WEBVTT

00:00:05.772 --> 00:00:08.035
PROFESSOR: Welcome
to this recitation.

00:00:08.035 --> 00:00:10.410
In this recitation, we're
going to look at linear systems

00:00:10.410 --> 00:00:11.630
with complex roots.

00:00:11.630 --> 00:00:14.040
So the system we're
examining is the one given

00:00:14.040 --> 00:00:19.640
as x dot equals minus 3x minus
2, and y dot equals 5x minus y.

00:00:19.640 --> 00:00:21.440
And you're asked
to use the matrix

00:00:21.440 --> 00:00:23.570
methods to solve this system.

00:00:23.570 --> 00:00:25.660
So why don't you
take a pause here

00:00:25.660 --> 00:00:27.432
and try to solve this problem?

00:00:27.432 --> 00:00:28.390
And I'll be right back.

00:00:37.460 --> 00:00:39.000
Welcome back.

00:00:39.000 --> 00:00:42.520
So the first step is to write
this system in matrix form.

00:00:45.510 --> 00:00:50.370
So we introduced
a vector [x, y],

00:00:50.370 --> 00:00:55.030
the matrix multiplying
column vector [x, y] again.

00:00:55.030 --> 00:01:02.440
The coefficients are going to
be minus 3, minus 2, 5, minus 1.

00:01:04.959 --> 00:01:07.610
So the first step in
solving this system

00:01:07.610 --> 00:01:18.380
is to find the eigenvalues
of the matrix A.

00:01:18.380 --> 00:01:20.950
So the eigenvalues of
matrix A are basically

00:01:20.950 --> 00:01:26.030
the solutions to this following
determinant equal to 0, minus 3

00:01:26.030 --> 00:01:32.050
minus lambda, minus 2,
5, minus 1 minus lambda,

00:01:32.050 --> 00:01:34.280
determinant equals to 0.

00:01:34.280 --> 00:01:37.500
Here, the lambda are
the unknown eigenvalues.

00:01:37.500 --> 00:01:39.100
And to get this
determinant, we're

00:01:39.100 --> 00:01:48.160
basically multiplying these
two terms, minus minus 2 dot 5,

00:01:48.160 --> 00:01:51.180
which gives us a plus sign.

00:01:51.180 --> 00:01:54.340
So here, we're going
to have lambda squared.

00:01:54.340 --> 00:01:59.450
3*lambda plus 1*lambda
gives us 4*lambda.

00:01:59.450 --> 00:02:05.450
And 3 plus 10 gives us 13.

00:02:10.410 --> 00:02:16.140
So this second-order
polynomial in lambda

00:02:16.140 --> 00:02:18.715
will give us the two
eigenvalues for this matrix.

00:02:18.715 --> 00:02:21.260
So let's examine
the discriminant.

00:02:21.260 --> 00:02:28.690
So we have b squared minus 4a*b.

00:02:28.690 --> 00:02:32.160
And this gives us minus 36.

00:02:32.160 --> 00:02:35.800
So the discriminant is negative.

00:02:35.800 --> 00:02:37.380
And that tells us
that we are going

00:02:37.380 --> 00:02:43.389
to have two complex roots, which
is the title of the recitation.

00:02:43.389 --> 00:02:44.930
And these two complex
roots are going

00:02:44.930 --> 00:02:47.740
to be complex conjugate
of each other.

00:02:47.740 --> 00:02:52.720
So the formula gives
us plus or minus i of 6

00:02:52.720 --> 00:02:54.480
for the root of
the discriminant.

00:02:54.480 --> 00:02:58.100
Here, we have minus 4 over 2.

00:02:58.100 --> 00:03:05.520
So the two roots are basically
minus 2 plus or minus i*3

00:03:05.520 --> 00:03:06.110
or 3i.

00:03:08.920 --> 00:03:10.610
So these are our two roots.

00:03:10.610 --> 00:03:14.190
So now, let's focus
on one of the roots

00:03:14.190 --> 00:03:18.370
to get the eigenvector
associated with the eigenvalue.

00:03:21.490 --> 00:03:24.395
So let's focus on the
positive one, for example.

