sm-transcript 0.0.6 → 0.0.7

data/test/results/18.06-03.srt

@@ -0,0 +1,2614 @@
+00:07.121
+I've been multiplying matrices
+already, but certainly time for
+
+00:12.725
+me to discuss the rules for
+matrix multiplication.
+
+00:17.227
+And the interesting part is the
+many ways you can do it,
+
+00:22.281
+and they all give the same
+answer.
+
+00:25.313
+And they're all important.
+
+00:28.529
+So matrix multiplication,
+and then, come inverses.
+
+00:32.488
+So we mentioned the inverse of
+a matrix.
+
+00:35.639
+That's a big deal.
+
+00:37.094
+Lots to do about inverses and
+how to find them.
+
+00:40.811
+Okay, so I'll begin with how to
+multiply two matrices.
+
+00:45.821
+First way, okay,
+so suppose I have a matrix A
+
+00:52.515
+multiplying a matrix B and --
+giving me a result -- well,
+
+01:01.035
+I could call it C.
+
+01:03.774
+A times B.
+
+01:05.295
+Okay.
+
+01:06.056
+So, let me just review the rule
+for this entry.
+
+01:13.054
+That's the entry in row i and
+column j.
+
+01:20.51
+So that's the i j entry.
+
+01:22.766
+Right there is C i j.
+
+01:24.74
+We always write the row number
+and then the column number.
+
+01:30.098
+So I might -- I might -- maybe
+I take it C 3 4,
+
+01:34.422
+just to make it specific.
+
+01:36.772
+So instead of i j,
+let me use numbers.
+
+01:40.25
+C 3 4.
+
+01:41.754
+So where does that come from,
+the three four entry?
+
+01:47.12
+It comes from row three,
+here, row three and column
+
+01:52.485
+four, as you know.
+
+01:54.416
+Column four.
+
+01:55.704
+And can I just write down,
+or can we write down the
+
+02:01.069
+formula for it?
+
+02:03.644
+If we look at the whole row and
+the whole column,
+
+02:08.247
+the quick way for me to say it
+is row three of A -- I could use
+
+02:14.192
+a dot for dot product.
+
+02:16.301
+I won't often use that,
+actually.
+
+02:19.369
+Dot column four of B.
+
+02:21.383
+But this gives us a chance to
+just, like, use a little matrix
+
+02:27.136
+notation.
+
+02:29.054
+What are the entries?
+
+02:31.441
+What's this first entry in row
+three?
+
+02:35.534
+That number that's sitting
+right there is...
+
+02:40.422
+A, so it's got two indices and
+what are they?
+
+02:46.221
+3 1.
+
+02:46.746
+So there's an a 3 1 there.
+
+02:50.165
+Now what's the first guy at the
+top of column four?
+
+02:56.739
+So what's sitting up there?
+
+03:00.289
+B 1 4, right.
+
+03:01.998
+So that this dot product starts
+with A 3 1 times B 1 4.
+
+03:10.15
+And then what's the next -- so
+this is like I'm accumulating
+
+03:16.156
+this sum, then comes the next
+guy, A 3 2, second column,
+
+03:21.755
+times B 2 4,
+second row.
+
+03:24.096
+So it's b A 3 2,
+B 2 4 and so on.
+
+03:27.354
+Just practice with indices.
+
+03:30.102
+Oh, let me even practice with a
+summation formula.
+
+03:36.211
+So this is -- most of the
+course, I use whole vectors.
+
+03:42.102
+I very seldom,
+get down to the details of
+
+03:46.549
+these particular entries,
+but here we'd better do it.
+
+03:52.329
+So it's some kind of a sum,
+right?
+
+03:56.887
+Of things in row three,
+column K shall I say?
+
+04:01.555
+Times things in row K,
+column four.
+
+04:05.163
+Do you see that that's what
+we're seeing here?
+
+04:09.937
+This is K is one,
+here K is two,
+
+04:13.226
+on along -- so the sum goes all
+the way along the row and down
+
+04:19.699
+the column, say,
+one to N.
+
+04:23.625
+So that's what the C three four
+entry looks like.
+
+04:28.399
+A sum of a three K b K four.
+
+04:31.185
+Just takes a little practice to
+do that.
+
+04:35.064
+Okay.
+
+04:35.561
+And -- well,
+maybe I should say -- when are
+
+04:39.739
+we allowed to multiply these
+matrices?
+
+04:44.415
+What are the shapes of these
+things?
+
+04:47.647
+The shapes are -- if we allow
+them to be not necessarily
+
+04:52.726
+square matrices.
+
+04:54.204
+If they're square,
+they've got to be the same
+
+04:58.267
+size.
+
+04:58.729
+If they're rectangular,
+they're not the same size.
+
+05:04.177
+If they're rectangular,
+this might be -- well,
+
+05:07.18
+I always think of A as m by n.
+m rows, n columns.
+
+05:10.383
+So that sum goes to n.
+
+05:11.85
+Now what's the point -- how
+many rows does B have to have?
+
+05:15.654
+n.
+
+05:15.787
+The number of rows in B,
+the number of guys that we meet
+
+05:19.457
+coming down has to match the
+number of ones across.
+
+05:23.593
+So B will have to be n by
+something.
+
+05:26.903
+Whatever.
+
+05:27.754
+P.
+
+05:27.943
+So the number of columns here
+has to match the number of rows
+
+05:33.617
+there, and then what's the
+result?
+
+05:36.737
+What's the shape of the result?
