WEBVTT

00:00:06.310 --> 00:00:07.350
LINAN CHEN: Hi everyone.

00:00:07.350 --> 00:00:08.300
I'm Linan.

00:00:08.300 --> 00:00:10.300
Welcome back to recitation.

00:00:10.300 --> 00:00:13.090
In recent lectures, we
have studied the properties

00:00:13.090 --> 00:00:14.050
of the determinant.

00:00:14.050 --> 00:00:17.130
And we also derived the
formula to compute it.

00:00:17.130 --> 00:00:19.280
Today we're going to
put what we learned

00:00:19.280 --> 00:00:23.480
into practice by considering
these two examples.

00:00:23.480 --> 00:00:27.630
So we want to find out the
determinants of these two 5

00:00:27.630 --> 00:00:29.880
by 5 matrices.

00:00:29.880 --> 00:00:33.700
And as you can
see, matrix A has x

00:00:33.700 --> 00:00:38.990
along this diagonal, and in the
first four rows, y to the right

00:00:38.990 --> 00:00:41.750
of x, except for the last row.

00:00:41.750 --> 00:00:44.730
And zero entries
everywhere else.

00:00:44.730 --> 00:00:49.040
And matrix B also has
x along this diagonal

00:00:49.040 --> 00:00:52.220
and y everywhere else.

00:00:52.220 --> 00:00:55.910
Before starting, let me
help you review what you

00:00:55.910 --> 00:00:58.740
can do to compute determinants.

00:00:58.740 --> 00:01:01.810
Of course, you can
carry out elimination

00:01:01.810 --> 00:01:05.459
to transform your original
matrix into upper triangular

00:01:05.459 --> 00:01:07.040
matrix.

00:01:07.040 --> 00:01:11.570
Or you can use this
big summation formula.

00:01:11.570 --> 00:01:16.010
Another choice would be
you can do it by cofactors.

00:01:16.010 --> 00:01:19.220
Namely, you can expand
your original matrix

00:01:19.220 --> 00:01:23.090
along any row or any column,
and then the determinant

00:01:23.090 --> 00:01:25.580
is simply given
by the dot product

00:01:25.580 --> 00:01:30.150
of that row or that
column with its cofactors.

00:01:30.150 --> 00:01:33.370
Why don't you pause the video
now and try to work on them

00:01:33.370 --> 00:01:34.440
yourself.

00:01:34.440 --> 00:01:37.490
Whenever you're ready, I'll
come back and show you my way.

00:01:46.990 --> 00:01:49.850
I hope you just had some
fun with these two problems.

00:01:49.850 --> 00:01:52.290
Now let's look at them together.

00:01:52.290 --> 00:01:55.110
Let's look at matrix A first.

00:01:55.110 --> 00:01:59.370
As you can see, there are a lot
of zero entries in matrix A.

00:01:59.370 --> 00:02:02.280
So perhaps you don't need
elimination to introduce more

00:02:02.280 --> 00:02:04.270
0's.

00:02:04.270 --> 00:02:07.580
Furthermore, we observe
this pattern of A,

00:02:07.580 --> 00:02:12.640
and you notice that if
I cover the last row

00:02:12.640 --> 00:02:16.680
and the first column, so
if this column and this row

00:02:16.680 --> 00:02:20.730
are not here, what is left
over is simply a 4 by 4

00:02:20.730 --> 00:02:23.140
lower triangular matrix.

00:02:23.140 --> 00:02:27.530
And similarly, if you
cover the first column

00:02:27.530 --> 00:02:33.140
and the first row, what is
left over here is simply a 4

00:02:33.140 --> 00:02:36.030
by 4 upper triangular matrix.

00:02:36.030 --> 00:02:37.910
This is telling
us that we should

00:02:37.910 --> 00:02:42.850
calculate the determinant
of A by the third method.

00:02:42.850 --> 00:02:46.630
So we should expand along
the first column of A,

00:02:46.630 --> 00:02:48.920
and we calculate the cofactors.

