WEBVTT

00:00:05.489 --> 00:00:06.960
PROFESSOR: Welcome back.

00:00:06.960 --> 00:00:09.220
So in this session, we're
going to look at unit step

00:00:09.220 --> 00:00:10.910
and impulse responses.

00:00:10.910 --> 00:00:13.800
So in this question, we ask
you to find the unit impulse

00:00:13.800 --> 00:00:17.680
response to these two equations,
x dot plus 2x equals f of t,

00:00:17.680 --> 00:00:24.660
and 2 x dot dot plus 27 x
dot-- oops, it should be a 7--

00:00:24.660 --> 00:00:27.840
plus 7 x dot plus
3x equals f of t.

00:00:27.840 --> 00:00:31.660
In the second part, you're asked
to find the unit step response

00:00:31.660 --> 00:00:33.450
for the first equation.

00:00:33.450 --> 00:00:35.230
So here, the key
points are really

00:00:35.230 --> 00:00:38.140
to remember what do we mean
by unit impulse and unit step

00:00:38.140 --> 00:00:38.890
response.

00:00:38.890 --> 00:00:41.740
Which initial condition
correspond to these responses?

00:00:41.740 --> 00:00:44.906
And what functions of f of t
do you choose in each case?

00:00:44.906 --> 00:00:47.530
So why don't you pause the video
and work through this problem?

00:00:47.530 --> 00:00:48.488
And I'll be right back.

00:00:56.230 --> 00:00:57.460
Welcome back.

00:00:57.460 --> 00:01:05.190
So let's look at the equation
A. So the unit impulse response

00:01:05.190 --> 00:01:09.540
is simply-- I'm going
to write this down,

00:01:09.540 --> 00:01:15.210
unit impulse response--
is simply the solution

00:01:15.210 --> 00:01:21.470
to the following problem, to
our differential equation,

00:01:21.470 --> 00:01:24.650
x dot plus 2x that we're
given, with the forcing

00:01:24.650 --> 00:01:32.740
in a delta function of magnitude
1 with rest initial conditions,

00:01:32.740 --> 00:01:35.330
which means in this case,
that x of 0 is equal to 0.

00:01:35.330 --> 00:01:37.990
x of 0 minus is equal to 0.

00:01:37.990 --> 00:01:40.690
So how do we go about
solving this equation?

00:01:40.690 --> 00:01:43.790
So here, first of all, let's
look at what this means.

00:01:43.790 --> 00:01:45.720
This equation could be
modeling, for example,

00:01:45.720 --> 00:01:48.670
the quantity of radioactive
chemical in a tank.

00:01:48.670 --> 00:01:53.900
And delta(t) here means that we
are giving a high disturbance

00:01:53.900 --> 00:01:55.650
on a very short
time on this system.

00:01:55.650 --> 00:01:57.590
So for example,
dumping a huge amount

00:01:57.590 --> 00:02:00.600
of chemical in that tank, and
then, just letting the system

00:02:00.600 --> 00:02:03.850
evolve after dumping all
that chemical in a tank.

00:02:03.850 --> 00:02:06.240
So really, what we could
see is that the delta

00:02:06.240 --> 00:02:10.539
can be identified to x
dot, which is the highest

00:02:10.539 --> 00:02:11.830
derivative in the system.

00:02:11.830 --> 00:02:13.430
And that would then
mean that we have

00:02:13.430 --> 00:02:15.450
a jump in the second
highest derivative, which

00:02:15.450 --> 00:02:16.630
would be in x.

00:02:16.630 --> 00:02:28.220
So delta t introduces
a jump in x.

00:02:28.220 --> 00:02:32.470
And what do I mean by
that is that we had rest

00:02:32.470 --> 00:02:34.710
initial conditions initially.

00:02:34.710 --> 00:02:41.140
Now, we have to go into a jump
where the quantity in the tank

00:02:41.140 --> 00:02:44.200
was raised to 1,
which is basically

00:02:44.200 --> 00:02:46.280
the magnitude on the
right-hand side and the fact

00:02:46.280 --> 00:02:48.113
that we have no
coefficients for the highest

00:02:48.113 --> 00:02:50.030
derivative in the system.

00:02:50.030 --> 00:02:52.210
So basically, we can
solve this equation

00:02:52.210 --> 00:03:04.500
by solving the equivalent
problem x dot plus 2x equals

00:03:04.500 --> 00:03:07.240
to 0, which is just
the homogeneous part,

00:03:07.240 --> 00:03:14.850
with our new initial conditions,
x of 0 minus equals to 1.