+
+05:39.668
+What's the shape of C,
+the output?
+
+05:42.789
+Well, it's got these same m
+rows -- it's got m rows.
+
+05:48.841
+And how many columns?
+
+05:51.027
+P.
+m by P.
+
+05:51.964
+Okay.
+
+05:52.485
+So there are m times P little
+numbers in there,
+
+05:57.275
+entries, and each one,
+looks like that.
+
+06:01.231
+Okay.
+
+06:01.752
+So that's the standard rule.
+
+06:04.668
+That's the way people think of
+multiplying matrices.
+
+06:11.228
+I do it too.
+
+06:12.922
+But I want to talk about other
+ways to look at that same
+
+06:20.689
+calculation, looking at whole
+columns and whole rows.
+
+06:28.031
+Okay.
+
+06:28.737
+So can I do A B C again?
+
+06:32.126
+A B equaling C again?
+
+06:36.221
+But now, tell me about...
+
+06:40.224
+I'll put it up here.
+
+06:43.426
+So here goes A,
+again, times B producing C.
+
+06:50.149
+And again, this is m by n.
+
+06:54.312
+This is n by P and this is m by
+P.
+
+06:59.595
+Okay.
+
+07:00.395
+Now I want to look at whole
+columns.
+
+07:07.6
+I want to look at the columns
+of -- here's the second way to
+
+07:12.836
+multiply matrices.
+
+07:14.434
+Because I'm going to build on
+what I know already.
+
+07:18.783
+How do I multiply a matrix by a
+column?
+
+07:22.156
+I know how to multiply this
+matrix by that column.
+
+07:27.303
+Shall I call that column one?
+
+07:29.858
+That tells me column one of the
+answer.
+
+07:33.206
+The matrix times the first
+column is that first column.
+
+07:37.963
+Because none of this stuff
+entered that part of the answer.
+
+07:43.073
+The matrix times the second
+column is the second column of
+
+07:48.094
+the answer.
+
+07:49.944
+Do you see what I'm saying?
+
+07:52.773
+That I could think of
+multiplying a matrix by a
+
+07:57.591
+vector, which I already knew how
+to do, and I can think of just P
+
+08:04.294
+columns sitting side by side,
+just like resting next to each
+
+08:10.474
+other.
+
+08:11.94
+And I multiply A times each one
+of those.
+
+08:14.775
+And I get the P columns of the
+answer.
+
+08:17.397
+Do you see this as -- this is
+quite nice, to be able to think,
+
+08:21.72
+okay, matrix multiplication
+works so that I can just think
+
+08:25.76
+of having several columns,
+multiplying by A and getting
+
+08:29.586
+the columns of the answer.
+
+08:32.208
+So, like, here's column one
+shall I call that column one?
+
+08:41.236
+And what's going in there is A
+times column one.
+
+08:48.813
+Okay.
+
+08:49.619
+So that's the picture a column
+at a time.
+
+08:57.197
+So what does that tell me?
+
+09:00.769
+What does that tell me about
+these columns?
+
+09:06.541
+These columns of C are
+combinations,
+
+09:11.35
+because we've seen that before,
+of columns of A.
+
+09:17.809
+Every one of these comes from A
+times this, and A times a vector
+
+09:26.466
+is a combination of the columns
+of A.
+
+09:32.924
+And it makes sense,
+because the columns of A have
+
+09:37.446
+length m and the columns of C
+have length m.
+
+09:41.496
+And every column of C is some
+combination of the columns of A.
+
+09:47.241
+And it's these numbers in here
+that tell me what combination it
+
+09:53.081
+is.
+
+09:54.211
+Do you see that?
+
+09:56.013
+That in that answer,
+C, I'm seeing stuff that's
+
+10:01.194
+combinations of these columns.
+
+10:04.572
+Now, suppose I look at it --
+that's two ways now.
+
+10:09.978
+The third way is look at it by
+rows.
+
+10:13.919
+So now let me change to rows.
+
+10:17.185
+Okay.
+
+10:18.875
+So now I can think of a row of
+A -- a row of A multiplying all
+
+10:28.038
+these rows here and producing a
+row of the product.
+
+10:35.55
+So this row takes a combination
+of these rows and that's the
+
+10:44.413
+answer.
+
+10:45.464
+So these rows of C are
+combinations of what?
+
+10:51.924
+Tell me how to finish that.
+
+10:57.482
+The rows of C,
+when I have a matrix B,
+
+11:00.883
+it's got its rows and I
+multiply by A,
+
+11:04.284
+and what does that do?
+
+11:06.305
+It mixes the rows up.
+
+11:08.235
+It creates combinations of the
+rows of B, thanks.
+
+11:12.647
+Rows of B.
+
+11:13.566
+That's what I wanted to see,
+that this answer --
+
+11:18.897
+I can see where the pieces are
+coming from.
+
+11:23.292
+The rows in the answer are
+coming as combinations of these
+
+11:29.258
+rows.
+
+11:29.782
+The columns in the answer are
+coming as combinations of those
+
+11:36.062
+columns.
+
+11:36.899
+And so that's three ways.
+
+11:40.457
+Now you can say,
+okay, what's the fourth way?
+
+11:46.462
+The fourth way -- so that's --
+now we've got,
+
+11:52.466
+like, the regular way,
+the column way,
+
+11:57.516
+the row way and -- what's left?
+
+12:01.746
+The one that I can -- well,
+one way is columns times rows.