00:02:48.920 --> 00:02:51.020
Let's do it now.

00:02:51.020 --> 00:03:00.110
So determinant of A,
is equal to the (1, 1)

00:03:00.110 --> 00:03:06.270
entry of A, which is
x, times the cofactor

00:03:06.270 --> 00:03:10.690
of that spot, which is the
determinant of the leftover 4

00:03:10.690 --> 00:03:13.080
by 4 matrix.

00:03:13.080 --> 00:03:16.200
And it's upper triangular,
so its determinant is simply

00:03:16.200 --> 00:03:21.450
given by x to the power 4.

00:03:21.450 --> 00:03:26.260
Plus, the only other
nonzero entry in that column

00:03:26.260 --> 00:03:28.860
is the y at the very bottom.

00:03:28.860 --> 00:03:32.530
So you put y here.

00:03:32.530 --> 00:03:35.390
And you multiply
y by the cofactor

00:03:35.390 --> 00:03:40.290
of that spot, which is the
determinant of the leftover 4

00:03:40.290 --> 00:03:42.130
by 4 matrix again.

00:03:42.130 --> 00:03:44.580
In this case, it's
lower triangular

00:03:44.580 --> 00:03:47.660
and its determinant
is y to the power 4.

00:03:47.660 --> 00:03:51.480
So I have a y to the power 4.

00:03:51.480 --> 00:03:53.520
But not quite.

00:03:53.520 --> 00:03:56.720
In principle, there
should be another factor

00:03:56.720 --> 00:03:59.150
here indicating the sign.

00:03:59.150 --> 00:04:03.430
And the sign in this
case, well because the y

00:04:03.430 --> 00:04:08.110
is the entry in the fifth
row and the first column,

00:04:08.110 --> 00:04:11.190
so this factor
should be negative 1

00:04:11.190 --> 00:04:14.670
to the power 5 plus 1.

00:04:14.670 --> 00:04:16.839
And of course, it's just 1.

00:04:16.839 --> 00:04:20.410
So the determinant
of A is simply

00:04:20.410 --> 00:04:29.780
equal to x to the fifth power
plus y to the fifth power.

00:04:29.780 --> 00:04:32.180
Did you get the correct answer?

00:04:32.180 --> 00:04:34.700
Well, the determinant
of A is not too bad,

00:04:34.700 --> 00:04:37.310
because A has a lot
of zero entries.

00:04:37.310 --> 00:04:42.060
Now let's look at the
determinant of matrix B.

00:04:42.060 --> 00:04:45.320
I have another copy of B here.

00:04:45.320 --> 00:04:48.710
So B also has a very
clear structure.

00:04:48.710 --> 00:04:53.080
It has x along its diagonal,
and y everywhere else.

00:04:53.080 --> 00:04:57.520
But in general, B does
not have any zero entry.

00:04:57.520 --> 00:05:02.630
So perhaps our first step should
be carrying out elimination

00:05:02.630 --> 00:05:07.000
to introduce zero
entries into matrix B.

00:05:07.000 --> 00:05:10.370
Of course, you can do it
by the regular routine.

00:05:10.370 --> 00:05:13.080
You start with the
first row, find a pivot,

00:05:13.080 --> 00:05:15.840
and eliminate the second
row and the third row

00:05:15.840 --> 00:05:17.750
and so on and so forth.

00:05:17.750 --> 00:05:20.870
But in this case,
there's a shortcut.

00:05:20.870 --> 00:05:24.630
If you compare two rows
that are next to each other,

00:05:24.630 --> 00:05:28.920
for example, if we compare the
fourth row and the fifth row,

00:05:28.920 --> 00:05:33.030
you notice that they have
a lot of entries in common.

00:05:33.030 --> 00:05:37.450
And they're only different
at these two spots.

00:05:37.450 --> 00:05:42.830
So imagine if I subtract the
fourth row from the fifth row.