00:03:14.850 --> 00:03:16.890
And so here,
clearly from before,

00:03:16.890 --> 00:03:22.300
we saw that the solution would
be a decaying exponential.

00:03:22.300 --> 00:03:25.380
We would have a constant
of integration here, which

00:03:25.380 --> 00:03:28.180
would be just 1 given our
new initial condition.

00:03:28.180 --> 00:03:30.990
So this would be valid
for t larger than 0.

00:03:30.990 --> 00:03:34.180
And given that we started from
rest initial conditions prior

00:03:34.180 --> 00:03:36.430
to this perturbance
introduced by the delta,

00:03:36.430 --> 00:03:40.720
we could also write this
with a u of t step function

00:03:40.720 --> 00:03:44.860
if we were to consider
t in R. So if you

00:03:44.860 --> 00:03:49.210
want to look at how the solution
looks like, we would have

00:03:49.210 --> 00:03:52.510
basically rest
initial conditions,

00:03:52.510 --> 00:03:55.950
and then, a jump
introduced by our forcing,

00:03:55.950 --> 00:04:01.030
and after that, just a
decaying exponential,

00:04:01.030 --> 00:04:05.630
and a jump here to a new
initial condition of 1.

00:04:05.630 --> 00:04:07.130
So let's move to
the second problem.

00:04:09.800 --> 00:04:12.170
Now, we're dealing with a
second-order differential

00:04:12.170 --> 00:04:14.410
equation.

00:04:14.410 --> 00:04:23.880
2 x dot dot plus 7 x dot
plus 3x equals f of t.

00:04:23.880 --> 00:04:27.810
And we are asked again
to seek the unit impulse

00:04:27.810 --> 00:04:30.630
response to this problem.

00:04:30.630 --> 00:04:39.275
So the unit impulse response
is simply the solution of 2

00:04:39.275 --> 00:04:46.420
x dot dot plus 7 x dot
plus 3x equals-- Again,

00:04:46.420 --> 00:04:48.610
the unit impulse
means that we're just

00:04:48.610 --> 00:04:52.910
kicking the system with an
impulse or delta function.

00:04:52.910 --> 00:04:57.780
And we're starting from
rest initial conditions.

00:04:57.780 --> 00:05:01.690
So here, let's just look at
what could this be modeling.

00:05:01.690 --> 00:05:04.310
So this could be just
basically Newton's Second Law

00:05:04.310 --> 00:05:08.360
with acceleration and forces
applied to the system.

00:05:08.360 --> 00:05:11.040
For example, it could
be a mass hanged

00:05:11.040 --> 00:05:13.970
on a spring with certain
damping due to this term.

00:05:13.970 --> 00:05:17.080
And this delta here would
have basically the units

00:05:17.080 --> 00:05:20.294
corresponding to a force that
would be felt by the highest

00:05:20.294 --> 00:05:22.460
order of term in the integral
which could correspond

00:05:22.460 --> 00:05:26.760
to the acceleration multiplied,
for example, by a mass of 2.

00:05:26.760 --> 00:05:29.790
So really, the delta
can be identified

00:05:29.790 --> 00:05:31.820
to the highest-order
derivative, which

00:05:31.820 --> 00:05:34.860
means that we would have
a discontinuity from 0

00:05:34.860 --> 00:05:39.280
to 1 on the derivative one
degree lower, so in x dot.

00:05:39.280 --> 00:05:50.910
And so the delta introduces
the discontinuity

00:05:50.910 --> 00:05:53.460
on the second
highest derivative.

00:05:53.460 --> 00:05:55.290
And what I mean
by that is that we

00:05:55.290 --> 00:06:01.900
have a jump from our
original initial conditions

00:06:01.900 --> 00:06:08.560
of x dot 0 minus equals to
0 to new initial conditions,

00:06:08.560 --> 00:06:11.420
x dot of 0 minus
equals one half.

00:06:11.420 --> 00:06:13.190
Where did this one
half come from?

00:06:13.190 --> 00:06:16.012
It came from the fact
that I have a factor of 2

00:06:16.012 --> 00:06:17.720
in front of the
highest-order derivative.

00:06:17.720 --> 00:06:20.500
So if I'm identifying
this with delta,

00:06:20.500 --> 00:06:22.920
then integration of
this term would give me

00:06:22.920 --> 00:06:27.700
an x dot that corresponds to
the jump from 0 to 1 over 2,

00:06:27.700 --> 00:06:29.276
so one half.

00:06:29.276 --> 00:06:30.400
That's where it comes from.