+
+12:09.524
+What happens if I multiply --
+So this was row times column,
+
+12:18.155
+it gave a number.
+
+12:20.094
+Okay.
+
+12:20.665
+Now I want to ask you about
+column times row.
+
+12:25.684
+If I multiply a column of A
+times a row of B,
+
+12:30.704
+what shape I ending up with?
+
+12:33.898
+So if I take a column times a
+row, that's definitely different
+
+12:40.857
+from taking a row times a
+column.
+
+12:45.876
+So a column of A was -- what's
+the shape of a column of A?
+
+12:53.085
+n by one.
+
+12:54.223
+A column of A is a column.
+
+12:57.512
+It's got m entries and one
+column.
+
+13:01.685
+And what's a row of B?
+
+13:04.467
+It's got one row and P columns.
+
+13:08.388
+So what's the shape --
+what do I get if I multiply a
+
+13:16.017
+column by a row?
+
+13:18.062
+I get a big matrix.
+
+13:20.492
+I get a full-sized matrix.
+
+13:23.816
+If I multiply a column by a row
+-- should we just do one?
+
+13:30.977
+Let me take the column two
+three four times the row one
+
+13:37.881
+six.
+
+13:38.393
+That product there --
+I mean, when I'm just following
+
+13:44.359
+the rules of matrix
+multiplication,
+
+13:46.555
+those rules are just looking
+like -- kind of petite,
+
+13:49.849
+kind of small,
+because the rows here are so
+
+13:52.562
+short and the columns there are
+so short, but they're the same
+
+13:56.503
+length, one entry.
+
+13:57.665
+So what's the answer?
+
+13:59.732
+What's the answer if I do two
+three four times one six,
+
+14:05.35
+just for practice?
+
+14:07.222
+Well, what's the first row of
+the answer?
+
+14:11.383
+Two twelve.
+
+14:12.527
+And the second row of the
+answer is three eighteen.
+
+14:17.728
+And the third row of the answer
+is four twenty four.
+
+14:24.073
+That's a very special matrix,
+there.
+
+14:27.372
+Very special matrix.
+
+14:29.257
+What can you tell me about its
+columns, the columns of that
+
+14:34.723
+matrix?
+
+14:35.383
+They're multiples of this guy,
+right?
+
+14:38.776
+They're multiples of that one.
+
+14:41.603
+Which follows our rule.
+
+14:44.714
+We said that the columns of the
+answer were combinations,
+
+14:49.15
+but there's only -- to take a
+combination of one guy,
+
+14:53.269
+it's just a multiple.
+
+14:54.933
+The rows of the answer,
+what can you tell me about
+
+14:58.814
+those three rows?
+
+15:00.161
+They're all multiples of this
+row.
+
+15:03.567
+They're all multiples of one
+six, as we expected.
+
+15:11.05
+But I'm getting a full-sized
+matrix.
+
+15:16.506
+And now, just to complete this
+thought, if I have -- let me
+
+15:25.547
+write down the fourth way.
+
+15:29.6
+A B is a sum of columns of A
+times rows of B.
+
+15:37.861
+So that, for example,
+if my matrix was two three four
+
+15:43.219
+and then had another column,
+say, seven eight nine,
+
+15:48.371
+and my matrix here has -- say,
+started with one six and then
+
+15:54.449
+had another column like zero
+zero, then -- here's the fourth
+
+16:00.528
+way, okay?
+
+16:01.559
+I've got two columns there,
+I've got two rows there.
+
+16:07.947
+So the beautiful rule is --
+see, the whole thing by columns
+
+16:14.595
+and rows is that I can take the
+first column times the first row
+
+16:21.817
+and add the second column times
+the second row.
+
+16:27.089
+So that's the fourth way --
+that I can take columns times
+
+16:33.509
+rows, first column times first
+row, second column times second
+
+16:40.501
+row and add.
+
+16:43.138
+Actually, what will I get?
+
+16:45.561
+What will the answer be for
+that matrix multiplication?
+
+16:50.594
+Well, this one it's just going
+to give us zero,
+
+16:54.881
+so in fact I'm back to this --
+that's the answer,
+
+16:59.355
+for that matrix multiplication.
+
+17:02.99
+I'm happy to put up here these
+facts about matrix
+
+17:06.69
+multiplication,
+because it gives me a chance to
+
+17:10.237
+write down special matrices like
+this.
+
+17:13.089
+This is a special matrix.
+
+17:15.017
+All those rows lie on the same
+line.
+
+17:17.715
+All those rows lie on the line
+through one six.
+
+17:22.109
+If I draw a picture of all
+these row vectors,
+
+17:24.973
+they're all the same direction.
+
+17:26.991
+If I draw a picture of these
+two column vectors,
+
+17:30.05
+they're in the same direction.
+
+17:32.002
+Later, I would use this
+language.
+
+17:34.085
+Not too much later,
+either.
+
+17:35.777
+I would say the row space,
+which is like all the
+
+17:38.836
+combinations of the rows,
+is just a line for this matrix.
+
+17:43.392
+The row space is the line
+through the vector one six.
+
+17:48.836
+All the rows lie on that line.
+
+17:51.978
+And the column space is also a
+line.
+
+17:55.642
+All the columns lie on the line
+through the vector two three
+
+18:01.82
+four.
+
+18:02.344
+So this is like a really
+minimal matrix.
+
+18:07.474
+And it's because of these ones.
+
+18:11.157
+Okay.
+
+18:11.751
+So that's a third way.
