WEBVTT

00:00:06.650 --> 00:00:07.900
DAVID SHIROKOFF: Hi, everyone.

00:00:07.900 --> 00:00:10.920
So today I'd like to take
a look at linear equations,

00:00:10.920 --> 00:00:13.690
linear first-order
differential equations.

00:00:13.690 --> 00:00:17.280
And specifically, we're going
to look at just exactly, what

00:00:17.280 --> 00:00:18.440
is a linear equation?

00:00:18.440 --> 00:00:21.302
So for example, which of
these equations are linear?

00:00:21.302 --> 00:00:22.760
And then second of
all, we're going

00:00:22.760 --> 00:00:25.090
to take a look at superposition.

00:00:25.090 --> 00:00:28.300
And we ask to consider
the differential equation

00:00:28.300 --> 00:00:31.550
y dot plus y squared equals
some function q of t.

00:00:31.550 --> 00:00:34.640
And just to demonstrate that
this equation does not satisfy

00:00:34.640 --> 00:00:36.337
the superposition principle.

00:00:36.337 --> 00:00:38.170
So I'll let you take a
look at this problem,

00:00:38.170 --> 00:00:39.640
and try and work it
out for yourself.

00:00:39.640 --> 00:00:40.848
And I'll be back in a minute.

00:00:54.444 --> 00:00:55.610
Hi, everyone.

00:00:55.610 --> 00:00:57.160
Welcome back.

00:00:57.160 --> 00:00:58.350
All right.

00:00:58.350 --> 00:01:00.770
So we're asked to figure
out which of these equations

00:01:00.770 --> 00:01:02.010
are linear.

00:01:02.010 --> 00:01:07.860
So just to recap,
what does it mean

00:01:07.860 --> 00:01:09.120
for an equation to be linear?

00:01:09.120 --> 00:01:10.953
What does it mean for
a first-order equation

00:01:10.953 --> 00:01:11.910
to be linear?

00:01:11.910 --> 00:01:17.740
Well, it means that we can
write it in the general form,

00:01:17.740 --> 00:01:23.230
so if it's a
function of x, dy/dx

00:01:23.230 --> 00:01:25.400
plus some general function of x.

00:01:25.400 --> 00:01:27.690
And this could be anything.

00:01:27.690 --> 00:01:29.862
But the point is that
it's multiplied just by y.

00:01:32.570 --> 00:01:34.720
And then it can equal
some arbitrary function

00:01:34.720 --> 00:01:37.230
q of x on the right-hand side.

00:01:37.230 --> 00:01:39.170
So every first-order
linear equation

00:01:39.170 --> 00:01:40.420
can be rewritten in this form.

00:01:43.640 --> 00:01:49.130
So let's tackle part A. All
right, y dot plus k*y equals--

00:01:49.130 --> 00:01:51.470
or y dot equals k*y.

00:01:51.470 --> 00:01:57.410
We can just rewrite it as
y dot equals negative k*y.

00:01:57.410 --> 00:02:01.885
Note that just some constant
k is a function of time.

00:02:01.885 --> 00:02:03.260
It's a constant
function of time,

00:02:03.260 --> 00:02:04.870
but it can just be thought
of as a function of time.

00:02:04.870 --> 00:02:06.790
Same with the
right-hand side, it's 0.

00:02:06.790 --> 00:02:09.130
And so this equation
has the same form

00:02:09.130 --> 00:02:13.320
as our general first-order
linear equation.

00:02:13.320 --> 00:02:17.100
So this is a linear equation.

00:02:17.100 --> 00:02:20.540
Let's take a look at
question 2, part 2.

00:02:20.540 --> 00:02:24.810
We have y dot plus y squared t.

00:02:24.810 --> 00:02:28.770
I'll bring the y to
the left-hand side.

00:02:28.770 --> 00:02:30.990
If we do that, we get this.

00:02:30.990 --> 00:02:34.730
And just by comparing this to
our generic first-order linear

00:02:34.730 --> 00:02:37.750
ODE, we note that
it's not of this form.

00:02:37.750 --> 00:02:42.570
So specifically, this is
the bad term in question.

00:02:42.570 --> 00:02:45.960
We see that it's
y squared times t.

00:02:45.960 --> 00:02:48.620
Notice how we're not allowed
to have any term that's y

00:02:48.620 --> 00:02:52.730
squared in our first-order
linear equation.

00:02:52.730 --> 00:02:54.345
Now I will say this
just as an aside.

00:02:59.000 --> 00:03:01.810
This equation's also known
as a Bernoulli equation.

