1
00:00:10,740 --> 00:00:11,700
OK.

2
00:00:11,700 --> 00:00:17,050
This is linear algebra
lecture eleven.

3
00:00:17,050 --> 00:00:22,760
And at the end of lecture ten,
I was talking about some vector

4
00:00:22,760 --> 00:00:26,880
spaces, but they're --

5
00:00:26,880 --> 00:00:29,070
the things in
those vector spaces

6
00:00:29,070 --> 00:00:33,180
were not what we
usually call vectors.

7
00:00:33,180 --> 00:00:36,010
Nevertheless, you
could add them and you

8
00:00:36,010 --> 00:00:40,230
could multiply by numbers,
so we can call them vectors.

9
00:00:40,230 --> 00:00:41,970
I think the example
I was working

10
00:00:41,970 --> 00:00:44,100
with they were matrices.

11
00:00:44,100 --> 00:00:46,990
So the -- so we had
like a matrix space,

12
00:00:46,990 --> 00:00:50,230
the space of all three
by three matrices.

13
00:00:50,230 --> 00:00:55,530
And I'd like to just pick
up on that, because --

14
00:00:55,530 --> 00:01:00,600
we've been so specific about
n dimensional space here,

15
00:01:00,600 --> 00:01:05,640
and you really want to see that
the same ideas work as long

16
00:01:05,640 --> 00:01:09,270
as you can add and
multiply by scalars.

17
00:01:09,270 --> 00:01:16,640
So these new, new vector
spaces, the example I took

18
00:01:16,640 --> 00:01:24,085
was the space M of all
three by three matrices.

19
00:01:26,900 --> 00:01:29,230
OK.

20
00:01:29,230 --> 00:01:32,630
I can add them, I can
multiply by scalars.

21
00:01:32,630 --> 00:01:36,380
I can multiply two of them
together, but I don't do that.

22
00:01:36,380 --> 00:01:38,710
That's not part of the
vector space picture.

23
00:01:38,710 --> 00:01:43,420
The vector space part is
just adding the matrices

24
00:01:43,420 --> 00:01:45,690
and multiplying by numbers.

25
00:01:45,690 --> 00:01:50,540
And that's fine, we stay
within this space of three

26
00:01:50,540 --> 00:01:52,220
by three matrices.

27
00:01:52,220 --> 00:01:56,050
And I had some subspaces
that were interesting,

28
00:01:56,050 --> 00:02:02,850
like the symmetric, the
subspace of symmetric matrices,

29
00:02:02,850 --> 00:02:05,600
symmetric three by threes.

30
00:02:05,600 --> 00:02:13,420
Or the subspace of upper
triangular three by threes.

31
00:02:13,420 --> 00:02:21,540
Now I, I use the word subspace
because it follows the rule.

32
00:02:21,540 --> 00:02:25,240
If I add two symmetric
matrices, I'm still symmetric.

33
00:02:25,240 --> 00:02:27,680
If I multiply two
symmetric matrices,

34
00:02:27,680 --> 00:02:30,900
is the product
automatically symmetric?

35
00:02:30,900 --> 00:02:31,870
No.

36
00:02:31,870 --> 00:02:34,510
But I'm not
multiplying matrices.

37
00:02:34,510 --> 00:02:35,180
I'm just adding.

38
00:02:35,180 --> 00:02:36,900
So I'm fine.

39
00:02:36,900 --> 00:02:38,430
This is a subspace.

40
00:02:38,430 --> 00:02:42,320
Similarly, if I add two
upper triangular matrices,

41
00:02:42,320 --> 00:02:45,400
I'm still upper triangular.

42
00:02:45,400 --> 00:02:48,630
And, that's a subspace.

43
00:02:48,630 --> 00:02:52,900
Now I just want to take these
as example and ask, well,

44
00:02:52,900 --> 00:02:55,240
what's a basis
for that subspace?

45
00:02:55,240 --> 00:02:57,570
What's the dimension
of that subspace?

46
00:02:57,570 --> 00:03:00,900
And what's bd- dimension
of the whole space?

47
00:03:00,900 --> 00:03:07,840
So, there's a natural basis for
all three by three matrices,

48
00:03:07,840 --> 00:03:11,120
and why don't we
just write it down.

49
00:03:11,120 --> 00:03:21,370
So, so M, a basis for M.

50
00:03:21,370 --> 00:03:23,760
Again, all three by threes.

51
00:03:27,940 --> 00:03:30,380
OK.

52
00:03:30,380 --> 00:03:34,400
And then I'll just count how
many members are in that basis

53
00:03:34,400 --> 00:03:36,490
and I'll know the dimension.

54
00:03:36,490 --> 00:03:40,790
And OK, it's going to
take me a little time.

55
00:03:40,790 --> 00:03:43,720
In fact, what is the dimension?

56
00:03:43,720 --> 00:03:47,840
Any idea of what I'm
coming up with next?

57
00:03:47,840 --> 00:03:52,570
How many numbers does it
take to specify that three

58
00:03:52,570 --> 00:03:53,250
by three matrix?

59
00:03:53,250 --> 00:03:53,750
Nine.

60
00:03:56,310 --> 00:03:59,550
Nine is the, is the
dimension I'm going to find.

61
00:03:59,550 --> 00:04:04,630
And the most obvious basis
would be the matrix that's

62
00:04:04,630 --> 00:04:12,880
that matrix and then this
matrix with a one there

63
00:04:12,880 --> 00:04:21,490
and that's two of them,
shall I put in the third one,

64
00:04:21,490 --> 00:04:27,130
and then onwards, and
the last one maybe

65
00:04:27,130 --> 00:04:29,900
would end with the one.

66
00:04:29,900 --> 00:04:30,400
OK.

67
00:04:34,340 --> 00:04:36,240
That's like the standard basis.

68
00:04:36,240 --> 00:04:40,820
In fact, our space is
practically the same

69
00:04:40,820 --> 00:04:43,850
as nine dimensional space.

70
00:04:43,850 --> 00:04:48,310
It's just the nine numbers
are written in a square

71
00:04:48,310 --> 00:04:50,150
instead of in a column.

72
00:04:50,150 --> 00:04:54,510
But somehow it's different
and, and ought to be thought

73
00:04:54,510 --> 00:04:56,186
of as --

74
00:04:59,310 --> 00:05:01,280
natural for itself.

75
00:05:01,280 --> 00:05:05,760
Because now what about the
symmetric three by threes?

76
00:05:05,760 --> 00:05:09,100
So that's a subspace.

77
00:05:09,100 --> 00:05:12,300
Just let's just think, what's
the dimension of that subspace

78
00:05:12,300 --> 00:05:14,050
and what's a basis
for that subspace.

79
00:05:16,630 --> 00:05:17,820
OK.

80
00:05:17,820 --> 00:05:21,300
And I guess this
question occurs to me.

81
00:05:21,300 --> 00:05:24,830
If I look at this subspace
of symmetric three

82
00:05:24,830 --> 00:05:32,040
by threes, well, how many
of these original basis

83
00:05:32,040 --> 00:05:36,060
members belong to the subspace?

84
00:05:36,060 --> 00:05:38,050
I think only three of them do.

85
00:05:38,050 --> 00:05:40,660
This one is symmetric.

86
00:05:40,660 --> 00:05:43,490
This last one is symmetric.