+
+18:14.364
+Now I want to say one more
+thing about matrix
+
+18:19.591
+multiplication while we're on
+the subject.
+
+18:24.461
+And it's this.
+
+18:26.124
+You could also multiply --
+You could also cut the matrix
+
+18:33.121
+into blocks and do the
+multiplication by blocks.
+
+18:37.723
+Yet that's actually so,
+useful that I want to mention
+
+18:42.815
+it.
+
+18:43.108
+Block multiplication.
+
+18:45.165
+So I could take my matrix A and
+I could chop it up,
+
+18:50.06
+like, maybe just for
+simplicity, let me chop it into
+
+18:55.054
+two --
+into four square blocks.
+
+18:59.191
+Suppose it's square.
+
+19:01.17
+Let's just take a nice case.
+
+19:03.94
+And B, suppose it's square
+also, same size.
+
+19:08.095
+So these sizes don't have to be
+the same.
+
+19:12.053
+What they have to do is match
+properly.
+
+19:16.702
+Here they certainly will match.
+
+19:20.23
+So here's the rule for block
+multiplication,
+
+19:25.123
+that if this has blocks like,
+A -- so maybe A1,
+
+19:30.357
+A2, A3, A4 are the blocks here,
+and these blocks are B1,
+
+19:36.615
+B2,3 and B4?
+
+19:37.98
+Then the answer I can find
+block.
+
+19:42.759
+And if you tell me what's in
+that block, then I'm going to be
+
+19:46.042
+quiet about matrix
+multiplication for the rest of
+
+19:48.667
+the day.
+
+19:49.105
+What goes into that block?
+
+19:50.527
+You see, these might be -- this
+matrix might be -- these
+
+19:53.536
+matrices might be,
+like, twenty by twenty with
+
+19:55.998
+blocks that are ten by ten,
+to take the easy case where all
+
+19:59.17
+the blocks are the same shape.
+
+20:01.522
+And the point is that I could
+multiply those by blocks.
+
+20:08.591
+And what goes in here?
+
+20:11.47
+What's that block in the
+answer?
+
+20:15.528
+A1 B1, that's a matrix times a
+matrix, it's the right size,
+
+20:23.12
+ten by ten.
+
+20:25.606
+Any more?
+
+20:26.735
+Plus, what else goes in there?
+
+20:30.498
+A2 B3, right?
+
+20:32.128
+It's just like block rows times
+block columns.
+
+20:37.772
+Nobody, I think,
+not even Gauss could see
+
+20:42.789
+instantly that it works.
+
+20:46.803
+But somehow,
+if we check it through,
+
+20:50.29
+all five ways we're doing the
+same multiplications.
+
+20:55.271
+So this familiar multiplication
+is what we're really doing when
+
+21:01.448
+we do it by columns,
+by rows by columns times rows
+
+21:06.33
+and by blocks.
+
+21:08.622
+Okay.
+
+21:09.357
+I just have to,
+like, get the rules straight
+
+21:15.681
+for matrix multiplication.
+
+21:19.504
+Okay.
+
+21:20.24
+All right, I'm ready for the
+second topic,
+
+21:26.269
+which is inverses.
+
+21:28.916
+Okay.
+
+21:29.652
+Ready for inverses.
+
+21:32.446
+And let me do it for square
+matrices first.
+
+21:40.387
+Okay.
+
+21:40.848
+So I've got a square matrix A.
+
+21:43.612
+And it may or may not have an
+inverse, right?
+
+21:47.665
+Not all matrices have inverses.
+
+21:50.52
+In fact, that's the most
+important question you can ask
+
+21:55.494
+about the matrix,
+is if it's -- if you know it's
+
+21:59.824
+square, is it invertible or not?
+
+22:03.692
+If it is invertible,
+then there is some other
+
+22:10.081
+matrix, shall I call it A
+inverse?
+
+22:14.873
+And what's the -- if A inverse
+exists -- there's a big "if"
+
+22:23.295
+here.
+
+22:25.038
+If this matrix exists,
+and it'll be really central to
+
+22:31.143
+figure out when does it exist?
+
+22:34.665
+And then if it does exist,
+how would you find it?
+
+22:40.3
+But what's the equation here
+that I haven't -- that I have to
+
+22:47.345
+finish now?
+
+22:49.576
+This matrix,
+if it exists multiplies A and
+
+23:00.786
+produces, I think,
+the identity.
+
+23:10.356
+But a real -- an inverse for a
+square matrix could be on the
+
+23:17.481
+right as well -- this is true,
+too, that it's -- if I have a
+
+23:24.605
+-- yeah in fact,
+this is not -- this is probably
+
+23:30.28
+the --
+this is something that's not
+
+23:34.311
+easy to prove,
+but it works.
+
+23:36.684
+That a left -- square matrices,
+a left inverse is also a right
+
+23:42.046
+inverse.
+
+23:42.749
+If I can find a matrix on the
+left that gets the identity,
+
+23:47.759
+then also that matrix on the
+right will produce that
+
+23:52.242
+identity.
+
+23:54
+For rectangular matrices,
+we'll see a left inverse that
+
+23:58.691
+isn't a right inverse.
+
+24:00.602
+In fact, the shapes wouldn't
+allow it.
+
+24:03.817
+But for square matrices,
+the shapes allow it and it
+
+24:08.161
+happens, if A has an inverse.
+
+24:10.68
+Okay, so give me some cases --
+let's see.
+
+24:15.025
+I hate to be negative here,
+but let's talk about the case
+
+24:21.574
+with no inverse.