00:03:01.810 --> 00:03:06.650
And as an aside, if you were
to make a substitution u equals

00:03:06.650 --> 00:03:11.670
1 over y, and rewrote this
equation in terms of u

00:03:11.670 --> 00:03:13.630
as the independent
variable, you'd

00:03:13.630 --> 00:03:17.855
find out that this equation
would be linear in terms of u.

00:03:17.855 --> 00:03:21.450
But as it's written now in
terms of y, it's not linear.

00:03:21.450 --> 00:03:23.200
So this is a kind of
an interesting point.

00:03:23.200 --> 00:03:25.980
That sometimes we can get
lucky and do a nonlinear

00:03:25.980 --> 00:03:28.190
transformation and
convert a nonlinear

00:03:28.190 --> 00:03:29.720
equation into a linear one.

00:03:36.020 --> 00:03:49.560
So for part 3 we have y prime
plus cos(x)*y equals x cubed.

00:03:49.560 --> 00:03:51.810
So again, we see that
this is a linear equation

00:03:51.810 --> 00:03:59.290
because we can identify
p of x with cosine x, q

00:03:59.290 --> 00:04:02.840
of x with x cubed.

00:04:02.840 --> 00:04:06.455
So this equation has the
general form of a linear ODE.

00:04:10.260 --> 00:04:13.730
For part 4, well if we take
a look at it right away,

00:04:13.730 --> 00:04:17.449
y dot divided by y is
equal to t squared.

00:04:17.449 --> 00:04:18.740
It doesn't appear to be linear.

00:04:18.740 --> 00:04:21.820
But of course, we can
multiply through by y

00:04:21.820 --> 00:04:22.820
and rewrite things.

00:04:25.430 --> 00:04:27.760
And of course, this
equation is equivalent to y

00:04:27.760 --> 00:04:32.160
dot minus t squared y equals 0.

00:04:32.160 --> 00:04:38.670
Which again, is
linear because p of t

00:04:38.670 --> 00:04:41.630
can be identified with
negative t squared and q of t

00:04:41.630 --> 00:04:45.120
can be identified with 0.

00:04:45.120 --> 00:04:54.010
And then lastly,
for part 5, we have

00:04:54.010 --> 00:05:04.780
x squared y y prime
plus 4x equals x cubed.

00:05:07.320 --> 00:05:11.220
And again, if we were to try
to write it in the general form

00:05:11.220 --> 00:05:18.650
of a linear ODE, which we have
here, it would look like...

00:05:23.460 --> 00:05:30.400
4x divided by x squared y minus
x cubed divided by x squared

00:05:30.400 --> 00:05:34.500
y equals 0.

00:05:34.500 --> 00:05:37.399
And because there's a y
on the denominator here,

00:05:37.399 --> 00:05:39.190
this equation doesn't
have the general form

00:05:39.190 --> 00:05:42.030
of a first-order
linear equation.

00:05:42.030 --> 00:05:46.680
Now again, I note that if you
were to make a substitution,

00:05:46.680 --> 00:05:52.247
u equal to y squared,
that substitution

00:05:52.247 --> 00:05:53.580
would make this equation linear.

00:05:56.870 --> 00:05:58.120
So this concludes part a.

00:06:00.980 --> 00:06:03.460
So for part b, we're given
a differential equation

00:06:03.460 --> 00:06:11.720
y dot plus y squared
is equal to q of t.

00:06:11.720 --> 00:06:14.960
And we want to show that
this equation doesn't satisfy

00:06:14.960 --> 00:06:17.360
the superposition principle.

00:06:17.360 --> 00:06:22.030
Now, you might note that
this equation is not linear.

00:06:22.030 --> 00:06:25.150
And second of all, we know
that linear equations satisfy

00:06:25.150 --> 00:06:26.370
the superposition principle.

00:06:29.210 --> 00:06:33.840
So we need to exploit the fact
that it's a nonlinear equation

00:06:33.840 --> 00:06:37.410
to make it fail the
superposition principle.

00:06:37.410 --> 00:06:39.890
And just as an example,
what we can do is I

00:06:39.890 --> 00:06:42.030
can just pick a couple
right-hand sides

00:06:42.030 --> 00:06:47.980
q as examples of this ODE.

00:06:47.980 --> 00:06:51.260
So we can imagine we
have a solution y_1 which

00:06:51.260 --> 00:06:56.450
solves the ODE y_1
dot plus y_1 squared

00:06:56.450 --> 00:07:00.610
equals-- I'm just going to
say 1 on the right-hand side.

00:07:00.610 --> 00:07:03.320
And I can also take
another function

00:07:03.320 --> 00:07:09.500
y_2, which say satisfies this
differential equation, say t.