87
00:05:43,490 --> 00:05:48,250
And the one in the middle with
a, with a one in that position

88
00:05:48,250 --> 00:05:52,210
-- in the two two position,
would be symmetric.

89
00:05:52,210 --> 00:05:56,680
But so I've got three of these
original nine are symmetric,

90
00:05:56,680 --> 00:06:03,800
but, so this is an
example where --

91
00:06:03,800 --> 00:06:05,600
but that's, that's
not all, right?

92
00:06:05,600 --> 00:06:06,610
What's the dimension?

93
00:06:06,610 --> 00:06:08,450
Let's put the dimensions down.

94
00:06:08,450 --> 00:06:13,580
Dimension of the,
of M, was nine.

95
00:06:13,580 --> 00:06:16,830
What's the dimension of --
shall we call this S -- is what?

96
00:06:19,630 --> 00:06:21,340
What's the dimension of this?

97
00:06:21,340 --> 00:06:25,660
I'm sort of taking simple
examples where we can, we can,

98
00:06:25,660 --> 00:06:30,130
spot the answer to
these questions.

99
00:06:30,130 --> 00:06:32,690
So how many -- if I
have a symmetric --

100
00:06:32,690 --> 00:06:36,750
think of all symmetric
matrices as a subspace,

101
00:06:36,750 --> 00:06:40,760
how many parameters do I choose
in three by three symmetric

102
00:06:40,760 --> 00:06:42,720
matrices?

103
00:06:42,720 --> 00:06:45,020
Six, right.

104
00:06:45,020 --> 00:06:49,220
If I choose the
diagonal that's three,

105
00:06:49,220 --> 00:06:53,340
and the three entries
above the diagonal,

106
00:06:53,340 --> 00:06:55,230
then I know what the
three entries below.

107
00:06:55,230 --> 00:06:58,550
So the dimension is six.

108
00:06:58,550 --> 00:07:00,820
I guess what's the
dimension of this here?

109
00:07:00,820 --> 00:07:05,180
Let's call this space
U for upper triangular.

110
00:07:05,180 --> 00:07:09,440
So what's the dimension of that
space of all upper triangular

111
00:07:09,440 --> 00:07:11,390
three by threes?

112
00:07:11,390 --> 00:07:13,870
Again six.

113
00:07:13,870 --> 00:07:14,585
Again six.

114
00:07:17,620 --> 00:07:24,490
And, but we haven't got a -- we
haven't seen -- well, actually,

115
00:07:24,490 --> 00:07:29,620
maybe we have got a basis here
for, the upper triangulars.

116
00:07:29,620 --> 00:07:35,070
I guess six of these guys,
one, two, three, four,

117
00:07:35,070 --> 00:07:39,120
and a, and a couple more,
would be upper triangular.

118
00:07:39,120 --> 00:07:44,390
So there's a accidental case
where the big basis contains

119
00:07:44,390 --> 00:07:46,930
in it a basis for the subspace.

120
00:07:46,930 --> 00:07:50,720
But with the symmetric
guy, it didn't have.

121
00:07:50,720 --> 00:07:53,570
The symmetric guy the,
basis -- so you see --

122
00:07:53,570 --> 00:07:56,690
a basis is the basis
for the big space,

123
00:07:56,690 --> 00:08:00,800
we generally need to think it
all over again to get a basis

124
00:08:00,800 --> 00:08:03,020
for the subspace.

125
00:08:03,020 --> 00:08:06,050
And then how do I
get other subspaces?

126
00:08:06,050 --> 00:08:15,100
Well, we spoke before about, the
subspace the symmetric matrices

127
00:08:15,100 --> 00:08:18,050
and the upper triangular.

128
00:08:18,050 --> 00:08:24,920
This is symmetric
and upper triangular.

129
00:08:30,280 --> 00:08:36,409
What's the, what's the
dimension of that space?

130
00:08:36,409 --> 00:08:36,909
OK.

131
00:08:36,909 --> 00:08:39,390
Well, what's in that space?

132
00:08:39,390 --> 00:08:42,480
So what's -- if a matrix
is symmetric and also upper

133
00:08:42,480 --> 00:08:46,950
triangular, that
makes it diagonal.

134
00:08:46,950 --> 00:08:51,640
So this is the same as
the diagonal matrices,

135
00:08:51,640 --> 00:08:55,230
diagonal three by threes.

136
00:08:55,230 --> 00:09:03,260
And the dimension of this,
of S intersect U, right --

137
00:09:03,260 --> 00:09:06,090
you're OK with that symbol?

138
00:09:06,090 --> 00:09:10,040
That's, that's the vectors
that are in both S and U,

139
00:09:10,040 --> 00:09:12,010
and that's D.

140
00:09:12,010 --> 00:09:15,320
So S intersect U
is the diagonals.

141
00:09:15,320 --> 00:09:21,250
And the dimension of the
diagonal matrices is three.

142
00:09:21,250 --> 00:09:24,880
And we've got a
basis, no problem.

143
00:09:24,880 --> 00:09:33,430
OK, as I write that, I think,
OK, what about putting --

144
00:09:33,430 --> 00:09:36,460
so this is like,
this intersection --

145
00:09:36,460 --> 00:09:40,820
is taking all the
vectors that are

146
00:09:40,820 --> 00:09:45,720
in both, that are symmetric
and also upper triangular.

147
00:09:45,720 --> 00:09:50,620
Now we looked at the union.

148
00:09:50,620 --> 00:09:55,070
Suppose I take the
matrices that are symmetric

149
00:09:55,070 --> 00:09:59,070
or upper triangular.

150
00:09:59,070 --> 00:10:01,620
What -- why was that no good?

151
00:10:01,620 --> 00:10:08,900
So why is it no -- why I
not interested in the union,

152
00:10:08,900 --> 00:10:13,060
putting together
those two subspaces?

153
00:10:13,060 --> 00:10:18,740
So this, these are matrices that
are in S or in U, or possibly

154
00:10:18,740 --> 00:10:21,760
both, so they, the
diagonals included.

155
00:10:21,760 --> 00:10:22,970
But what's bad about this?

156
00:10:22,970 --> 00:10:27,670
It's not a subspace.

157
00:10:27,670 --> 00:10:30,420
It's like having,
taking, you know,

158
00:10:30,420 --> 00:10:34,145
a couple of lines in the
plane and stopping there.

159
00:10:36,680 --> 00:10:40,550
A line -- this is -- so there's
a three dimensional subspace

160
00:10:40,550 --> 00:10:44,030
of a nine dimensional space,
there's -- ooh, sorry, six.

161
00:10:44,030 --> 00:10:46,150
There's a six
dimensional subspace

162
00:10:46,150 --> 00:10:47,910
of a nine dimensional space.

163
00:10:47,910 --> 00:10:49,410
There's another one.

164
00:10:49,410 --> 00:10:52,670
But they, they're headed
in different directions,

165
00:10:52,670 --> 00:10:55,360
so we, we can't just
put them together.

166
00:10:55,360 --> 00:10:56,990
We have to fill in.

167
00:10:56,990 --> 00:10:58,870
So that's what we do.

168
00:10:58,870 --> 00:11:03,000
To get this bigger space that
I'll write with a plus sign,

169
00:11:03,000 --> 00:11:09,590
this is combinations of
things in S and things in U.

170
00:11:09,590 --> 00:11:10,410
OK.

171
00:11:10,410 --> 00:11:15,300
So that's the final space
I'm going to introduce.