00:05:42.830 --> 00:05:48.360
So if I do the
following operation--

00:05:48.360 --> 00:05:54.620
so I subtract this
row from the last row.

00:05:54.620 --> 00:06:00.290
Then the new fifth row
should become the following.

00:06:00.290 --> 00:06:09.580
So this row will become 0,
0, 0, y minus x, x minus y.

00:06:12.860 --> 00:06:15.710
You see, just by this
simple operation,

00:06:15.710 --> 00:06:18.830
I have introduced three
zero entries at once.

00:06:22.420 --> 00:06:25.640
And it's similar with the
fourth row and the third row.

00:06:25.640 --> 00:06:30.030
They have common entries
here, here, and here.

00:06:30.030 --> 00:06:33.670
So you subtract the third
row from the fourth row.

00:06:36.520 --> 00:06:46.410
You update the fourth row to
0, 0, y minus x, x minus y, 0.

00:06:46.410 --> 00:06:49.210
Again, three zero entries.

00:06:49.210 --> 00:06:52.990
And same thing happened to the
second row and the third row.

00:06:52.990 --> 00:06:56.780
So you subtract the second
row from the third row

00:06:56.780 --> 00:07:01.070
and your new third
row will become 0,

00:07:01.070 --> 00:07:08.290
y minus x, x minus y, 0, 0.

00:07:08.290 --> 00:07:14.860
Finally, you subtract the
first row from the second row,

00:07:14.860 --> 00:07:16.925
and then you update
the second row

00:07:16.925 --> 00:07:25.210
to y minus x, x
minus y, 0, 0, 0.

00:07:25.210 --> 00:07:27.190
Let's keep the
first row unchanged.

00:07:27.190 --> 00:07:28.720
So I'm going to copy here.

00:07:33.990 --> 00:07:34.630
All right.

00:07:34.630 --> 00:07:38.920
By row elimination, we have
introduced many zero entries

00:07:38.920 --> 00:07:42.540
to matrix B. Is
there anything else

00:07:42.540 --> 00:07:46.640
that I can take advantage of?

00:07:46.640 --> 00:07:50.090
Let's observe the pattern
of this new matrix.

00:07:50.090 --> 00:07:54.740
As you can see, in each row,
you have two nonzero entries,

00:07:54.740 --> 00:07:56.420
except for the first row.

00:07:56.420 --> 00:07:58.850
And they're only
different by a sign.

00:07:58.850 --> 00:08:02.760
So if, somehow, you can figure
out a way to sum them up,

00:08:02.760 --> 00:08:05.540
you will get even
more zero entries.

00:08:05.540 --> 00:08:06.900
So let's do it.

00:08:06.900 --> 00:08:10.930
That's going to involve
the operations on column.

00:08:10.930 --> 00:08:13.880
So here is how I do it.

00:08:16.720 --> 00:08:21.310
I'm going to keep the
last column unchanged.

00:08:21.310 --> 00:08:28.540
So the last column is
y, 0, 0, 0, x minus y.

00:08:32.870 --> 00:08:37.620
What I will do is I will add
a copy of the last column

00:08:37.620 --> 00:08:39.220
to the fourth column.

00:08:39.220 --> 00:08:41.230
So this is what I'm going to do.

00:08:41.230 --> 00:08:46.450
Add one copy of the last
column to the fourth column.

00:08:46.450 --> 00:08:57.890
Now the new fourth column will
become 2y, 0, 0, x minus y, 0.

00:08:57.890 --> 00:09:02.440
As you can see, by doing
this, I have killed this spot.

00:09:02.440 --> 00:09:05.160
So I have introduced
one more zero entry

00:09:05.160 --> 00:09:07.420
into my fourth column.

00:09:07.420 --> 00:09:11.870
If you continue, you may
want to add the fourth column

00:09:11.870 --> 00:09:13.200
to the third column.

00:09:13.200 --> 00:09:15.840
Let's see what will
happen if you do that.