+
+24:23.445
+So -- these matrices are called
+invertible or non-singular --
+
+24:30.463
+those are the good ones.
+
+24:33.27
+And we want to be able to
+identify how -- if we're given a
+
+24:39.936
+matrix, has it got an inverse?
+
+24:44.498
+Can I talk about the singular
+case?
+
+24:49.132
+No inverse.
+
+24:50.631
+All right.
+
+24:51.994
+Best to start with an example.
+
+24:56.083
+Tell me an example -- let's get
+an example up here.
+
+25:02.898
+Let's make it two by two -- of
+a matrix that has not got an
+
+25:10.803
+inverse.
+
+25:11.894
+And let's see why.
+
+25:14.347
+Let me write one up.
+
+25:18.709
+No inverse.
+
+25:20.69
+Let's see why.
+
+25:23.211
+Let me write up -- one three
+two six.
+
+25:29.695
+Why does that matrix have no
+inverse?
+
+25:36.178
+You could answer that various
+ways.
+
+25:42.302
+Give me one reason.
+
+25:45.724
+Well, you could --
+if you know about determinants,
+
+25:54.11
+which you're not supposed to,
+you could take its determinant
+
+26:00.474
+and you would get -- Zero.
+
+26:03.278
+Okay.
+
+26:03.818
+Now -- all right.
+
+26:05.651
+Let me ask you other reasons.
+
+26:08.779
+I mean, as for other reasons
+that that matrix isn't
+
+26:14.172
+invertible.
+
+26:16.438
+Here, I could use what I'm
+saying here.
+
+26:21.318
+Suppose A times other matrix
+gave the identity.
+
+26:27.226
+Why is that not possible?
+
+26:30.437
+Because -- oh,
+yeah -- I'm thinking about
+
+26:35.574
+columns here.
+
+26:37.244
+If I multiply this matrix A by
+some other matrix,
+
+26:43.409
+then the --
+the result -- what can you tell
+
+26:48.883
+me about the columns?
+
+26:50.677
+They're all multiples of those
+columns, right?
+
+26:54.521
+If I multiply A by another
+matrix that -- the product has
+
+26:59.304
+columns that come from those
+columns.
+
+27:02.38
+So can I get the identity
+matrix?
+
+27:05.968
+No way.
+
+27:06.499
+The columns of the identity
+matrix, like one zero -- it's
+
+27:10.752
+not a combination of those
+columns, because those two
+
+27:14.702
+columns lie on the -- both lie
+on the same line.
+
+27:18.272
+Every combination is just going
+to be on that line and I can't
+
+27:22.905
+get one zero.
+
+27:24.652
+So, do you see that sort of
+column picture of the matrix not
+
+27:30.497
+being invertible.
+
+27:32.181
+In fact, here's another reason.
+
+27:35.252
+This is even a more important
+reason.
+
+27:38.819
+Well, how can I say more
+important?
+
+27:42.187
+All those are important.
+
+27:44.565
+This is another way to see it.
+
+27:48.528
+A matrix has no inverse -- yeah
+-- here -- now this is
+
+27:55.417
+important.
+
+27:56.718
+A matrix has no -- a square
+matrix won't have an inverse if
+
+28:04.258
+there's no inverse because I can
+solve -- I can find an X of -- a
+
+28:12.577
+vector X with A times -- this A
+times X giving zero.
+
+28:20.377
+This is the reason I like best.
+
+28:24.973
+That matrix won't have an
+inverse.
+
+28:29.866
+Can you -- well,
+let me change I to U.
+
+28:35.351
+So tell me a vector X that,
+solves A X equals zero.
+
+28:42.763
+I mean, this is,
+like, the key equation.
+
+28:49.88
+In mathematics,
+all the key equations have zero
+
+28:53.485
+on the right-hand side.
+
+28:55.287
+So what's the X?
+
+28:56.541
+Tell me an X here -- so now I'm
+going to put -- slip in the X
+
+29:01.244
+that you tell me and I'm going
+to get zero.
+
+29:04.536
+What X would do that job?
+
+29:07.201
+Three and negative one?
+
+29:09.088
+Is that the one you picked,
+or -- yeah.
+
+29:12.208
+Or another -- well,
+if you picked zero with zero,
+
+29:16.148
+I'm not so excited,
+right?
+
+29:18.2
+Because that would always work.
+
+29:20.745
+So it's really the fact that
+this vector isn't zero that's
+
+29:25.424
+important.
+
+29:27.147
+It's a non-zero vector and
+three negative one would do it.
+
+29:32.196
+That just says three of this
+column minus one of that column
+
+29:37.422
+is the zero column.
+
+29:39.105
+Okay.
+
+29:39.548
+So now I know that A couldn't
+be invertible.
+
+29:43.357
+But what's the reasoning?
+
+29:45.571
+If A X is zero,
+suppose I multiplied by A
+
+29:49.114
+inverse.
+
+29:50.885
+Yeah, well here's the reason.
+
+29:53.826
+Here -- this is why this spells
+disaster for an inverse.
+
+29:59.403
+The matrix can't have an
+inverse if some combination of
+
+30:04.878
+the columns gives z- it gives
+nothing.
+
+30:09.339
+Because, I could take A X
+equals zero, I could multiply by
+
+30:12.837
+A inverse and what would I
+discover?