172
00:11:15,300 --> 00:11:17,420
I have a couple of subspaces.

173
00:11:17,420 --> 00:11:19,830
I can take their intersection.

174
00:11:19,830 --> 00:11:24,840
And now I'm interested in not
their union but their sum.

175
00:11:24,840 --> 00:11:28,130
So this would be the,
this is the intersection,

176
00:11:28,130 --> 00:11:31,580
and this will be their sum.

177
00:11:31,580 --> 00:11:35,195
So what do I need
for a subspace here?

178
00:11:38,010 --> 00:11:43,260
I take anything in S
plus anything in U.

179
00:11:43,260 --> 00:11:46,010
I don't just take things
that are in S and pop

180
00:11:46,010 --> 00:11:48,600
in also, separately,
things that are in U.

181
00:11:48,600 --> 00:11:56,190
This is the sum of
any element of S,

182
00:11:56,190 --> 00:12:02,310
that is, any symmetric
matrix, plus any

183
00:12:02,310 --> 00:12:08,450
in U, any element of U.

184
00:12:08,450 --> 00:12:09,050
OK.

185
00:12:09,050 --> 00:12:11,950
Now as long as we've
got an example here,

186
00:12:11,950 --> 00:12:15,020
tell me what we get.

187
00:12:15,020 --> 00:12:18,270
If I take every
symmetric matrix,

188
00:12:18,270 --> 00:12:20,310
take all symmetric
matrices, and add them

189
00:12:20,310 --> 00:12:22,770
to all upper
triangular matrices,

190
00:12:22,770 --> 00:12:27,520
then I've got a whole lot of
matrices and it is a subspace.

191
00:12:27,520 --> 00:12:30,170
And what's -- it's
a vector space,

192
00:12:30,170 --> 00:12:34,280
and what vector space
would I then have?

193
00:12:34,280 --> 00:12:36,670
Any idea what,
what matrices can I

194
00:12:36,670 --> 00:12:42,200
get out of a symmetric
plus an upper triangular?

195
00:12:42,200 --> 00:12:44,200
I can get anything.

196
00:12:44,200 --> 00:12:45,600
I get all matrices.

197
00:12:45,600 --> 00:12:50,300
I get all three by threes.

198
00:12:50,300 --> 00:12:54,380
It's worth thinking about that.

199
00:12:54,380 --> 00:12:58,330
It's just like stretch your
mind a little, just a little,

200
00:12:58,330 --> 00:13:05,700
to, to think of these subspaces
and what their intersection is

201
00:13:05,700 --> 00:13:07,110
and what their sum is.

202
00:13:07,110 --> 00:13:08,950
And now can I give you
a little -- oh, well,

203
00:13:08,950 --> 00:13:10,810
let's figure out the dimension.

204
00:13:10,810 --> 00:13:17,220
So what's the
dimension of S plus U?

205
00:13:17,220 --> 00:13:25,470
In this example is nine, because
we got all three by threes.

206
00:13:25,470 --> 00:13:31,690
So the original spaces had, the
original symmetric space had

207
00:13:31,690 --> 00:13:36,010
dimension six and the original
upper triangular space

208
00:13:36,010 --> 00:13:38,010
had dimension six.

209
00:13:38,010 --> 00:13:42,975
And actually I'm seeing
here a nice formula.

210
00:13:48,360 --> 00:13:54,810
That the dimension of S
plus the dimension of U --

211
00:13:54,810 --> 00:13:58,020
if I have two subspaces,
the dimension of one plus

212
00:13:58,020 --> 00:14:03,750
the dimension of the other
-- equals the dimension

213
00:14:03,750 --> 00:14:07,850
of their intersection plus
the dimension of their sum.

214
00:14:07,850 --> 00:14:12,530
Six plus six is three plus nine.

215
00:14:16,550 --> 00:14:20,850
That's kind of satisfying, that
these natural operations --

216
00:14:20,850 --> 00:14:23,360
and we've -- this
is it, actually,

217
00:14:23,360 --> 00:14:27,650
this is the set of
natural things to do with,

218
00:14:27,650 --> 00:14:29,390
with subspaces.

219
00:14:29,390 --> 00:14:36,730
That, the dimensions
come out in a good way.

220
00:14:36,730 --> 00:14:37,640
OK.

221
00:14:37,640 --> 00:14:43,540
Maybe I'll take just one more
example of a vector space

222
00:14:43,540 --> 00:14:48,270
that doesn't have vectors in it.

223
00:14:48,270 --> 00:14:51,296
It's come from
differential equations.

224
00:14:54,200 --> 00:14:57,120
So this is a one
more new vector space

225
00:14:57,120 --> 00:14:59,580
that we'll give just
a few minutes to.

226
00:14:59,580 --> 00:15:06,380
Suppose I have a differential
equation like d^2y/dx^2+ y=0.

227
00:15:09,590 --> 00:15:12,200
OK.

228
00:15:12,200 --> 00:15:14,175
I look at the solutions
to that equation.

229
00:15:16,790 --> 00:15:21,230
So what are the solutions
to that equation?

230
00:15:21,230 --> 00:15:24,815
y=cos(x) is a solution.

231
00:15:28,250 --> 00:15:31,960
y=sin(x) is a solution.

232
00:15:31,960 --> 00:15:37,730
y equals -- well, e to the (ix)
is a solution, if you want,

233
00:15:37,730 --> 00:15:41,150
if you allow me to put that in.

234
00:15:41,150 --> 00:15:42,520
But why should I put that in?

235
00:15:42,520 --> 00:15:45,890
It's already there.

236
00:15:45,890 --> 00:15:48,650
You see, I'm really looking
at a null space here.

237
00:15:48,650 --> 00:15:52,730
I'm looking at the null space
of a differential equation.

238
00:15:52,730 --> 00:15:55,210
That's the solution space.

239
00:15:55,210 --> 00:16:01,000
And describe the solution
space, all solutions

240
00:16:01,000 --> 00:16:03,290
to this differential equation.

241
00:16:03,290 --> 00:16:07,450
So the equation is y''+y=0.

242
00:16:07,450 --> 00:16:11,630
Cosine's, cosine's a
solution, sine is a solution.

243
00:16:11,630 --> 00:16:14,430
Now tell me all the solutions.

244
00:16:14,430 --> 00:16:18,930
They're -- so I
don't need e^(ix).

245
00:16:18,930 --> 00:16:21,220
Forget that.

246
00:16:21,220 --> 00:16:23,420
What are all the
complete solutions?

247
00:16:30,510 --> 00:16:32,860
Is what?

248
00:16:32,860 --> 00:16:34,840
A combination of these.

249
00:16:34,840 --> 00:16:38,070
The complete
solution is y equals

250
00:16:38,070 --> 00:16:48,350
some multiple of the cosine
plus some multiple of the sine.

251
00:16:48,350 --> 00:16:51,520
That's a vector space.

252
00:16:51,520 --> 00:16:52,660
That's a vector space.

253
00:16:52,660 --> 00:16:54,530
What's the dimension
of that space?

254
00:16:54,530 --> 00:16:57,500
What's a basis for that space?

255
00:16:57,500 --> 00:17:00,100
OK, let me ask
you a basis first.

256
00:17:00,100 --> 00:17:03,850
If I take the set of
solutions to that second order

257
00:17:03,850 --> 00:17:06,450
differential equation --

258
00:17:06,450 --> 00:17:10,520
there it is, those
are the solutions.