00:03:27.680 --> 00:03:29.870
And we could do all
the following again--

00:03:29.870 --> 00:03:30.760
AUDIENCE: Minus 2.

00:03:30.760 --> 00:03:32.210
PROFESSOR: Minus 2.

00:03:32.210 --> 00:03:33.110
Thank you.

00:03:33.110 --> 00:03:34.930
We can do all the
following calculation

00:03:34.930 --> 00:03:39.120
that I'm going to go do now
for the complex conjugate,

00:03:39.120 --> 00:03:42.220
and I will explain at the
end how that basically

00:03:42.220 --> 00:03:46.000
not change the result.
So for this eigenvalue,

00:03:46.000 --> 00:03:48.410
we need to compute
now the eigenvector.

00:03:48.410 --> 00:03:54.460
So to do that, we basically have
to use minus 3 minus-- I'm just

00:03:54.460 --> 00:03:57.810
going to write the system here,
let you see what I'm doing.

00:04:02.060 --> 00:04:10.340
And we are solving this system.

00:04:10.340 --> 00:04:12.950
So where does this
system come from?

00:04:12.950 --> 00:04:15.770
It comes from the fact
that we're looking

00:04:15.770 --> 00:04:22.900
for an eigenvector, v_plus, that
is defined as a*v_plus equals

00:04:22.900 --> 00:04:24.370
lambda_plus v_plus.

00:04:24.370 --> 00:04:26.240
And you can then
bring everything

00:04:26.240 --> 00:04:29.170
on the left-hand
side, a minus lambda_i

00:04:29.170 --> 00:04:31.680
applied to v_plus gives
us the zero vector.

00:04:31.680 --> 00:04:33.070
So that's what we have here.

00:04:33.070 --> 00:04:34.910
The unknowns are
a_1, a_2, and we're

00:04:34.910 --> 00:04:36.310
going to try to solve for this.

00:04:36.310 --> 00:04:38.520
So if we plug in now for
the value of lambda_plus

00:04:38.520 --> 00:04:43.850
that we have, we have minus 3
plus 2, which gives us minus 1.

00:04:43.850 --> 00:04:49.640
And then, we have a
minus 3i and minus 2.

00:04:49.640 --> 00:04:52.760
And for the second line,
second entry of this matrix

00:04:52.760 --> 00:04:56.530
you have 5, minus
1 minus minus 2.

00:04:56.530 --> 00:04:59.600
So we have 2 minus
1, which is 1.

00:04:59.600 --> 00:05:00.890
And then, we have minus 3i.

00:05:04.770 --> 00:05:09.200
[a 1, a 2] equals to [0, 0].

00:05:09.200 --> 00:05:11.210
So here, you can
check for yourself

00:05:11.210 --> 00:05:13.370
that these two equations
given by the first line

00:05:13.370 --> 00:05:15.360
and the second line
are actually the same.

00:05:15.360 --> 00:05:19.620
And so basically,
to get a_1 and a_2,

00:05:19.620 --> 00:05:25.930
it is sufficient to just solve,
for example, the first one,

00:05:25.930 --> 00:05:28.760
where here, I just wrote
minus 1 minus 3i multiplied

00:05:28.760 --> 00:05:32.080
by a_1 minus 2a_2 equals to 0.

00:05:32.080 --> 00:05:34.110
And I just brought the
minus 2 on this side.

00:05:34.110 --> 00:05:36.920
So here, you can see that
if we pick a_1 equals to 0--

00:05:36.920 --> 00:05:41.560
equals to plus 2, which
would be our first entry,

00:05:41.560 --> 00:05:43.660
we can then cancel out
these two and just have

00:05:43.660 --> 00:05:47.860
a_2 equals to minus 1 minus 3i.

00:05:47.860 --> 00:05:51.710
So this would be one
eigenvector associated

00:05:51.710 --> 00:05:53.210
with this eigenvalue.

00:05:53.210 --> 00:05:54.850
We could have picked other ones.

00:05:54.850 --> 00:05:56.810
They're basically
parallel to this one.

