WEBVTT

00:00:06.739 --> 00:00:07.530
ANA RITA PIRES: Hi.

00:00:07.530 --> 00:00:09.132
Welcome back to recitation.

00:00:09.132 --> 00:00:11.340
In lecture, you've been
learning about the properties

00:00:11.340 --> 00:00:12.600
of determinants.

00:00:12.600 --> 00:00:15.040
To remember, there were
three main properties,

00:00:15.040 --> 00:00:17.680
and then seven more that
fall out of those three.

00:00:17.680 --> 00:00:19.309
I'll tell you what
these three were.

00:00:19.309 --> 00:00:21.350
The first one was the
determinant of the identity

00:00:21.350 --> 00:00:23.850
matrix is always equal to 1.

00:00:23.850 --> 00:00:25.780
If you switch two
rows in a matrix,

00:00:25.780 --> 00:00:27.730
the determinant switches sign.

00:00:27.730 --> 00:00:30.770
And the determinant
is a function

00:00:30.770 --> 00:00:34.630
of each-- it's a linear
function of each row separately.

00:00:34.630 --> 00:00:36.780
And there's seven more.

00:00:36.780 --> 00:00:38.430
We'll use them here.

00:00:38.430 --> 00:00:41.630
Today's problem is about finding
the determinants of matrices

00:00:41.630 --> 00:00:43.380
by using these properties.

00:00:43.380 --> 00:00:46.640
So here you have four matrices.

00:00:46.640 --> 00:00:50.600
A has lots of 100's,
200's, and 300's numbers.

00:00:50.600 --> 00:00:54.180
B is called a
Vandermonde matrix.

00:00:54.180 --> 00:00:57.560
It has a very nice structure
with 1's, and then a, b, c;

00:00:57.560 --> 00:00:59.130
a squared, b squared, c squared.

00:00:59.130 --> 00:01:00.830
It can be bigger,
and you'll just

00:01:00.830 --> 00:01:04.750
have cubes and more
letters down here.

00:01:04.750 --> 00:01:07.380
C is given by the
product of these two,

00:01:07.380 --> 00:01:11.510
and D is this matrix.

00:01:11.510 --> 00:01:12.050
Good luck.

00:01:12.050 --> 00:01:14.669
Hit pause, work on them, and
when you're ready, come back

00:01:14.669 --> 00:01:15.960
and I'll show you how I did it.

00:01:25.290 --> 00:01:26.440
Did you get some?

00:01:26.440 --> 00:01:28.910
OK, let's do it.

00:01:28.910 --> 00:01:36.770
Starting with matrix A. I
have lots of big numbers.

00:01:41.400 --> 00:01:44.960
I suggest that we do a
little bit of elimination,

00:01:44.960 --> 00:01:47.250
because as you know,
doing elimination steps,

00:01:47.250 --> 00:01:49.870
except for permuting
rows, doesn't change

00:01:49.870 --> 00:01:52.100
the determinant of the matrix.

00:01:52.100 --> 00:02:04.660
So let's do determinant of A
is equal to-- 101, 201, 301.

00:02:04.660 --> 00:02:08.580
Then if I subtract off the
first row from the second one,

00:02:08.580 --> 00:02:12.000
I'll get 1, 1, 1, which
is very convenient.

00:02:12.000 --> 00:02:15.490
And actually, if you
subtract the second row

00:02:15.490 --> 00:02:21.480
from the third one,
you'll get 1, 1, 1.

00:02:21.480 --> 00:02:23.690
Here's the property
of the determinant:

00:02:23.690 --> 00:02:25.990
if you have two equal
rows on your matrix,

00:02:25.990 --> 00:02:28.105
the determinant is
automatically equal to zero.

00:02:30.750 --> 00:02:31.310
All right.

00:02:31.310 --> 00:02:32.420
All done with one of them.

00:02:32.420 --> 00:02:35.660
Let's work on the second one.

00:02:35.660 --> 00:02:42.940
The determinant of B. Well,
let's try elimination again.

00:02:42.940 --> 00:02:47.310
1, a, a squared.

00:02:47.310 --> 00:02:53.700
1 minus 1 is 0, b minus a,
and b squared minus a squared.

00:02:53.700 --> 00:02:56.060
b squared minus a squared--
let me factor that

00:02:56.060 --> 00:03:01.420
into b minus a, b plus
a, which will be very

00:03:01.420 --> 00:03:03.440
convenient in the next step.

00:03:03.440 --> 00:03:06.290
And then I'm going to
subtract the first row again

00:03:06.290 --> 00:03:07.935
from the third one.