259
00:17:10,520 --> 00:17:12,010
What's a basis for that space?

260
00:17:14,680 --> 00:17:16,480
Now remember, what's
the, what question I

261
00:17:16,480 --> 00:17:16,980
asking?

262
00:17:16,980 --> 00:17:18,890
Because if you know the
question I'm asking,

263
00:17:18,890 --> 00:17:21,750
you'll see the answer.

264
00:17:21,750 --> 00:17:25,990
A basis means all
the guys in the space

265
00:17:25,990 --> 00:17:29,170
are combinations of
these basis vectors.

266
00:17:29,170 --> 00:17:31,320
Well, this is a basis.

267
00:17:31,320 --> 00:17:35,870
sin x, cos x there is a basis.

268
00:17:38,640 --> 00:17:44,090
Those two -- they're like
the special solutions, right?

269
00:17:44,090 --> 00:17:47,320
We had special
solutions to Ax=b.

270
00:17:47,320 --> 00:17:53,730
Now we've got special solutions
to differential equations.

271
00:17:53,730 --> 00:17:59,270
Sorry, we had special
solutions to Ax=0, I misspoke.

272
00:17:59,270 --> 00:18:01,760
The special solutions
were for the null space

273
00:18:01,760 --> 00:18:04,040
just as here we're talking
about the null space.

274
00:18:04,040 --> 00:18:07,360
Do you see that here
is a -- those two --

275
00:18:07,360 --> 00:18:13,170
and what's the dimension
of the solution space?

276
00:18:13,170 --> 00:18:23,360
How many vectors in this basis?

277
00:18:23,360 --> 00:18:27,430
Two, the sine and cosine.

278
00:18:27,430 --> 00:18:30,960
Are those the only
basis for this space?

279
00:18:30,960 --> 00:18:32,720
By no means.

280
00:18:32,720 --> 00:18:37,070
e^(ix) and e^(-ix)
would be another basis.

281
00:18:37,070 --> 00:18:38,220
Lots of bases.

282
00:18:38,220 --> 00:18:43,960
But do you see that really what
a course in differential --

283
00:18:43,960 --> 00:18:50,720
in linear differential equations
is about is finding a basis

284
00:18:50,720 --> 00:18:52,460
for the solution space.

285
00:18:52,460 --> 00:18:56,640
The dimension of the solution
space will always be --

286
00:18:56,640 --> 00:19:01,410
will be two, because we have
a second order equation.

287
00:19:01,410 --> 00:19:04,470
So that's, like
there's 18.03 in --

288
00:19:04,470 --> 00:19:10,310
five minutes of 18.06 is enough
to, to take care of 18.03.

289
00:19:10,310 --> 00:19:12,590
So there's a -- that's
one more example.

290
00:19:12,590 --> 00:19:13,090
OK.

291
00:19:13,090 --> 00:19:16,200
And of course the
point of the example

292
00:19:16,200 --> 00:19:24,020
is these things don't
look like vectors.

293
00:19:24,020 --> 00:19:26,080
They look like functions.

294
00:19:26,080 --> 00:19:31,520
But we can call them vectors,
because we can add them

295
00:19:31,520 --> 00:19:34,320
and we can multiply
by constants,

296
00:19:34,320 --> 00:19:36,470
so we can take
linear combinations.

297
00:19:36,470 --> 00:19:39,800
That's all we have
to be allowed to do.

298
00:19:39,800 --> 00:19:43,920
So that's really why this idea
of linear algebra and basis

299
00:19:43,920 --> 00:19:51,120
and dimension and so on
plays a wider role than --

300
00:19:51,120 --> 00:19:56,820
our constant discussions
of m by n matrices.

301
00:19:56,820 --> 00:19:57,520
OK.

302
00:19:57,520 --> 00:20:00,360
That's what I wanted to
say about that topic.

303
00:20:00,360 --> 00:20:09,740
Now of course the key, number
associated with matrices,

304
00:20:09,740 --> 00:20:13,510
to go back to that
number, is the rank.

305
00:20:13,510 --> 00:20:17,990
And the rank, what do
we know about the rank?

306
00:20:20,970 --> 00:20:23,360
Well, we know it's
not bigger than m

307
00:20:23,360 --> 00:20:25,310
and it's not bigger than n.

308
00:20:25,310 --> 00:20:29,060
So but I'd like to have a
little discussion on the rank.

309
00:20:29,060 --> 00:20:30,780
Maybe I'll put that here.

310
00:20:30,780 --> 00:20:33,895
So I'm picking up this
topic of rank one matrices.

311
00:20:38,360 --> 00:20:46,070
And the reason I'm interested
in rank one matrices

312
00:20:46,070 --> 00:20:48,900
is that they ought to be simple.

313
00:20:48,900 --> 00:20:55,450
If the rank is only one, the
matrix can't get away from

314
00:20:55,450 --> 00:20:55,950
us.

315
00:20:55,950 --> 00:20:59,840
So for example, let me take
-- let me create a rank one

316
00:20:59,840 --> 00:21:00,960
matrix.

317
00:21:00,960 --> 00:21:01,460
OK.

318
00:21:01,460 --> 00:21:04,840
Suppose it's three --
suppose it's two by three.

319
00:21:07,820 --> 00:21:10,943
And let me give
you the first row.

320
00:21:16,080 --> 00:21:19,750
What can the second row be?

321
00:21:19,750 --> 00:21:24,060
Tell me a possible second row
here, for, for this matrix

322
00:21:24,060 --> 00:21:27,570
to have rank one.

323
00:21:27,570 --> 00:21:30,380
A possible second row is?

324
00:21:30,380 --> 00:21:33,150
Two eight ten.

325
00:21:33,150 --> 00:21:41,390
The second row is a
multiple of the first row.

326
00:21:41,390 --> 00:21:43,380
It's not independent.

327
00:21:43,380 --> 00:21:45,580
So tell me a basis
for the -- oh yeah,

328
00:21:45,580 --> 00:21:50,250
sorry to keep bringing
up these same questions.

329
00:21:50,250 --> 00:21:53,750
After the quiz I'll
stop, but for now,

330
00:21:53,750 --> 00:21:57,200
tell me a basis
for the row space.

331
00:21:57,200 --> 00:22:03,470
A basis for the row space of
that matrix is the first row,

332
00:22:03,470 --> 00:22:03,980
right?

333
00:22:03,980 --> 00:22:06,860
The first row, one four five.

334
00:22:06,860 --> 00:22:11,020
A basis for the column
space of this matrix is?

335
00:22:11,020 --> 00:22:15,110
What's the dimension
of the column space?

336
00:22:15,110 --> 00:22:19,140
The dimension of the
column space is also one,

337
00:22:19,140 --> 00:22:19,640
right?

338
00:22:19,640 --> 00:22:21,130
Because it's also the rank.

339
00:22:21,130 --> 00:22:24,980
The dimension -- you remember
the dimension of the column

340
00:22:24,980 --> 00:22:33,090
space equals the rank equals the
dimension of the column space

341
00:22:33,090 --> 00:22:38,510
of the transpose, which
is the row space of A.

342
00:22:38,510 --> 00:22:44,620
OK, and in this case
it's one, r is one.

343
00:22:44,620 --> 00:22:48,080
And sure enough, all
the columns are --

344
00:22:48,080 --> 00:22:51,390
all the other columns are
multiples of that column.