00:07:09.500 --> 00:07:11.200
So all I've done
is I've just picked

00:07:11.200 --> 00:07:14.860
one q_1 on the right-hand
side and another function q_2.

00:07:14.860 --> 00:07:18.180
And I imagined that I
have a solution y_1 which

00:07:18.180 --> 00:07:21.440
satisfies this differential
equation and y_2 which

00:07:21.440 --> 00:07:24.670
satisfies this
differential equation.

00:07:24.670 --> 00:07:28.680
So what does the
superposition principle say?

00:07:28.680 --> 00:07:38.510
Well, superposition
says that if y_1

00:07:38.510 --> 00:07:42.350
solves this equation and
y_2 solves this equation,

00:07:42.350 --> 00:07:51.970
then the function y equals
y_1 plus y_2 must solve

00:07:51.970 --> 00:07:59.900
the equation y dot
plus y squared equals

00:07:59.900 --> 00:08:04.400
the sum of the right-hand
side, 1 plus t.

00:08:04.400 --> 00:08:08.670
So if we try and
substitute in y_1 plus y_2

00:08:08.670 --> 00:08:12.520
into this equation, we're going
to come to a contradiction.

00:08:12.520 --> 00:08:14.160
So let's do that now.

00:08:14.160 --> 00:08:20.220
So we take d by dt
of y_1 plus y_2,

00:08:20.220 --> 00:08:25.200
and I'm going to add it to
y_1 plus y_2 quantity squared.

00:08:29.490 --> 00:08:31.620
And now what I can do is
I can just simplify this.

00:08:31.620 --> 00:08:36.630
So I get y_1 dot plus y_2 dot.

00:08:36.630 --> 00:08:39.350
And I can expand
out this square.

00:08:39.350 --> 00:08:48.542
So I get y_1 squared plus
y_2 squared plus 2y_1*y_2.

00:08:48.542 --> 00:08:50.000
And now what I'm
going to do is I'm

00:08:50.000 --> 00:08:56.550
going to combine the term
y_1 with y_1 squared.

00:08:56.550 --> 00:09:01.490
And by construction, y_1 dot
plus y_1 squared is equal to 1,

00:09:01.490 --> 00:09:03.880
because I assumed that it
satisfied this differential

00:09:03.880 --> 00:09:06.020
equation.

00:09:06.020 --> 00:09:11.940
And then in addition by
construction, y_2 dot plus y_2

00:09:11.940 --> 00:09:16.150
squared satisfies
the right-hand side

00:09:16.150 --> 00:09:18.340
of the second equation with t.

00:09:21.160 --> 00:09:29.540
So I get 1 plus t, but I'm
left over with this other piece

00:09:29.540 --> 00:09:30.040
2y_1*y_2.

00:09:33.120 --> 00:09:36.160
So in general, if I
take a function which

00:09:36.160 --> 00:09:38.580
solves this differential
equation and another function

00:09:38.580 --> 00:09:40.940
which solves this
differential equation,

00:09:40.940 --> 00:09:43.270
and I add them
together, and plug it

00:09:43.270 --> 00:09:46.390
into the left-hand side of
this differential equation,

00:09:46.390 --> 00:09:51.750
I get something which is 1
plus t plus some other stuff.

00:09:51.750 --> 00:09:58.510
And if it did satisfy the
principle of superposition,

00:09:58.510 --> 00:10:01.710
it must equal 1 plus t.

00:10:01.710 --> 00:10:03.460
So we arrive at a
contradiction because it

00:10:03.460 --> 00:10:05.830
doesn't equal 1 plus t.

00:10:05.830 --> 00:10:11.460
In fact, it fails to equal 1
plus t by this term 2y_1*y_2.

00:10:11.460 --> 00:10:14.140
And this term comes directly
from the fact that we had

00:10:14.140 --> 00:10:17.460
a nonlinearity in the equation.

00:10:17.460 --> 00:10:20.290
So this is just one
illustration of the fact

00:10:20.290 --> 00:10:23.080
that a nonlinear equation
doesn't necessarily satisfy

00:10:23.080 --> 00:10:25.270
the superposition principle.

00:10:25.270 --> 00:10:28.150
Whereas, every linear equation
satisfies the superposition

00:10:28.150 --> 00:10:29.130
principle.

00:10:29.130 --> 00:10:31.730
This is one reason we
love linear equations

00:10:31.730 --> 00:10:34.760
and we study them extensively.

00:10:34.760 --> 00:10:38.320
OK, so I'd just like
to conclude here.

00:10:38.320 --> 00:10:41.570
And I'll see you next time.