31.13 --> 00:41:34
So don't miss this property.

627
00:41:34 --> 00:41:39
It's sort of,
like, amazing that it can...

628
00:41:39 --> 00:41:46
And the tenth property is
equally simple to state,

629
00:41:46 --> 00:41:52
that the determinant of A
transposed equals the

630
00:41:52 --> 00:41:54
determinant of A.

631
00:41:54 --> 00:42:00
And of course,
let's just check it on the ab

632
00:42:00 --> 00:42:01
cd guy.

633
00:42:01 --> 00:42:06
We could check that sure
enough, that's ab cd,

634
00:42:06 --> 00:42:09
it works.

635
00:42:09 --> 00:42:12
It's ad - bc,
it's ad - bc,

636
00:42:12 --> 00:42:14
sure enough.

637
00:42:14 --> 00:42:19
So that transposing did not
change the determinant.

638
00:42:19 --> 00:42:26
But what it does change is --
well, what it does is it lists,

639
00:42:26 --> 00:42:30
so all -- I've been working
with rows.

640
00:42:30 --> 00:42:37
If a row is all zeroes,
the determinant is zero.

641
00:42:37 --> 00:42:41
But now, with rule ten,
I know what to do is a column

642
00:42:41 --> 00:42:42
is all zero.

643
00:42:42 --> 00:42:46
If a column is all zero,
what's the determinant?

644
00:42:46 --> 00:42:47
Zero, again.

645
00:42:47 --> 00:42:52
So, like all those properties
about rows, exchanging two rows

646
00:42:52 --> 00:42:53
reverses the sign.

647
00:42:53 --> 00:42:57
Now, exchanging two columns
reverses the sign,

648
00:42:57 --> 00:43:02
because I can always,
if I want to see why,

649
00:43:02 --> 00:43:05
I can transpose,
those columns become rows,

650
00:43:05 --> 00:43:08
I do the exchange,
I transpose back.

651
00:43:08 --> 00:43:11
And I've done a column
operation.

652
00:43:11 --> 00:43:15
So, in, in conclusion,
there was nothing special about

653
00:43:15 --> 00:43:20
row one, 'cause I could exchange
rows, and now there's nothing

654
00:43:20 --> 00:43:25.09
special about rows that isn't
equally true for

655
00:43:25.09 --> 00:43:28
columns because this is the
same.

656
00:43:28 --> 00:43:28
OK.

657
00:43:28 --> 00:43:33
And again, maybe I won't -- oh,
let's see.

658
00:43:33 --> 00:43:34
Do we...?

659
00:43:34 --> 00:43:40
Maybe it's worth seeing a quick
proof of this number ten,

660
00:43:40 --> 00:43:44
quick, quick,
er, proof of number ten.

661
00:43:44 --> 00:43:48
Er, let me take the -- this is
number ten.

662
00:43:48 --> 00:43:52
A transposed
equals A.

663
00:43:52 --> 00:43:59.99
Determinate of A transposed
equals determinate of A.

664
00:43:59.99 --> 00:44:04
That's what I should have said.

665
00:44:04 --> 00:44:05
OK.

666
00:44:05 --> 00:44:07
So, let's just,
er.

667
00:44:07 --> 00:44:13
A typical matrix A,
if I use elimination,

668
00:44:13 --> 00:44:16
this factors into LU.

669
00:44:16 --> 00:44:23
And the transpose is U
transpose, l transpose.

670
00:44:23 --> 00:44:25.21
Er...
let me.

671
00:44:25.21 --> .


672
. --> 00:44:30
this is to prove.

673
00:44:30 --> 00:44:33
So this is proof,
this is proof number ten,

674
00:44:33 --> 00:44:37
using -- well,
I don't know which ones I'll

675
00:44:37 --> 00:44:40
use, so I'll put 'em all in,
one to nine.

676
00:44:40 --> 00:44:40
OK.

677
00:44:40 --> 00:44:44
I'm going to prove number ten
by using one to nine.

678
00:44:44 --> 00:44:49
I won't cover every case,
but I'll cover almost every

679
00:44:49 --> 00:44:49
case.

680
00:44:49 --> 00:44:52
So in almost every case,
A can

681
00:44:52 --> 00:44:57
factor into LU,
and A transposed can factor

682
00:44:57 --> 00:44:58
into that.

683
00:44:58 --> 00:45:00
Now, what do I do next?

684
00:45:00 --> 00:45:04
So I want to prove that these
are the same.

685
00:45:04 --> 00:45:06
I see a product here.

686
00:45:06 --> 00:45:08
So I use rule nine.

687
00:45:08 --> 00:45:14
So, now what I want to prove
is, so now I know that this is

688
00:45:14 --> 00:45:19
LU, this is U transposed and l
transposed.

689
00:45:19 --> 00:45:23
Now, just for a practice,
what are all those

690
00:45:23 --> 00:45:25
determinants?