345
00:22:51,390 --> 00:22:57,000
Now there's -- there ought
to be a nice way to see that,

346
00:22:57,000 --> 00:22:59,700
and here it is.

347
00:22:59,700 --> 00:23:06,530
I can write that matrix as
its pivot column, one two,

348
00:23:06,530 --> 00:23:10,520
times its --

349
00:23:10,520 --> 00:23:11,730
times one four five.

350
00:23:14,470 --> 00:23:18,570
A column times a row,
one column times one row

351
00:23:18,570 --> 00:23:21,340
gives me a matrix, right?

352
00:23:21,340 --> 00:23:25,370
If I multiply a
column by a row, that,

353
00:23:25,370 --> 00:23:30,500
g- that's a two by one matrix
times a one by three matrix,

354
00:23:30,500 --> 00:23:34,780
and the result of the
multiplication is two by three.

355
00:23:34,780 --> 00:23:36,630
And it comes out right.

356
00:23:36,630 --> 00:23:45,450
So what I want to -- my point
is the rank one matrices that

357
00:23:45,450 --> 00:23:52,830
every rank one matrix has the
form some column times some

358
00:23:52,830 --> 00:23:55,520
row.

359
00:23:55,520 --> 00:23:59,180
So U is a column vector,
V is a column vector --

360
00:23:59,180 --> 00:24:03,570
but I make it into a row
by putting in V transpose.

361
00:24:03,570 --> 00:24:12,330
So that's the -- complete
picture of rank one matrices.

362
00:24:12,330 --> 00:24:14,580
We'll be interested
in rank one matrices.

363
00:24:14,580 --> 00:24:19,270
Later we'll find, oh, their
determinant, that'll be easy,

364
00:24:19,270 --> 00:24:23,620
their eigenvalues,
that'll be interesting.

365
00:24:23,620 --> 00:24:26,450
Rank one matrices are
like the building blocks

366
00:24:26,450 --> 00:24:27,410
for all matrices.

367
00:24:27,410 --> 00:24:30,620
And actually maybe
you can guess.

368
00:24:37,220 --> 00:24:47,890
If I took any matrix, a five by
seventeen matrix of rank four,

369
00:24:47,890 --> 00:24:51,890
then it seems pretty
likely -- and it's true,

370
00:24:51,890 --> 00:24:56,060
that I could break that five
by seventeen matrix down

371
00:24:56,060 --> 00:24:59,840
as a combination of
rank one matrices.

372
00:24:59,840 --> 00:25:03,310
And probably how many
of those would I need?

373
00:25:03,310 --> 00:25:07,810
If I have a five by seventeen
matrix of rank four,

374
00:25:07,810 --> 00:25:11,300
I'll need four of them, right.

375
00:25:11,300 --> 00:25:13,490
Four rank one matrices.

376
00:25:13,490 --> 00:25:17,710
So the rank one matrices are
the, are the building blocks.

377
00:25:17,710 --> 00:25:23,680
And out -- I can produce every,
I can produce every five by --

378
00:25:23,680 --> 00:25:29,070
every rank four matrix out
of four rank one matrices.

379
00:25:29,070 --> 00:25:33,170
That brings me to a
question, of course.

380
00:25:33,170 --> 00:25:33,670
OK.

381
00:25:33,670 --> 00:25:36,406
Would the rank four
matrices form a subspace?

382
00:25:38,990 --> 00:25:43,450
Let me take all five by
seventeen matrices and think

383
00:25:43,450 --> 00:25:48,589
about rank four -- the
subset of rank four matrices.

384
00:25:48,589 --> 00:25:49,880
Let me -- I'll write this down.

385
00:25:53,670 --> 00:25:57,010
You seem I'm reviewing
for the quiz,

386
00:25:57,010 --> 00:26:00,800
because I'm asking the kind of
questions that are short enough

387
00:26:00,800 --> 00:26:04,780
but -- that bring out do you
know what these words mean.

388
00:26:04,780 --> 00:26:06,400
So I take --

389
00:26:06,400 --> 00:26:13,080
my matrix space M now is all
five by seventeen matrices.

390
00:26:13,080 --> 00:26:27,010
And now the question I ask is
the subset of, of rank four

391
00:26:27,010 --> 00:26:34,340
matrices, is that a subspace?

392
00:26:34,340 --> 00:26:38,650
If I add a matrix of -- so if I
multiply a matrix of rank four

393
00:26:38,650 --> 00:26:39,880
by --

394
00:26:39,880 --> 00:26:43,520
of rank four or less,
let's say, because I

395
00:26:43,520 --> 00:26:49,490
have to let the zero matrix in
if it's going to be a subspace.

396
00:26:49,490 --> 00:26:52,330
But, but that doesn't just
because the zero matrix

397
00:26:52,330 --> 00:26:56,610
got in there doesn't
mean I have a subspace.

398
00:26:56,610 --> 00:26:59,840
So if I -- so the, the question
really comes down to --

399
00:26:59,840 --> 00:27:05,605
if I add two rank four
matrices, is the sum rank four?

400
00:27:09,050 --> 00:27:09,800
What do you think?

401
00:27:12,730 --> 00:27:15,110
If -- no, not usually.

402
00:27:15,110 --> 00:27:16,540
Not usually.

403
00:27:16,540 --> 00:27:20,200
If I add two rank four
matrices, the sum is probably --

404
00:27:20,200 --> 00:27:22,750
what could I say about the sum?

405
00:27:22,750 --> 00:27:27,905
Well, actually, well,
the rank could be five.

406
00:27:30,790 --> 00:27:35,610
It's a general fact, actually,
that the rank of A plus B

407
00:27:35,610 --> 00:27:40,880
can't be more than rank
of A plus the rank of B.

408
00:27:40,880 --> 00:27:42,880
So this would say if
I added two of those,

409
00:27:42,880 --> 00:27:45,820
the rank couldn't be larger
than eight, but I know actually

410
00:27:45,820 --> 00:27:48,420
the rank couldn't be as
large as eight anyway.

411
00:27:48,420 --> 00:27:50,650
What -- how big
could the rank be,

412
00:27:50,650 --> 00:27:52,480
for, for the rank
of a matrix in M?

413
00:27:52,480 --> 00:27:57,450
Could be as large as
five, right, right.

414
00:27:57,450 --> 00:28:00,460
So they're all sort
of natural ideas.

415
00:28:00,460 --> 00:28:05,570
So it's rank four matrices
or rank one matrices --

416
00:28:05,570 --> 00:28:09,600
let me, let me change
that to rank one.

417
00:28:09,600 --> 00:28:12,670
Let me take the subset
of rank one matrices.

418
00:28:12,670 --> 00:28:15,660
Is that a vector space?

419
00:28:15,660 --> 00:28:19,860
If I add a rank one matrix
to a rank one matrix?

420
00:28:19,860 --> 00:28:20,460
No.

421
00:28:20,460 --> 00:28:23,050
It's most likely going
to have rank two.

422
00:28:23,050 --> 00:28:23,790
So this is --

423
00:28:23,790 --> 00:28:25,090
So I'll just make that point.

424
00:28:25,090 --> 00:28:27,690
Not a subspace.

425
00:28:32,030 --> 00:28:34,640
OK.

426
00:28:34,640 --> 00:28:35,140
OK.

427
00:28:35,140 --> 00:28:38,210
Those are topics that
I wanted to, just

428
00:28:38,210 --> 00:28:42,530
fill out the, the
previous lectures.