00:06:00.290 --> 00:06:02.240
So now what?

00:06:02.240 --> 00:06:06.100
So what we need to remember
is the meaning of all of this.

00:06:06.100 --> 00:06:08.790
Seeking the eigenvalues and
the eigenvectors is basically

00:06:08.790 --> 00:06:14.860
equivalent to seeking a
solution in the form exponential

00:06:14.860 --> 00:06:18.850
lambda*t with the direction of
the eigenvector associated with

00:06:18.850 --> 00:06:19.940
this eigenvalue.

00:06:19.940 --> 00:06:22.120
So now that we actually
have this eigenvector

00:06:22.120 --> 00:06:26.160
and this eigenvalue, we can
write down the solution.

00:06:26.160 --> 00:06:29.770
And I'm just going to write
the solution in x, which

00:06:29.770 --> 00:06:32.870
has entries basically x and y.

00:06:32.870 --> 00:06:34.670
And one way of writing
it would be just

00:06:34.670 --> 00:06:37.870
to basically first start by
writing what we have there.

00:06:42.940 --> 00:06:44.520
I'm just going to spell it out.

00:06:44.520 --> 00:06:53.230
So we have this
multiplied by 2 minus 3i.

00:06:53.230 --> 00:06:54.550
So what do we do with this?

00:06:54.550 --> 00:06:56.520
Well, we remember
our earlier formula.

00:06:56.520 --> 00:06:59.860
So this is exponential minus
2t plus exponential 3i*t.

00:06:59.860 --> 00:07:03.190
So we can split the exponential
3i*t into a cosine and a sine.

00:07:03.190 --> 00:07:05.610
And this, we're going to
also be able to split it

00:07:05.610 --> 00:07:08.220
into the complex part
and the real part.

00:07:08.220 --> 00:07:11.940
And then, we're going
to combine the real part

00:07:11.940 --> 00:07:13.220
and the complex part.

00:07:13.220 --> 00:07:14.640
So let's do that.

00:07:14.640 --> 00:07:19.820
Exponential minus
2t multiplying,

00:07:19.820 --> 00:07:34.620
basically, cosine 3t plus i
sine 3t for the entry 2 minus 1

00:07:34.620 --> 00:07:35.440
minus 3i.

00:07:35.440 --> 00:07:37.370
So we have an i
here and an i here.

00:07:37.370 --> 00:07:47.490
So things can be combined
into a real part.

00:07:47.490 --> 00:07:50.530
So in the first entry here,
what are we going to have?

00:07:50.530 --> 00:07:52.460
We're going to have
cos 3t multiplying 2.

00:07:52.460 --> 00:07:53.960
That's going to be
in the real part.

00:07:57.396 --> 00:07:58.770
And I'm going to
keep some space.

00:08:05.260 --> 00:08:10.660
And another entry here at the
second entry of this vector

00:08:10.660 --> 00:08:16.410
is going to give us cosine
3t multiplied by minus 1.

00:08:16.410 --> 00:08:17.836
Oops, here, it should be a 3t.

00:08:17.836 --> 00:08:18.336
Sorry.

00:08:22.710 --> 00:08:26.840
So minus cosine 3t.

00:08:26.840 --> 00:08:30.420
Now, where are we going to
have another real part here?

00:08:30.420 --> 00:08:33.440
It's going to come from a
multiplication of i sine 3t

00:08:33.440 --> 00:08:34.580
by 3i.

00:08:34.580 --> 00:08:36.620
So the two i's together
gives a minus 1.

00:08:36.620 --> 00:08:44.030
And we end up with
a plus 3 sine 3t.

00:08:44.030 --> 00:08:45.827
So we're done for the real part.

00:08:45.827 --> 00:08:47.410
Now let's focus on
the imaginary part.

00:08:47.410 --> 00:08:48.076
What do we have?

00:08:48.076 --> 00:08:50.540
We have an i sine
3t multiplying a 2.

00:08:56.280 --> 00:09:01.810
And we have a minus 3i
here multiplying cosine 3t.