+
+30:14.984
+Suppose I take that equation
+and I multiply by -- if A
+
+30:18.236
+inverse existed,
+which of course I'm going to
+
+30:20.935
+come to the conclusion it can't
+because if it existed,
+
+30:24.187
+if there was an A inverse to
+this dopey matrix,
+
+30:27.009
+I would multiply that equation
+by that inverse and I would
+
+30:30.506
+discover X is zero.
+
+30:32.593
+If I multiply A by A inverse on
+the left, I get X.
+
+30:38.802
+If I multiply by A inverse on
+the right, I get zero.
+
+30:45.265
+So I would discover X was zero.
+
+30:49.194
+But it -- X is not zero.
+
+30:52.996
+X -- this guy wasn't zero.
+
+30:54.943
+There it is.
+
+30:55.842
+It's three minus one.
+
+30:57.416
+So, conclusion -- only,
+it takes us some time to really
+
+31:01.462
+work with that conclusion -- our
+conclusion will be that
+
+31:05.582
+non-invertible matrices,
+singular matrices,
+
+31:08.729
+some combinations of their
+columns gives the zero column.
+
+31:13.749
+They they take some vector X
+into zero.
+
+31:17.999
+And there's no way A inverse
+can recover, right?
+
+31:23.255
+That's what this equation says.
+
+31:26.723
+This equation says I take this
+vector X and multiplying by A
+
+31:33.321
+gives zero.
+
+31:35.446
+But then when I multiply by A
+inverse, I can never escape from
+
+31:41.664
+zero.
+
+31:42.174
+So there couldn't be an A
+inverse.
+
+31:45.538
+Where here -- okay,
+now fix -- all right.
+
+31:49.615
+Now let me take -- all right,
+back to the positive side.
+
+31:56.138
+Let's take a matrix that does
+have an inverse.
+
+32:00.35
+And why not invert it?
+
+32:02.409
+Okay.
+
+32:02.877
+Can I -- so let me take on this
+third board a matrix -- shall I
+
+32:08.679
+fix that up a little?
+
+32:10.644
+Tell me a matrix that has got
+an inverse.
+
+32:14.387
+Well, let me say one three two
+-- what shall I put there?
+
+32:20.657
+Well, don't put six,
+I guess is -- right?
+
+32:27.307
+Do I any favorites here?
+
+32:31.296
+One?
+
+32:31.961
+Or eight?
+
+32:33.457
+I don't care.
+
+32:35.618
+What, seven?
+
+32:37.613
+Seven.
+
+32:38.61
+Okay.
+
+32:39.442
+Seven is a lucky number.
+
+32:45.094
+All right, seven,
+okay.
+
+32:46.649
+Okay.
+
+32:47.002
+So -- now what's our idea?
+
+32:48.84
+We believe that this matrix is
+invertible.
+
+32:51.738
+Those who like determinants
+have quickly taken its
+
+32:55.202
+determinant and found it wasn't
+zero.
+
+32:57.747
+Those who like columns,
+and probably that --
+
+33:01.634
+that department is not totally
+popular yet -- but those who
+
+33:06.827
+like columns will look at those
+two columns and say,
+
+33:11.394
+hey, they point in different
+directions.
+
+33:14.885
+So I can get anything.
+
+33:16.855
+Now, let me see,
+what do I mean?
+
+33:20.436
+How I going to computer A
+inverse?
+
+33:24.119
+So A inverse -- here's A
+inverse, now,
+
+33:28.249
+and I have to find it.
+
+33:30.704
+And what do I get when I do
+this multiplication?
+
+33:35.949
+The identity.
+
+33:37.4
+You know, forgive me for taking
+two by two-s,
+
+33:42.31
+but --
+lt's good to keep the
+
+33:46.542
+computations manageable and let
+the ideas come out.
+
+33:52.099
+Okay, now what's the idea I
+want?
+
+33:55.656
+I'm looking for this matrix A
+inverse, how I going to find it?
+
+34:02.436
+Right now, I've got four
+numbers to find.
+
+34:07.883
+I'm going to look at the first
+column.
+
+34:12.001
+Let me take this first column,
+A B.
+
+34:15.787
+What's up there?
+
+34:17.568
+What -- tell me this.
+
+34:19.906
+What equation does the first
+column satisfy?
+
+34:24.692
+The first column satisfies A
+times that column is one zero.
+
+34:32.262
+The first column of the answer.
+
+34:34.862
+And the second column,
+C D, satisfies A times that
+
+34:38.97
+second column is zero one.
+
+34:41.15
+You see that finding the
+inverse is like solving two
+
+34:45.426
+systems.
+
+34:46.097
+One system, when the right-hand
+side is one zero -- I'm just
+
+34:51.043
+going to split it into two
+pieces.
+
+34:54.733
+I don't even need to rewrite
+it.
+
+34:59.927
+I can take A times -- so let me
+put it here.
+
+35:07.131
+A times column j of A inverse
+is column j of the identity.
+
+35:16.681
+I've got n equations.
+
+35:20.2
+I've got, well,
+two in this case.
+
+35:27.07
+And they have the same matrix,
+A, but they have different
+
+35:30.917
+right-hand sides.
+
+35:32.084
+The right-hand sides are just
+the columns of the identity,
+
+35:36
+this guy and this guy.
+
+35:37.511
+And these are the two
+solutions.
+
+35:39.641
+Do you see what I'm going --
+I'm looking at that equation by
+
+35:43.694
+columns.
+
+35:45
+I'm looking at A times this
+column, giving that guy,
+
+35:48.348
+and A times that column giving
+that guy.