429
00:28:42,530 --> 00:28:46,730
The I'll ask one more
subspace question, a,

430
00:28:46,730 --> 00:28:50,570
a more, a more, likely example.

431
00:28:50,570 --> 00:28:53,751
Suppose I'm in -- let me put,
put this example on a new

432
00:28:53,751 --> 00:28:54,250
board.

433
00:28:56,850 --> 00:28:58,823
Suppose I'm in R, in R^4.

434
00:29:03,980 --> 00:29:10,850
So my typical vector in R^4 has
four components, v1, v2, v3,

435
00:29:10,850 --> 00:29:11,350
and v4.

436
00:29:15,560 --> 00:29:21,280
Suppose I take the
subspace of vectors

437
00:29:21,280 --> 00:29:23,850
whose components add to zero.

438
00:29:23,850 --> 00:29:34,070
So I let S be all v, all vectors
v in four dimensional space

439
00:29:34,070 --> 00:29:37,050
with v1+v2+v3+v4=0.

440
00:29:37,050 --> 00:29:39,320
So I just want to consider
that bunch of vectors.

441
00:29:39,320 --> 00:29:41,560
Is it a subspace, first of all?

442
00:29:41,560 --> 00:29:42,430
It is a subspace.

443
00:29:42,430 --> 00:29:57,880
It is a subspace.

444
00:29:57,880 --> 00:30:01,780
What's -- how do we see that?

445
00:30:01,780 --> 00:30:04,380
It is a subspace.

446
00:30:04,380 --> 00:30:06,890
I -- formally I should check.

447
00:30:06,890 --> 00:30:11,430
If I have one vector that with
whose components add to zero

448
00:30:11,430 --> 00:30:13,990
and I multiply that
vector by six --

449
00:30:13,990 --> 00:30:16,980
the components still add to
zero, just six times as --

450
00:30:16,980 --> 00:30:19,250
six times zero.

451
00:30:19,250 --> 00:30:22,840
If I have a couple of v
and a w and I add them,

452
00:30:22,840 --> 00:30:25,150
the, the components
still add to zero.

453
00:30:25,150 --> 00:30:27,170
OK, it's a subspace.

454
00:30:27,170 --> 00:30:29,500
What's the dimension
of that space

455
00:30:29,500 --> 00:30:32,730
and what's a basis
for that space?

456
00:30:32,730 --> 00:30:37,370
So you see how I can just
describe a space and we --

457
00:30:37,370 --> 00:30:41,530
we can ask for the dimension
-- ask for the basis first

458
00:30:41,530 --> 00:30:42,800
and the dimension.

459
00:30:42,800 --> 00:30:44,620
Of course, the
dimension's the one

460
00:30:44,620 --> 00:30:48,710
that's easy to tell
me in a single word.

461
00:30:48,710 --> 00:30:51,790
What's the dimension
of our subspace S here?

462
00:30:55,910 --> 00:30:57,940
And a basis tell me --

463
00:30:57,940 --> 00:31:01,250
some vectors in it.

464
00:31:01,250 --> 00:31:07,210
Well, I'm going to make ask you
again to guess the dimension.

465
00:31:07,210 --> 00:31:09,190
Again I think I heard it.

466
00:31:09,190 --> 00:31:11,270
The dimension is three.

467
00:31:11,270 --> 00:31:12,650
Three.

468
00:31:12,650 --> 00:31:18,380
Now how does this
connect to our Ax=0?

469
00:31:18,380 --> 00:31:21,370
Is this the null
space of something?

470
00:31:21,370 --> 00:31:24,500
Is that the null
space of a matrix?

471
00:31:24,500 --> 00:31:26,210
And then we can
look at the matrix

472
00:31:26,210 --> 00:31:30,240
and, and we know everything
about those subspaces.

473
00:31:30,240 --> 00:31:39,665
This is the null
space of what matrix?

474
00:31:47,010 --> 00:31:53,030
What's the matrix where the
null space is then Ab=0.

475
00:31:53,030 --> 00:31:57,195
So I want this
equation to be Ab=0.

476
00:31:59,920 --> 00:32:02,510
b is now the vector.

477
00:32:02,510 --> 00:32:07,850
And what's the matrix that,
that we're seeing there?

478
00:32:07,850 --> 00:32:14,580
It's the matrix of four ones.

479
00:32:14,580 --> 00:32:20,120
Do you see that that's -- that
if I look at Ab=0 for this

480
00:32:20,120 --> 00:32:27,180
matrix A, I multiply by b
and I get this requirement,

481
00:32:27,180 --> 00:32:29,280
that the components add to zero.

482
00:32:29,280 --> 00:32:33,670
So I'm really when
I speak about S --

483
00:32:33,670 --> 00:32:37,280
I'm speaking about the
null space of that matrix.

484
00:32:37,280 --> 00:32:38,050
OK.

485
00:32:38,050 --> 00:32:41,350
Let's just say we've
got a matrix now,

486
00:32:41,350 --> 00:32:44,000
we want its null space.

487
00:32:44,000 --> 00:32:47,610
Well, we -- tell
me its rank first.

488
00:32:47,610 --> 00:32:54,900
The rank of that
matrix is one, thanks.

489
00:32:54,900 --> 00:32:57,210
So r is one.

490
00:32:57,210 --> 00:32:59,610
What's the general
formula for the dimension

491
00:32:59,610 --> 00:33:01,530
of the null space?

492
00:33:01,530 --> 00:33:07,640
The dimension of the null
space of a matrix is --

493
00:33:07,640 --> 00:33:12,420
in general, an m by
n matrix of rank r?

494
00:33:12,420 --> 00:33:16,895
How many independent
guys in the null space?

495
00:33:21,140 --> 00:33:22,920
n-r, right?

496
00:33:22,920 --> 00:33:25,780
n-r.

497
00:33:25,780 --> 00:33:31,390
In this case, n is
four, four columns.

498
00:33:31,390 --> 00:33:35,330
The rank is one, so the null
space is three dimensions.

499
00:33:35,330 --> 00:33:39,440
So of course y- you could
see it in this case,

500
00:33:39,440 --> 00:33:44,160
but you can also see it
here in our systematic way

501
00:33:44,160 --> 00:33:49,450
of dealing with the four
fundamental subspaces

502
00:33:49,450 --> 00:33:51,730
of a matrix.

503
00:33:51,730 --> 00:33:55,410
So what actually what,
what are all four subspaces

504
00:33:55,410 --> 00:33:56,050
then?

505
00:33:56,050 --> 00:33:58,840
The row space is clear.

506
00:33:58,840 --> 00:34:01,420
The row space is in R^4.

507
00:34:01,420 --> 00:34:05,530
Yeah, can we take the
four fundamental subspaces

508
00:34:05,530 --> 00:34:06,910
of this matrix?

509
00:34:06,910 --> 00:34:08,815
Let's just kill this example.

510
00:34:11,860 --> 00:34:15,400
The row space is
one dimensional.

511
00:34:15,400 --> 00:34:19,830
It's all multiples
of that, of that row.

512
00:34:19,830 --> 00:34:22,780
The null space is
three dimensional.

513
00:34:22,780 --> 00:34:26,260
Oh, you better give me a
basis for the null space.

514
00:34:26,260 --> 00:34:28,080
So what's a basis
for the null space?

515
00:34:28,080 --> 00:34:30,489
The special solutions.