00:09:01.810 --> 00:09:08.310
So we want to have
a minus 3 cosine 3t,

00:09:08.310 --> 00:09:11.020
and finally, this minus 1
multiplying this sine 3t.

00:09:16.200 --> 00:09:17.320
So now, we did--

00:09:17.320 --> 00:09:18.849
AUDIENCE: [INAUDIBLE].

00:09:18.849 --> 00:09:19.640
PROFESSOR: Oh yeah.

00:09:19.640 --> 00:09:20.730
Thank you.

00:09:20.730 --> 00:09:23.470
2, from this operation.

00:09:23.470 --> 00:09:25.700
So now, we did
split our solution

00:09:25.700 --> 00:09:28.930
into a real part and
an imaginary part.

00:09:28.930 --> 00:09:32.400
So how can we write the
general solution of the system?

00:09:32.400 --> 00:09:35.880
Well, we knew that for this
linear system of equations,

00:09:35.880 --> 00:09:38.960
if we have a complex
number that is

00:09:38.960 --> 00:09:41.370
a solution to the
linear equation,

00:09:41.370 --> 00:09:44.700
then its real part
and its imaginary part

00:09:44.700 --> 00:09:47.180
are also two
independent solutions.

00:09:47.180 --> 00:09:51.860
So we can write the general
solution of the system

00:09:51.860 --> 00:09:54.610
as a linear combination
of the real part

00:09:54.610 --> 00:09:55.860
and the imaginary part.

00:09:55.860 --> 00:10:09.320
And I can just label this u_1 of
t and u_2 of t here and vector.

00:10:09.320 --> 00:10:11.960
And we can then write
the general solution

00:10:11.960 --> 00:10:14.790
in terms of any constant--
that would be determined

00:10:14.790 --> 00:10:18.760
by the initial condition if
we had one-- exponential minus

00:10:18.760 --> 00:10:28.110
2t along u_1 plus c_2
exponential minus 2t

00:10:28.110 --> 00:10:34.895
along vector u_2, which
are also functions of t,

00:10:34.895 --> 00:10:38.920
just the difference
from what we had before.

00:10:38.920 --> 00:10:43.080
So here, basically we seek
for the eigenvalues values

00:10:43.080 --> 00:10:44.590
of the matrix.

00:10:44.590 --> 00:10:47.590
We looked for the
eigenvector associated

00:10:47.590 --> 00:10:49.650
with the complex eigenvalue.

00:10:49.650 --> 00:10:52.470
We were able to write
the full solution.

00:10:52.470 --> 00:10:55.900
And then, because of
the linearity property,

00:10:55.900 --> 00:10:58.250
we were able to
just then extract

00:10:58.250 --> 00:11:00.810
two linearly independent
solutions, the real part

00:11:00.810 --> 00:11:03.690
of the solution we had and the
imaginary part of the solution

00:11:03.690 --> 00:11:04.360
we had.

00:11:04.360 --> 00:11:07.640
So what I mentioned
earlier was that we

00:11:07.640 --> 00:11:14.130
could do this whole calculation
for the other eigenvalue

00:11:14.130 --> 00:11:16.970
with a minus.

00:11:16.970 --> 00:11:19.369
If you try to do it and
trickle down your minus,

00:11:19.369 --> 00:11:21.160
you would see that
basically you would just

00:11:21.160 --> 00:11:24.890
end up with minus
signs here basically

00:11:24.890 --> 00:11:26.410
in front of the sines.

00:11:26.410 --> 00:11:29.570
And what you could do
then is just simply

00:11:29.570 --> 00:11:32.090
absorb that minus sign
for the general solution

00:11:32.090 --> 00:11:33.310
in c_1 and c_2.

00:11:33.310 --> 00:11:35.560
And basically, it gives
you exactly the same form

00:11:35.560 --> 00:11:36.620
for the general solution.

00:11:36.620 --> 00:11:39.380
So you don't need to redo
it for the second one.

00:11:39.380 --> 00:11:43.090
You would still end up with
only two linearly independent

00:11:43.090 --> 00:11:44.400
solutions, not four.

00:11:44.400 --> 00:11:47.020
OK, so that ends
this recitation.