691
00:45:25 --> 00:45:28
So this is, this is,
this is prove this,

692
00:45:28 --> 00:45:33
prove this, prove this,
and now I'm ready to do it.

693
00:45:33 --> 00:45:35
What's the determinant of l?

694
00:45:35 --> 00:45:39
You remember what l is,
it's this lower triangular

695
00:45:39 --> 00:45:44
matrix with ones on the
diagonals.

696
00:45:44 --> 00:45:47
So what is the determinant of
that guy?

697
00:45:47 --> 00:45:48
I- It's one.

698
00:45:48 --> 00:45:53
Any time I have this triangular
matrix, I can get rid of all the

699
00:45:53 --> 00:45:56.98
non-zeroes, down to the diagonal
case.

700
00:45:56.98 --> 00:45:59
The determinate of l is one.

701
00:45:59 --> 00:46:02
How about the determinant of l
transposed?

702
00:46:02 --> 00:46:06.5
That's triangular also,
right?

703
00:46:06.5 --> 00:38:45
Still got those ones on the
diagonal, it's just the

704
00:38:45 --> 00:31:07
non-zeroes flipped to the other
side of the diagonal,

705
00:31:07 --> 00:26:43
but they didn't matter anyway.

706
00:26:43 --> 00:20:33
That's my proof,
really, that once I've got

707
00:20:33 --> 00:14:49
triangular matrices,
l and l transposed,

708
00:14:49 --> 00:06:27
or U and U transposed,
when they're triangular,4

709
00:06:27 --> 00:15:30
I'm down to the product of the
diagonal and if I transpose,

710
00:15:30 --> 00:17:04
who cares?

711
00:17:04 --> 00:24:43
OK, that's not -- I didn't put
in every comma and,

712
00:24:43 --> 00:33:00
and cross every T in that
proof, but that's really the

713
00:33:00 --> 00:33:57
proof.

714
00:33:57 --> 00:39:44
That's the, like,
concrete proof that,

715
00:39:44 --> 00:46:55
that gets -- get down to
triangular

716
00:46:55 --> 00:39:25
matrices and then get down to
diagonal matrices.

717
00:39:25 --> 00:32:33
OK, one more coming,
which I I have to make,

718
00:32:33 --> 00:23:36
because all math professors
watching this will be waiting

719
00:23:36 --> 00:22:29
for it.

720
00:22:29 --> 00:15:18
OK, so they had to wait until
the last minute.

721
00:15:18 --> 00:07:20
What I -- way,
way back in property two,4

722
00:07:20 --> 00:14:29.14
I said that if you do a row
exchange, the determinant

723
00:14:29.14 --> 00:16:16
changes sign.

724
00:16:16 --> 00:24:31
So if I do seven row exchanges,
the determinant changes sign,

725
00:24:31 --> 00:32:38
but it -- would it be possible
t- to produce the same matrix

726
00:32:38 --> 00:39:39
with seven row exchanges and
with ten row exchanges?

727
00:39:39 --> 00:47:46
If that were possible,
that would be a bad thing,

728
00:47:46 --> 00:47:46
right?

729
00:47:46 --> 00:47:49
If If I could -- why would it
be bad?

730
00:47:49 --> 00:47:52
My whole lecture would die,
right?

731
00:47:52 --> 00:47:56
Because rule two said that if
you do seven row exchanges,

732
00:47:56 --> 00:48:00
then the sign of the
determinant reverses.

733
00:48:00 --> 00:48:04
But if you do ten row
exchanges, the sign of the

734
00:48:04 --> 00:48:08
determinant stays the same,
because minus one

735
00:48:08 --> 00:48:10
ten times is plus one.

736
00:48:10 --> 00:48:15
OK, so there's a hidden fact
here, that every -- like,

737
00:48:15 --> 00:48:19
every permutation,
the permutations are either odd

738
00:48:19 --> 00:48:20
or even.

739
00:48:20 --> 00:48:25
I could get the permutation
with seven row exchanges,

740
00:48:25 --> 00:48:29
then I could probably get it
with twenty-one,

741
00:48:29 --> 00:48:32
or twenty-three,
or a hundred and one,

742
00:48:32 --> 00:48:34
if it's an odd one.

743
00:48:34 --> 00:48:36
Or an even number of
permutations,

744
00:48:36 --> 00:48:41
so, but that's the key fact
that just takes another little

745
00:48:41 --> 00:48:44
algebraic trick to see,
and that means that the

746
00:48:44 --> 00:48:48
determinant is well-defined by
properties one,

747
00:48:48 --> 00:48:51
two, three and it's got
properties

748
00:48:51 --> 00:48:52
four to ten.

749
00:48:52 --> 00:48:56
OK, that's today and I'll try
to get the homework for next

750
00:48:56 --> 00:48:59
Wednesday onto the web this
afternoon.

751
00:48:59 --> 00:49:02
Thanks.