516
00:34:30,489 --> 00:34:34,830
To find the special solutions,
I look for the free variables.

517
00:34:34,830 --> 00:34:39,650
The free variables here
are -- there's the pivot.

518
00:34:39,650 --> 00:34:44,010
The free variables are
two, three, and four.

519
00:34:44,010 --> 00:34:52,730
So the basis, basis
for S, for S will be --

520
00:34:52,730 --> 00:35:01,710
I'm expecting three vectors,
three special solutions.

521
00:35:01,710 --> 00:35:06,470
I give the value one
to that free variable,

522
00:35:06,470 --> 00:35:12,000
and what's the pivot
variable if the --

523
00:35:12,000 --> 00:35:15,130
this is going to
be a vector in S?

524
00:35:15,130 --> 00:35:16,390
Minus one.

525
00:35:16,390 --> 00:35:20,250
Now they're always added to
-- the entries add to zero.

526
00:35:20,250 --> 00:35:22,870
The second special
solution has a one

527
00:35:22,870 --> 00:35:26,540
in the second free variable,
and again a minus one

528
00:35:26,540 --> 00:35:27,730
makes it right.

529
00:35:27,730 --> 00:35:30,940
The third one has a one in
the third free variable,

530
00:35:30,940 --> 00:35:34,910
and again a minus
one makes it right.

531
00:35:34,910 --> 00:35:35,780
That's my answer.

532
00:35:35,780 --> 00:35:39,770
That's the answer I
would be looking for.

533
00:35:39,770 --> 00:35:43,300
The -- a basis for
this subspace S,

534
00:35:43,300 --> 00:35:45,530
you would just
list three vectors,

535
00:35:45,530 --> 00:35:48,350
and those would be the
natural three to list.

536
00:35:48,350 --> 00:35:55,240
Not the only possible three,
but those are the special three.

537
00:35:55,240 --> 00:35:58,960
OK, tell me about
the column space,

538
00:35:58,960 --> 00:36:02,560
What's the column
space of this matrix A?

539
00:36:07,390 --> 00:36:13,500
So the column space
is a subspace of R^1,

540
00:36:13,500 --> 00:36:15,470
because m is only one.

541
00:36:15,470 --> 00:36:17,920
The columns only
have one component.

542
00:36:17,920 --> 00:36:23,360
So the column space of S, the
column space of A is somewhere

543
00:36:23,360 --> 00:36:26,410
in the space R^1,
because we only have --

544
00:36:26,410 --> 00:36:30,240
these columns are short.

545
00:36:30,240 --> 00:36:32,940
And what is the
column space actually?

546
00:36:36,330 --> 00:36:42,280
I just, it's just talking with
these words is what I'm doing.

547
00:36:42,280 --> 00:36:48,360
The column space for
that matrix is R^1.

548
00:36:48,360 --> 00:36:52,780
The column space
for that matrix is

549
00:36:52,780 --> 00:36:55,463
all multiples of that column.

550
00:36:58,410 --> 00:37:00,760
And all multiples
give you all of R^1.

551
00:37:03,630 --> 00:37:07,140
And what's the, the
remaining fourth space,

552
00:37:07,140 --> 00:37:12,680
the null space of A
transpose is what?

553
00:37:17,560 --> 00:37:21,880
So we transpose A.

554
00:37:21,880 --> 00:37:26,290
We look for combinations
of the columns

555
00:37:26,290 --> 00:37:30,310
now that give zero
for A transpose.

556
00:37:30,310 --> 00:37:32,190
And there aren't any.

557
00:37:32,190 --> 00:37:36,620
The only thing, the only
combination of these rows

558
00:37:36,620 --> 00:37:41,500
to give the zero row is
the zero combination.

559
00:37:41,500 --> 00:37:42,380
OK.

560
00:37:42,380 --> 00:37:44,415
So let's just check dimensions.

561
00:37:47,180 --> 00:37:51,320
The null space has
dimension three.

562
00:37:51,320 --> 00:37:53,420
The row space has dimension one.

563
00:37:53,420 --> 00:37:54,630
Three plus one is four.

564
00:37:57,180 --> 00:37:59,680
The column space
has dimension one,

565
00:37:59,680 --> 00:38:02,790
and what's the
dimension of this, like,

566
00:38:02,790 --> 00:38:05,510
smallest possible space?

567
00:38:05,510 --> 00:38:08,790
What's the dimension
of the zero space?

568
00:38:08,790 --> 00:38:09,675
It's a subspace.

569
00:38:13,520 --> 00:38:14,190
Zero.

570
00:38:14,190 --> 00:38:15,190
What else could it be?

571
00:38:15,190 --> 00:38:17,970
I mean, let's -- we have to
take a reasonable answer --

572
00:38:17,970 --> 00:38:20,190
and the only reasonable
answer is zero.

573
00:38:20,190 --> 00:38:25,900
So one plus zero gives -- this
was n, the number of columns,

574
00:38:25,900 --> 00:38:30,480
and this is m, the
number of rows.

575
00:38:30,480 --> 00:38:32,910
And let's just, let
me just say again

576
00:38:32,910 --> 00:38:37,030
then the, the, the subspace
that has only that one

577
00:38:37,030 --> 00:38:42,830
point, that point is zero
dimensional, of course.

578
00:38:42,830 --> 00:38:46,920
And the basis is empty, because
if the dimension is zero,

579
00:38:46,920 --> 00:38:49,480
there shouldn't be
anybody in the basis.

580
00:38:49,480 --> 00:38:56,160
So the basis of that smallest
subspace is the empty set.

581
00:38:56,160 --> 00:38:59,590
And the number of members
in the empty set is zero,

582
00:38:59,590 --> 00:39:01,400
so that's the dimension.

583
00:39:01,400 --> 00:39:02,610
OK.

584
00:39:02,610 --> 00:39:04,040
Good.

585
00:39:04,040 --> 00:39:10,740
Now I have just five
minutes to tell you about --

586
00:39:10,740 --> 00:39:16,010
well, actually, about some,
some, some, this is now,

587
00:39:16,010 --> 00:39:21,430
this last topic of small
world graphs, and leads into,

588
00:39:21,430 --> 00:39:27,140
a lecture about graphs
and linear algebra.

589
00:39:27,140 --> 00:39:29,700
But let me tell you --

590
00:39:29,700 --> 00:39:32,860
in these last minutes the
graph that I interested in.

591
00:39:32,860 --> 00:39:40,600
It's the graph where
-- so what is a graph?

592
00:39:40,600 --> 00:39:42,020
Better tell you that first.

593
00:39:42,020 --> 00:39:42,880
OK.

594
00:39:42,880 --> 00:39:43,516
What's a graph?

595
00:39:47,380 --> 00:39:48,180
OK.

596
00:39:48,180 --> 00:39:49,980
This isn't calculus.

597
00:39:49,980 --> 00:39:54,010
We're not, I'm not thinking
of, like, some sine curve.

598
00:39:54,010 --> 00:39:57,940
The word graph is used in
a completely different way.

599
00:39:57,940 --> 00:40:07,490
It's a set of, a bunch
of nodes and edges,

600
00:40:07,490 --> 00:40:10,060
edges connecting the nodes.

601
00:40:10,060 --> 00:40:17,280
So I have nodes like
five nodes and edges --

602
00:40:17,280 --> 00:40:21,880
I'll put in some edges, I
could put, include them all.