+
+35:50.909
+So -- Essentially -- so this is
+like the Gauss -- we're back to
+
+35:54.979
+Gauss.
+
+35:55.373
+We're back to solving systems
+of equations,
+
+35:58.131
+but we're solving -- we've got
+two right-hand sides instead of
+
+36:02.136
+one.
+
+36:03.186
+That's where Jordan comes in.
+
+36:08.371
+So at the very beginning of the
+lecture, I mentioned
+
+36:17.489
+Gauss-Jordan,
+let me write it up again.
+
+36:24.283
+Okay.
+
+36:25.177
+Here's the Gauss-Jordan idea.
+
+36:30.362
+Gauss-Jordan solve two
+equations at once.
+
+36:39.123
+Okay.
+
+36:39.793
+Let me show you how the
+mechanics go.
+
+36:44.624
+How do I solve a single
+equation?
+
+36:48.917
+So the two equations are one
+three two seven,
+
+36:54.821
+multiplying A B gives one zero.
+
+36:58.98
+And the other equation is the
+same one three two seven
+
+37:06.091
+multiplying C D gives zero one.
+
+37:11.726
+Okay.
+
+37:12.21
+That'll tell me the two columns
+of the inverse.
+
+37:16.66
+I'll have inverse.
+
+37:18.401
+In other words,
+if I can solve with this matrix
+
+37:22.85
+A, if I can solve with that
+right-hand side and that
+
+37:27.783
+right-hand side,
+I'm invertible.
+
+37:30.781
+I've got it.
+
+37:31.942
+Okay.
+
+37:33.586
+And Jordan sort of said to
+Gauss, solve them together,
+
+37:38.815
+look at the matrix -- if we
+just solve this one,
+
+37:43.452
+I would look at one three two
+seven, and how do I deal with
+
+37:49.174
+the right-hand side?
+
+37:51.147
+I stick it on as an extra
+column, right?
+
+37:54.995
+That's this augmented matrix.
+
+37:58.842
+That's the matrix when I'm
+watching the right-hand side at
+
+38:03.493
+the same time,
+doing the same thing to the
+
+38:06.838
+right side that I do to the
+left?
+
+38:09.449
+So I just carry it along as an
+extra column.
+
+38:12.957
+Now I'm going to carry along
+two extra columns.
+
+38:16.71
+And I'm going to do whatever
+Gauss wants, right?
+
+38:21.523
+I'm going to do elimination.
+
+38:23.891
+I'm going to get this to be
+simple and this thing will turn
+
+38:28.796
+into the inverse.
+
+38:30.233
+This is what's coming.
+
+38:32.093
+I'm going to do elimination
+steps to make this into the
+
+38:36.66
+identity, and lo and behold,
+the inverse will show up here.
+
+38:41.564
+K--- let's do it.
+
+38:43.002
+Okay.
+
+38:44.355
+So what are the elimination
+steps?
+
+38:46.541
+So you see -- here's my matrix
+A and here's the identity,
+
+38:50.252
+like, stuck on,
+augmented on.
+
+38:52.108
+STUDENT: I'm sorry...
+
+38:53.499
+STRANG: Yeah?
+
+38:54.361
+STUDENT: -- is the two and the
+three supposed to be switched?
+
+38:59
+STRANG: Did I -- oh,
+no, they weren't supposed to be
+
+39:03.775
+switched.
+
+39:04.618
+Sorry.
+
+39:05.179
+Thanks.
+
+39:05.835
+Okay.
+
+39:06.303
+Thank you very much.
+
+39:08.176
+And there -- I've got them
+right.
+
+39:11.172
+Okay, thanks.
+
+39:12.389
+Okay.
+
+39:12.857
+So let's do elimination.
+
+39:15.105
+All right, it's going to be
+simple, right?
+
+39:20.255
+So I take two of this row away
+from this row.
+
+39:23.616
+So this row stays the same and
+two of those come away from
+
+39:27.971
+this.
+
+39:28.353
+That leaves me with a zero and
+a one and two of these away from
+
+39:33.089
+this is that what you're getting
+-- after one elimination step --
+
+39:38.666
+Let me sort of separate the --
+the left half from the right
+
+39:43.815
+half.
+
+39:44.258
+So two of that first row got
+subtracted from the second row.
+
+39:49.495
+Now this is an upper triangular
+form.
+
+39:52.69
+Gauss would quit,
+but Jordan says keeps going.
+
+39:56.685
+Use elimination upwards.
+
+39:59.702
+Subtract a multiple of equation
+two from equation one to get rid
+
+40:04.293
+of the three.
+
+40:05.24
+So let's go the whole way.
+
+40:07.135
+So now I'm going to -- this guy
+is fine, but I'm going to --
+
+40:11.434
+what do I do now?
+
+40:12.673
+What's my final step that
+produces the inverse?
+
+40:16.025
+I multiply this by the right
+number to get up to ther to
+
+40:20.032
+remove that three.
+
+40:22.219
+So I guess, I -- since this is
+a one, there's the pivot sitting
+
+40:26.626
+there.
+
+40:27.053
+I multiply it by three and
+subtract from that,
+
+40:30.252
+so what do I get?
+
+40:31.46
+I'll have one zero -- oh,
+yeah that was my whole point.
+
+40:35.299
+I'll multiply this by three and
+subtract from that,
+
+40:38.853
+which will give me seven.
+
+40:41.413
+And I multiply this by three
+and subtract from that,
+
+40:50.784
+which gives me a minus three.