00:03:07.935 --> 00:03:10.430
I'll get 0, c minus a.

00:03:10.430 --> 00:03:13.550
And again, I'll get c
squared minus a squared, c

00:03:13.550 --> 00:03:16.742
minus a, c plus a.

00:03:23.350 --> 00:03:30.680
Let's use that third property
that was the determinant

00:03:30.680 --> 00:03:34.270
is linear on each
row separately.

00:03:34.270 --> 00:03:38.320
So what I'm going to do is,
see this factor of b minus a?

00:03:38.320 --> 00:03:41.170
It shows up in every
entry of this row.

00:03:41.170 --> 00:03:45.900
Well it's a 0, so it's the
zero multiple of b minus a.

00:03:45.900 --> 00:03:48.620
So I'm going to pull out
this factor of b minus a,

00:03:48.620 --> 00:03:53.710
and this row is going to
become 0, 1, b plus a.

00:03:53.710 --> 00:03:57.100
I will also, in the same
step, do the same thing

00:03:57.100 --> 00:03:58.230
with the third row.

00:03:58.230 --> 00:04:02.110
I'll pull out a
factor of c minus a,

00:04:02.110 --> 00:04:08.260
and it will become
0, 1, c plus a.

00:04:08.260 --> 00:04:11.540
Here's one factor
from the second row.

00:04:11.540 --> 00:04:14.640
Another factor
from the third row.

00:04:14.640 --> 00:04:18.420
And 1, a, a squared.

00:04:18.420 --> 00:04:22.200
0, 1, b plus a.

00:04:22.200 --> 00:04:25.410
0, 1, c plus a.

00:04:28.460 --> 00:04:29.760
Now what?

00:04:29.760 --> 00:04:32.250
Well remember,
you know how to do

00:04:32.250 --> 00:04:34.670
the determinant of upper
triangular matrices

00:04:34.670 --> 00:04:37.760
because all that you do
is multiply the pivots.

00:04:37.760 --> 00:04:40.610
This is almost upper triangular,
except there's a 1 over here.

00:04:40.610 --> 00:04:42.410
So let's do another
elimination step.

00:04:45.250 --> 00:04:50.170
b minus a, c minus a.

00:04:50.170 --> 00:05:00.850
1, a, a squared; 0, 1, b
plus a; 0, 0, c plus a minus

00:05:00.850 --> 00:05:06.400
b plus a is c minus b.

00:05:06.400 --> 00:05:10.540
So the determinant of this
matrix is now 1 times 1 times

00:05:10.540 --> 00:05:12.390
c minus b.

00:05:12.390 --> 00:05:20.670
So we get b minus a,
c minus a, c minus b,

00:05:20.670 --> 00:05:23.120
which has a really nice formula.

00:05:23.120 --> 00:05:25.540
This is called the
Vandermonde determinant.

00:05:25.540 --> 00:05:28.430
It's always like this, even if
your matrix is bigger than 3

00:05:28.430 --> 00:05:32.580
by 3, if it's 4 by 4, or
5 by 5 and so on, you just

00:05:32.580 --> 00:05:34.770
have more differences
of all the letters

00:05:34.770 --> 00:05:37.870
that show up in your matrix.

00:05:37.870 --> 00:05:41.190
On to matrix number 3.

00:05:41.190 --> 00:05:53.990
C equals [1, 2, 3] [1, -4, 5].

00:05:53.990 --> 00:05:58.010
How did you get the
determinant for this one?

00:05:58.010 --> 00:06:02.820
Well remember, this
is a rank one matrix

00:06:02.820 --> 00:06:05.900
because it's a column
vector times a row vector.

00:06:05.900 --> 00:06:08.850
So if you write out what
the matrix is it'll be a 3

00:06:08.850 --> 00:06:13.140
by 3 matrix where we can
think about it this way.

00:06:13.140 --> 00:06:17.210
The first row will be
1 times [1, -4, 5].

00:06:17.210 --> 00:06:20.260
The second row will be
2 times those numbers.

00:06:20.260 --> 00:06:23.870
And the third row will be
three times those numbers.

00:06:23.870 --> 00:06:30.960
So all the rows are going
to be linearly dependent,

00:06:30.960 --> 00:06:34.492
or another way of saying
the matrix is singular.

00:06:34.492 --> 00:06:36.560
When the matrix is
singular, the determinant

00:06:36.560 --> 00:06:37.750
is always equal to 0.

00:06:37.750 --> 00:06:40.580
That's also one
of the properties.

00:06:40.580 --> 00:06:42.490
C is 0.

