WEBVTT

00:00:08.320 --> 00:00:10.134
PROFESSOR: Hi, and welcome.

00:00:10.134 --> 00:00:11.550
Today, we're going
to do a problem

00:00:11.550 --> 00:00:15.630
about powers of a matrix.

00:00:15.630 --> 00:00:21.270
Our problem is first to find
a formula for the k-th power

00:00:21.270 --> 00:00:24.510
of this matrix C. This is
a two by two matrix that

00:00:24.510 --> 00:00:27.834
depends on variables a and b.

00:00:27.834 --> 00:00:29.250
And the second
part of our problem

00:00:29.250 --> 00:00:34.210
is to calculate C to the
100th in the special case

00:00:34.210 --> 00:00:38.710
where a and b are -1.

00:00:38.710 --> 00:00:40.840
You can hit pause now,
and I'll give you a minute

00:00:40.840 --> 00:00:42.990
to do the problem yourself.

00:00:42.990 --> 00:00:45.210
And then, I'll come back
and we can do it together.

00:00:57.620 --> 00:00:58.720
OK.

00:00:58.720 --> 00:01:00.220
We're back.

00:01:00.220 --> 00:01:05.980
Now, what's the first step in
finding powers of a matrix?

00:01:05.980 --> 00:01:09.130
Well, we need to find the
eigenvalues and eigenvectors

00:01:09.130 --> 00:01:10.900
of this matrix.

00:01:10.900 --> 00:01:13.030
So how do we do that?

00:01:13.030 --> 00:01:22.840
We compute the determinant of
C minus lambda I, which is just

00:01:22.840 --> 00:01:35.090
the determinant of this matrix:
2b minus a minus lambda,

00:01:35.090 --> 00:01:47.320
2b minus 2a; a minus b, and
2a minus b minus lambda.

00:01:47.320 --> 00:01:48.780
OK.

00:01:48.780 --> 00:01:54.940
If you compute this, well, we
have a lambda squared term.

00:01:54.940 --> 00:01:56.760
OK.

00:01:56.760 --> 00:01:59.500
Our lambda term,
if you look at it,

00:01:59.500 --> 00:02:06.800
you'll see we get 2b
minus a plus 2a minus b,

00:02:06.800 --> 00:02:08.470
which is just a plus b.

00:02:08.470 --> 00:02:10.680
And we have a negative sign.

00:02:10.680 --> 00:02:15.080
And it's negative a
plus b times lambda.

00:02:15.080 --> 00:02:18.930
And our last term is a
little tougher to compute.

00:02:18.930 --> 00:02:21.040
So I'll let you do it yourself.

00:02:21.040 --> 00:02:25.390
But you're just going
to get plus a*b.

00:02:25.390 --> 00:02:32.390
And this will factor as lambda
minus a times lambda minus b.

00:02:32.390 --> 00:02:38.820
So our eigenvalues
are just a and b.

00:02:38.820 --> 00:02:41.370
Now we need to find
our eigenvectors.

00:02:41.370 --> 00:02:43.720
So how do we do that?

00:02:43.720 --> 00:02:46.170
Well, what we need
to do is we need

00:02:46.170 --> 00:02:51.230
to look at C minus a
times the identity.

00:02:51.230 --> 00:02:54.770
And we need to find the
null space of this matrix.

00:02:54.770 --> 00:02:58.360
So what do we get here?

00:02:58.360 --> 00:03:02.010
We get 2b minus 2a.

00:03:07.330 --> 00:03:13.080
And then our next entry here,
we get 2a minus b minus a.

00:03:13.080 --> 00:03:15.810
So this is a minus b.

00:03:15.810 --> 00:03:16.850
Good.

00:03:16.850 --> 00:03:27.310
So you can see that this
matrix has the same columns

00:03:27.310 --> 00:03:30.740
and the same rows.

00:03:30.740 --> 00:03:37.670
And so you can see that a
vector in the null space,

00:03:37.670 --> 00:03:43.690
since this column is
-2 times this column,

00:03:43.690 --> 00:03:48.370
we can see that our first
eigenvector is just-- or 1, 2,

00:03:48.370 --> 00:03:51.050
I should say.

00:03:51.050 --> 00:03:54.964
It's just [1, 2].

00:03:54.964 --> 00:03:55.464
Good.

00:03:58.220 --> 00:04:02.280
Well, I guess we have space
to do the second one too.

00:04:02.280 --> 00:04:04.950
Why not?

00:04:04.950 --> 00:04:08.780
So let's write out
the second one also.

00:04:08.780 --> 00:04:11.976
Here, we're subtracting
b instead of a.

00:04:11.976 --> 00:04:14.020
You get b minus a.

00:04:14.020 --> 00:04:17.620
You get 2b minus 2a.

00:04:17.620 --> 00:04:20.870
We get a minus b.

00:04:20.870 --> 00:04:22.430
And what do we have here?

00:04:22.430 --> 00:04:25.071
We have 2a minus 2b.

00:04:28.400 --> 00:04:33.000
So now, what's in the
null space of this matrix?

00:04:33.000 --> 00:04:40.780
Well, what you can see
is that this column

00:04:40.780 --> 00:04:43.090
is -1 times that column.

00:04:43.090 --> 00:04:50.910
So our second eigenvector
is just going to be [1, 1].

00:04:50.910 --> 00:04:54.020
And I should remind you that
if you have a harder example,

00:04:54.020 --> 00:04:57.750
you can just find these
null spaces by elimination

00:04:57.750 --> 00:04:59.490
like we always do.