+
+40:56.113
+And what's my hope,
+belief?
+
+41:00.891
+Here I started with A and the
+identity, and I ended up with
+
+41:11.549
+the identity and who?
+
+41:16.878
+That better be A inverse.
+
+41:19.699
+That's the Gauss Jordan idea.
+
+41:22.972
+Start with this long matrix,
+double-length A I,
+
+41:28.164
+eliminate, eliminate until this
+part is down to I,
+
+41:33.695
+then this one will -- must be
+for some reason,
+
+41:38.773
+and we've got to find the
+reason -- must be A inverse.
+
+41:44.755
+Shall I just check that it
+works?
+
+41:49.722
+Let me just check that -- can I
+multiply this matrix this part
+
+41:54.113
+times A, I'll carry A over here
+and just do that multiplication.
+
+41:58.648
+You'll see I'll do it the old
+fashioned way.
+
+42:01.744
+Seven minus six is a one.
+
+42:03.544
+Twenty one minus twenty one is
+a zero, minus two plus two is a
+
+42:07.935
+zero, minus six plus seven is a
+one.
+
+42:11.246
+Check.
+
+42:11.871
+So that is the inverse.
+
+42:14.265
+That's the Gauss-Jordan idea.
+
+42:17.283
+So, you'll -- one of the
+homework problems or more than
+
+42:22.904
+one for Wednesday will ask you
+to go through those steps.
+
+42:28.733
+I think you just got to go
+through Gauss-Jordan a couple of
+
+42:34.769
+times, but I --
+yeah -- just to see the
+
+42:40.382
+mechanics.
+
+42:41.691
+But the, important thing is,
+why -- is, like,
+
+42:47.45
+what happened?
+
+42:49.283
+Why did we -- why did we get A
+inverse there?
+
+42:55.042
+Let me ask you that.
+
+42:57.66
+We got -- so we take --
+We do row reduction,
+
+43:04.155
+we do elimination on this long
+matrix A I until the first half
+
+43:11.19
+is up.
+
+43:11.882
+Then a second half is A
+inverse.
+
+43:15.458
+Well, how do I see that?
+
+43:18.226
+Let me put up here how I see
+that.
+
+43:22.032
+So here's my Gauss-Jordan
+thing, and I'm doing stuff to
+
+43:28.26
+it.
+
+43:29.989
+So I'm -- well,
+whole lot of E's.
+
+43:32.697
+Remember those are those
+elimination matrices.
+
+43:36.503
+Those are the -- those are the
+things that we figured out last
+
+43:41.663
+time.
+
+43:42.086
+Yes, that's what an elimination
+step is it's in matrix form,
+
+43:47.078
+I'm multiplying by some Es.
+
+43:50.208
+And the result -- well,
+so I'm multiplying by a whole
+
+43:54.728
+bunch of Es.
+
+43:55.772
+So, I get a -- can I call the
+overall matrix E?
+
+43:59.771
+That's the elimination matrix,
+the product of all those little
+
+44:05.074
+pieces.
+
+44:05.683
+What do I mean by little
+pieces?
+
+44:09.248
+Well, there was an elimination
+matrix that subtracted two of
+
+44:14.126
+that away from that.
+
+44:15.779
+Then there was an elimination
+matrix that subtracted three of
+
+44:20.74
+that away from that.
+
+44:22.394
+I guess in this case,
+that was all.
+
+44:25.867
+So there were just two Es in
+this case, one that did this
+
+44:30.872
+step and one that did this step
+and together they gave me an E
+
+44:36.324
+that does both steps.
+
+44:38.201
+And the net result was to get
+an I here.
+
+44:41.687
+And you can tell me what that
+has to be.
+
+44:45.173
+This is, like,
+the picture of what happened.
+
+44:50
+If E multiplied A,
+whatever that E is -- we never
+
+44:57.343
+figured it out in this way.
+
+45:01.474
+But whatever that E times that
+E is, E times A is -- What's E
+
+45:10.654
+times A?
+
+45:11.878
+It's I.
+
+45:14.019
+That E, whatever the heck it
+was, multiplied A and produced
+
+45:20.25
+I.
+
+45:20.465
+So E must be -- E A equaling I
+tells us what E is,
+
+45:25.729
+namely it is -- STUDENT:
+It's the inverse of A.
+
+45:30.67
+STRANG: It's the inverse of A.
+
+45:33.893
+Great.
+
+45:34.538
+And therefore,
+when the second half,
+
+45:38.298
+when E multiplies I,
+it's E --
+
+45:42.809
+Put this A inverse.
+
+45:44.09
+You see the picture looking
+that way?
+
+45:46.515
+E times A is the identity.
+
+45:48.266
+It tells us what E has to be.
+
+45:50.22
+It has to be the inverse,
+and therefore,
+
+45:52.848
+on the right-hand side,
+where E -- where we just
+
+45:56.014
+smartly tucked on the identity,
+it's turning in,
+
+45:59.181
+step by step --
+It's turning into A inverse.
+
+46:04.197
+There is the statement of
+Gauss-Jordan elimination.
+
+46:09.908
+That's how you find the
+inverse.
+
+46:13.449
+Where we can look at it as
+elimination, as solving n
+
+46:19.274
+equations at the same time --
+-- and tacking on n columns,
+
+46:26.682
+solving those equations and up
+goes the n columns of A inverse
+
+46:33.617
+.
+
+46:33.73
+Okay, thanks.
+
+46:35.208
+See you on Wednesday.