00:06:45.930 --> 00:06:48.240
Onto the next matrix.

00:06:48.240 --> 00:06:49.690
Last one.

00:06:49.690 --> 00:06:56.660
D is equal to 0, 0, 0.

00:06:56.660 --> 00:07:06.750
1, -1, 3, -3, 4, -4.

00:07:06.750 --> 00:07:10.800
Maybe you didn't get that from
just looking at the matrix

00:07:10.800 --> 00:07:13.105
the first time, but did you
see how I wrote it down?

00:07:13.105 --> 00:07:15.180
It has 0's down the diagonal.

00:07:15.180 --> 00:07:20.130
And then for each entry,
I have minus that entry.

00:07:20.130 --> 00:07:22.660
That means that this
matrix is skew symmetric.

00:07:22.660 --> 00:07:25.940
What that means is if
you do D transpose,

00:07:25.940 --> 00:07:27.590
it will not be
equal to D, but it

00:07:27.590 --> 00:07:37.430
will be equal to minus D. D
transpose is equal to minus D.

00:07:37.430 --> 00:07:39.890
Now, what does this give
us for a determinant?

00:07:39.890 --> 00:07:43.640
Well if these two matrices
are the same matrix,

00:07:43.640 --> 00:07:46.830
this determinant is equal
to that determinant.

00:07:46.830 --> 00:07:49.750
One of the properties
is that the determinant

00:07:49.750 --> 00:07:53.780
of a transpose of the matrix
is equal to the determinant

00:07:53.780 --> 00:07:56.100
of the original matrix.

00:07:56.100 --> 00:07:57.460
How about this side?

00:07:57.460 --> 00:08:07.350
Well, first temptation
would be to just write that.

00:08:07.350 --> 00:08:09.330
Is that always true?

00:08:09.330 --> 00:08:09.870
No.

00:08:09.870 --> 00:08:14.120
The determinant is linear
on each row separately.

00:08:14.120 --> 00:08:16.650
That means that you can't
pull out the factor that

00:08:16.650 --> 00:08:18.010
is multiplying the matrix.

00:08:18.010 --> 00:08:21.330
You have to pull it out once
for each row of the matrix.

00:08:21.330 --> 00:08:31.560
So what I should
have written was

00:08:31.560 --> 00:08:34.179
-1, that's my factor, minus.

00:08:34.179 --> 00:08:35.370
How many rows do I have?

00:08:35.370 --> 00:08:37.490
One, two, three.

00:08:37.490 --> 00:08:39.690
Pull it out once for each row.

00:08:39.690 --> 00:08:44.990
Times the determinant of D. Well
fortunately, -1 to the third

00:08:44.990 --> 00:08:48.370
is simply equal to -1.

00:08:48.370 --> 00:08:50.860
So here we go.

00:08:50.860 --> 00:08:54.490
It was correct, in fact.

00:08:54.490 --> 00:08:57.190
We have determinant of D is
equal to minus determinant

00:08:57.190 --> 00:08:59.380
of D. What is the
only number that

00:08:59.380 --> 00:09:01.970
is equal to minus that number?

00:09:01.970 --> 00:09:03.580
0.

00:09:03.580 --> 00:09:08.991
Determinant of D is
equal to 0 again.

00:09:08.991 --> 00:09:11.830
Let me ask you
one last question.

00:09:11.830 --> 00:09:14.330
Is it true that all skew
symmetric matrices have

00:09:14.330 --> 00:09:16.355
the determinant equal to 0?

00:09:16.355 --> 00:09:18.190
It was true for this one.

00:09:18.190 --> 00:09:21.010
Is it true in every case?

00:09:21.010 --> 00:09:24.490
Well, the key factor
here was that I

00:09:24.490 --> 00:09:29.260
had -1 to the third power
and I got a minus sign here,

00:09:29.260 --> 00:09:32.215
determinant of D is equal
to minus determinant of D.

00:09:32.215 --> 00:09:34.920
What if this number had
been an even number?

00:09:34.920 --> 00:09:37.500
Then I would just have
the determinant of D

00:09:37.500 --> 00:09:42.350
is equal minus 1 to an
even number, D. That's 1.

00:09:42.350 --> 00:09:46.420
So I would have determinant of
D is equal to determinant of D.

00:09:46.420 --> 00:09:48.380
There's nothing I can
say about that number.

00:09:48.380 --> 00:09:51.505
It can be your favorite
number, not necessarily 0.

00:09:51.505 --> 00:09:52.510
All right.

00:09:52.510 --> 00:09:53.500
We're done for today.

00:09:53.500 --> 00:09:54.848
Thank you.