00:04:59.490 --> 00:05:00.040
Great.

00:05:00.040 --> 00:05:02.610
Now we have our eigenvalues
and our eigenvectors.

00:05:02.610 --> 00:05:05.400
So now we can write C
in a nice easy way that

00:05:05.400 --> 00:05:09.810
allows us to take powers of it.

00:05:09.810 --> 00:05:11.300
So what's that way?

00:05:11.300 --> 00:05:21.060
So that's C equals
S lambda S inverse.

00:05:21.060 --> 00:05:24.910
So this is just, what is S?

00:05:24.910 --> 00:05:29.750
Remember, S is our
matrix of eigenvectors.

00:05:29.750 --> 00:05:36.700
So S is the matrix 1, 2; 1, 1.

00:05:36.700 --> 00:05:37.680
Good.

00:05:37.680 --> 00:05:39.250
Now what is lambda?

00:05:39.250 --> 00:05:40.951
Lambda is the matrix
of eigenvalues.

00:05:40.951 --> 00:05:41.450
Right?

00:05:41.450 --> 00:05:44.360
So it's just a and b.

00:05:44.360 --> 00:05:49.330
Those are the diagonal
entries of my lambda matrix.

00:05:49.330 --> 00:05:53.110
And then, we just
find S inverse.

00:05:53.110 --> 00:05:58.040
So we just take
negative signs here

00:05:58.040 --> 00:06:05.340
and recall that we have to
divide by the determinant.

00:06:05.340 --> 00:06:10.840
And the determinant of
this matrix is just -1.

00:06:10.840 --> 00:06:13.080
So we just change
the signs there.

00:06:13.080 --> 00:06:13.680
Good.

00:06:13.680 --> 00:06:18.310
So this is our nice
decomposition of C.

00:06:18.310 --> 00:06:20.740
Now how do we take powers of C?

00:06:20.740 --> 00:06:25.820
Well, C to the k is just S
lambda to the k S inverse.

00:06:28.360 --> 00:06:39.040
[1, 1; 2, 1]; a to the k,
b to the k; [-1, 1; 2, -1].

00:06:39.040 --> 00:06:40.020
Good.

00:06:40.020 --> 00:06:46.820
And multiplying these
matrices together,

00:06:46.820 --> 00:06:49.110
just do a little
arithmetic here.

00:06:54.660 --> 00:06:56.880
Got a bunch of
powers of a and b.

00:06:56.880 --> 00:07:02.040
Because we take powers
of the eigenvalues.

00:07:02.040 --> 00:07:09.050
We have here, we have 2 b
to the k minus a to the k.

00:07:09.050 --> 00:07:12.540
Have a to the k
minus b to the k.

00:07:15.390 --> 00:07:20.820
2 b to the k minus 2 a to the k.

00:07:20.820 --> 00:07:27.800
And finally, we get 2 a
to the k minus b to the k.

00:07:27.800 --> 00:07:30.540
And this is our
k-th power matrix.

00:07:33.213 --> 00:07:33.713
Good.

00:07:36.560 --> 00:07:37.305
A quick check.

00:07:37.305 --> 00:07:39.900
It's always good to
check your work here.

00:07:39.900 --> 00:07:42.880
Let's plug in k equals 1.

00:07:42.880 --> 00:07:43.880
And what do we get?

00:07:43.880 --> 00:07:51.080
We get 2b minus a, a minus b,
2b minus 2a, and 2a minus b.

00:07:51.080 --> 00:07:53.410
And if we can go all the
way back to our matrix

00:07:53.410 --> 00:07:57.160
at the very beginning,
all the way back here,

00:07:57.160 --> 00:08:00.550
that agrees perfectly
with what we started with.

00:08:00.550 --> 00:08:02.250
So that's good.

00:08:02.250 --> 00:08:05.300
That means that we did
this decomposition right.

00:08:05.300 --> 00:08:05.810
Good.

00:08:05.810 --> 00:08:09.160
So now, we've computed the
k-th power of this matrix.

00:08:09.160 --> 00:08:13.750
Let's do a particular example.

00:08:13.750 --> 00:08:18.380
So let's plug in a and b are -1.

00:08:18.380 --> 00:08:22.470
So a equals b equals -1.

00:08:22.470 --> 00:08:25.690
And k equals 100.

00:08:25.690 --> 00:08:27.240
Then what do we get?

00:08:27.240 --> 00:08:31.060
Well, -1 to the 100th is just 1.

00:08:31.060 --> 00:08:34.830
So we're just plugging in 1
for b to the k and a to the k

00:08:34.830 --> 00:08:35.950
everywhere.

00:08:35.950 --> 00:08:42.049
And we just get, in this
case, C to the 100th is just

00:08:42.049 --> 00:08:45.420
[1, 0; 0, 1].

00:08:45.420 --> 00:08:49.160
It's just the identity matrix.

00:08:49.160 --> 00:08:51.720
Great.

00:08:51.720 --> 00:08:52.290
Great.

00:08:52.290 --> 00:08:52.790
OK.

00:08:52.790 --> 00:08:57.800
Now to summarize, how do
we take powers of a matrix?

00:08:57.800 --> 00:08:59.940
Well, first we diagonalize.

00:08:59.940 --> 00:09:02.980
We write our matrix
as S lambda S inverse.

00:09:02.980 --> 00:09:06.358
And then, we just take powers
of the diagonal matrix.