00:09:15.840 --> 00:09:24.140
So if you add the fourth
column to the third column,

00:09:24.140 --> 00:09:35.311
now what should appear here is
2y, 0, x minus y, 0, y minus x.

00:09:38.770 --> 00:09:42.770
But in this new third
column, you still

00:09:42.770 --> 00:09:45.870
have two zero entries,
which is the same

00:09:45.870 --> 00:09:48.470
as the original third column.

00:09:48.470 --> 00:09:50.960
So although you've
killed this spot,

00:09:50.960 --> 00:09:55.070
but you've introduced
a new nonzero entry.

00:09:55.070 --> 00:10:00.260
So is there a way that we
can kill this spot too?

00:10:00.260 --> 00:10:06.000
You may have noticed that if you
add a copy of the fifth column

00:10:06.000 --> 00:10:11.560
to this column again, then that
spot will have been killed.

00:10:11.560 --> 00:10:12.780
So let's do it.

00:10:12.780 --> 00:10:15.800
If I add this
column to this one--

00:10:15.800 --> 00:10:18.140
I'm going to just
update it here-- then

00:10:18.140 --> 00:10:24.470
the first entry
should become 3y.

00:10:24.470 --> 00:10:27.060
These are unchanged.

00:10:27.060 --> 00:10:29.400
And the last spot becomes 0.

00:10:34.680 --> 00:10:38.190
It reflects here
as you are adding

00:10:38.190 --> 00:10:42.540
a copy of the fourth column
and a copy of the fifth column

00:10:42.540 --> 00:10:44.810
to the third column.

00:10:44.810 --> 00:10:46.560
Now you've got the idea.

00:10:46.560 --> 00:10:48.130
And you continue.

00:10:48.130 --> 00:10:51.050
What do you do with
the second column?

00:10:51.050 --> 00:10:58.550
This time you will have to add
a copy of the third column,

00:10:58.550 --> 00:11:02.020
a copy of the fourth column, and
the copy of the fifth column.

00:11:06.030 --> 00:11:17.040
So you update the second column
to be 4y, x minus y, 0, 0, 0.

00:11:19.630 --> 00:11:22.650
Eventually, what you will
do to the first column

00:11:22.650 --> 00:11:29.770
would be you add everything
to the first column.

00:11:29.770 --> 00:11:32.920
So a copy of each.

00:11:41.360 --> 00:11:43.710
Then the first
column will become--

00:11:43.710 --> 00:11:45.760
so I don't have
enough spot here,

00:11:45.760 --> 00:11:50.870
so make it here-- x plus 4y.

00:11:50.870 --> 00:11:56.070
Then everything else is 0.

00:11:56.070 --> 00:11:57.290
This is fun.

00:11:57.290 --> 00:11:59.110
And the result's really nice.

00:11:59.110 --> 00:12:01.260
This is wonderful
because this is simply

00:12:01.260 --> 00:12:03.080
upper triangular matrix.

00:12:03.080 --> 00:12:06.290
Now you tell me: what
is the determinant of B?

00:12:06.290 --> 00:12:12.070
The determinant of
B is the determinant

00:12:12.070 --> 00:12:14.240
of this upper triangular matrix.

00:12:14.240 --> 00:12:17.040
So you simply multiply
everything together,

00:12:17.040 --> 00:12:27.100
that's x plus 4y times x
minus y to the fourth power.

00:12:29.640 --> 00:12:33.480
So I hope you enjoyed
these two examples.

00:12:33.480 --> 00:12:35.980
Maybe your method is
different from mine,

00:12:35.980 --> 00:12:38.370
but at least these
two examples teach us

00:12:38.370 --> 00:12:41.390
that you can be flexible
in combining methods

00:12:41.390 --> 00:12:43.497
in calculating determinants.

00:12:43.497 --> 00:12:45.330
Thanks for watching,
and I'm looking forward

00:12:45.330 --> 00:12:47.280
to see you soon.