603
00:40:21,880 --> 00:40:23,970
There's -- well, let me
put in a couple more.

604
00:40:27,350 --> 00:40:31,380
There's a graph with five
nodes and one two three four

605
00:40:31,380 --> 00:40:34,690
five six edges.

606
00:40:34,690 --> 00:40:38,190
And some five by six
matrix is going to tell us

607
00:40:38,190 --> 00:40:41,100
everything about that graph.

608
00:40:41,100 --> 00:40:43,350
Let me leave that
matrix to next time

609
00:40:43,350 --> 00:40:46,350
and tell you about the
question I'm interested in.

610
00:40:46,350 --> 00:40:52,340
Suppose, suppose the
graph isn't just,

611
00:40:52,340 --> 00:40:56,230
just doesn't have just five
nodes, but suppose every,

612
00:40:56,230 --> 00:40:59,780
suppose every person
in this room is a node.

613
00:41:03,200 --> 00:41:07,310
And suppose there's an
edge between two nodes

614
00:41:07,310 --> 00:41:11,300
if those two people are friends.

615
00:41:11,300 --> 00:41:14,070
So have I described a graph?

616
00:41:14,070 --> 00:41:18,980
It's a pretty big graph,
hundred, hundred nodes.

617
00:41:18,980 --> 00:41:21,190
And I don't know how
many edges are in there.

618
00:41:24,920 --> 00:41:27,730
There's an edge
if you're friends.

619
00:41:27,730 --> 00:41:29,950
So that's the graph
for this class.

620
00:41:29,950 --> 00:41:35,120
A, a similar graph you could
take for the whole country,

621
00:41:35,120 --> 00:41:38,680
so two hundred and
sixty million nodes.

622
00:41:38,680 --> 00:41:43,340
And edges between friends.

623
00:41:43,340 --> 00:41:50,110
And the question for that graph
is how many steps does it take

624
00:41:50,110 --> 00:41:52,210
to get from anybody to anybody?

625
00:41:56,780 --> 00:42:01,950
What two people are furthest
apart in this friendship graph,

626
00:42:01,950 --> 00:42:04,070
say for the US?

627
00:42:04,070 --> 00:42:08,820
By furthest apart, I
mean the distance from --

628
00:42:08,820 --> 00:42:12,640
well, I'll tell you my
distance to Clinton.

629
00:42:12,640 --> 00:42:14,040
It's two.

630
00:42:14,040 --> 00:42:18,440
I happened to go to college
with somebody who knows Clinton.

631
00:42:18,440 --> 00:42:19,190
I don't know him.

632
00:42:19,190 --> 00:42:24,950
So my distance to Clinton is not
one, because I don't, happily

633
00:42:24,950 --> 00:42:26,680
or not, don't know him.

634
00:42:26,680 --> 00:42:29,050
But I know somebody who does.

635
00:42:29,050 --> 00:42:32,541
He's a Senator and so
I presume he knows him.

636
00:42:32,541 --> 00:42:33,040
OK.

637
00:42:33,040 --> 00:42:35,311
I don't know what your --
well, what's your distance

638
00:42:35,311 --> 00:42:35,810
to Clinton?

639
00:42:39,100 --> 00:42:40,950
Well, not more
than three, right.

640
00:42:40,950 --> 00:42:43,000
Actually, true.

641
00:42:43,000 --> 00:42:44,430
You know me.

642
00:42:44,430 --> 00:42:50,880
I take credit for reducing your
Clinton distance to three --

643
00:42:50,880 --> 00:42:52,195
what's your distance to Monica.

644
00:42:54,780 --> 00:43:04,090
Not, anybody below -- below
four is in trouble here.

645
00:43:04,090 --> 00:43:07,010
Or maybe three, but, right.

646
00:43:07,010 --> 00:43:14,740
So -- and what's Hillary's
distance to Monica?

647
00:43:14,740 --> 00:43:18,020
I don't think we'd better
put that on tape here.

648
00:43:18,020 --> 00:43:22,360
That's one or two, I guess.

649
00:43:22,360 --> 00:43:24,490
Is that right?

650
00:43:24,490 --> 00:43:28,980
I don't -- well, we won't,
think more about that.

651
00:43:28,980 --> 00:43:32,490
So actually, the,
the real question

652
00:43:32,490 --> 00:43:38,030
is what are large distances?

653
00:43:38,030 --> 00:43:41,910
How, how far apart could
people be separated?

654
00:43:41,910 --> 00:43:46,520
And roughly this number
six degrees of separation

655
00:43:46,520 --> 00:43:49,940
has kind of appeared as the
movie title, as the book title,

656
00:43:49,940 --> 00:43:52,050
and it's with this meaning.

657
00:43:52,050 --> 00:43:56,050
That roughly speaking --

658
00:43:56,050 --> 00:43:59,230
six might be a fairly --

659
00:43:59,230 --> 00:44:01,860
not too many people.

660
00:44:01,860 --> 00:44:04,860
If you sit next to
somebody on an airplane,

661
00:44:04,860 --> 00:44:07,850
you get talking to them.

662
00:44:07,850 --> 00:44:12,310
You begin to discuss mutual
friends to sort of find out,

663
00:44:12,310 --> 00:44:15,220
OK, what connections
do you have,

664
00:44:15,220 --> 00:44:17,450
and very often
you'll find you're

665
00:44:17,450 --> 00:44:21,880
connected in, like, two
or three or four steps.

666
00:44:21,880 --> 00:44:23,970
And you remark,
it's a small world,

667
00:44:23,970 --> 00:44:27,850
and that's how this expression
small world came up.

668
00:44:27,850 --> 00:44:31,450
But six, I don't know if you
could find -- if it took six,

669
00:44:31,450 --> 00:44:34,810
I don't know if you would
successfully discover those six

670
00:44:34,810 --> 00:44:36,980
in a, in an airplane
conversation.

671
00:44:36,980 --> 00:44:40,150
But here's the math
question, and I'll

672
00:44:40,150 --> 00:44:42,650
leave it for next,
for lecture twelve,

673
00:44:42,650 --> 00:44:46,010
and do a lot of linear
algebra in lecture twelve.

674
00:44:46,010 --> 00:44:54,860
But the interesting point is
that with a few shortcuts,

675
00:44:54,860 --> 00:44:58,290
the distances come
down dramatically.

676
00:44:58,290 --> 00:45:03,900
That, I mean, all your distances
to Clinton immediately drop

677
00:45:03,900 --> 00:45:06,460
to three by taking
linear algebra.

678
00:45:06,460 --> 00:45:11,580
That's, like, an extra bonus
for taking linear algebra.

679
00:45:11,580 --> 00:45:17,550
And to understand mathematically
what it is about these graphs

680
00:45:17,550 --> 00:45:18,120
--

681
00:45:18,120 --> 00:45:21,710
or like the graphs of
the World Wide Web.

682
00:45:21,710 --> 00:45:23,170
There's a fantastic graph.

683
00:45:23,170 --> 00:45:27,790
So many people would like to
understand and model the web.

684
00:45:27,790 --> 00:45:34,650
What the -- where the edges are
links and the nodes are, sites,

685
00:45:34,650 --> 00:45:37,530
websites.

686
00:45:37,530 --> 00:45:39,960
I'll leave you with that
graph, and I'll see you --

687
00:45:39,960 --> 00:45:42,740
have a good weekend,
and see you on Monday.