00:06:30.400 --> 00:06:32.733
And each time that you would
have a coefficient in front

00:06:32.733 --> 00:06:34.670
of your highest order
derivative associated

00:06:34.670 --> 00:06:37.480
with the delta on the right,
you would have a jump of 1

00:06:37.480 --> 00:06:40.270
over that coefficient.

00:06:40.270 --> 00:06:42.920
So from this point,
what are we solving?

00:06:42.920 --> 00:06:45.900
What do we need to solve to
get the unit impulse response?

00:06:45.900 --> 00:06:54.500
It's equivalent to solving our
new system where we can then

00:06:54.500 --> 00:07:04.400
get rid of the delta with
our new initial conditions,

00:07:04.400 --> 00:07:05.610
dot equals one half.

00:07:05.610 --> 00:07:07.440
And of course, here,
I did mentioned

00:07:07.440 --> 00:07:11.600
by the initial condition on
x 0 minus 1, because we need

00:07:11.600 --> 00:07:14.330
two initial conditions for
the second order derivative,

00:07:14.330 --> 00:07:17.480
would still be 0, as the
discontinuity would not be

00:07:17.480 --> 00:07:21.050
felt by the function x itself.

00:07:21.050 --> 00:07:23.550
So we can go ahead and
solve this problem.

00:07:23.550 --> 00:07:26.670
So we use the characteristic
polynomial as we did before.

00:07:31.440 --> 00:07:34.380
This characteristic polynomial
would have a discriminant

00:07:34.380 --> 00:07:40.520
of 25, which gives us simple
roots that we can compute,

00:07:40.520 --> 00:07:43.860
so minus 7 plus or
minus the square root

00:07:43.860 --> 00:07:47.910
of the discriminant over 4.

00:07:53.250 --> 00:07:57.580
So we have two roots, minus
7 minus 5 over 4, minus 12

00:07:57.580 --> 00:08:01.160
over 4, which is minus
3; minus 7 plus 5 over 4,

00:08:01.160 --> 00:08:04.630
which just gives
us minus one half.

00:08:04.630 --> 00:08:09.870
So we can write down the
solution to this problem

00:08:09.870 --> 00:08:14.660
as c_1 exponential the
root that we just found,

00:08:14.660 --> 00:08:19.390
minus 3t, plus c_2
exponential the other root,

00:08:19.390 --> 00:08:23.990
minus one half of t.

00:08:23.990 --> 00:08:27.200
Now, to get c_1, c_2, we
need to take into account

00:08:27.200 --> 00:08:28.520
the initial conditions.

00:08:28.520 --> 00:08:31.550
So the first one tells
us that x at 0 is 0.

00:08:31.550 --> 00:08:34.420
So that will give us c_1
plus c_2 equals to 0,

00:08:34.420 --> 00:08:38.200
so basically, c_2
equals minus c_1.

00:08:38.200 --> 00:08:40.510
And the second initial
condition tells us

00:08:40.510 --> 00:08:41.780
that the derivative is 0.

00:08:41.780 --> 00:08:48.320
So that gives us minus 3c_1
minus one half c_2 equals to 0.

00:08:48.320 --> 00:08:50.790
Here, we have that c_2
is equal to minus c_1.

00:08:50.790 --> 00:08:52.595
So we can just factor
out everything.

00:08:52.595 --> 00:08:57.850
And we end up with 3 minus
one half-- whoops, sorry.

00:08:57.850 --> 00:09:05.290
Here it should be one half, our
new modified initial condition,

00:09:05.290 --> 00:09:07.120
equals to one half.

00:09:07.120 --> 00:09:11.080
And here, you would
have 5/2 equal to 1/2.

00:09:11.080 --> 00:09:17.780
So c_2 equals 1/5
equals minus c_1.

00:09:17.780 --> 00:09:21.300
And so plugging in c_1
and c_2 in this formula

00:09:21.300 --> 00:09:26.930
would give us the
general solution

00:09:26.930 --> 00:09:33.280
with c_2 exponential minus one
half t minus exponential minus

00:09:33.280 --> 00:09:34.090
3t.

00:09:34.090 --> 00:09:36.980
And all of this, remember, we're
solving for t larger than 0

00:09:36.980 --> 00:09:40.320
in our modified system with
our new initial conditions.

00:09:40.320 --> 00:09:42.190
So this is for t larger than 0.

00:09:42.190 --> 00:09:46.230
And again, here, if we wanted
to just write it for t in R,

00:09:46.230 --> 00:09:49.850
then we could just add
the step function u of t

00:09:49.850 --> 00:09:52.460
that would just signify that
we took rest initial conditions

00:09:52.460 --> 00:09:56.640
before x equals 0 minus.

00:09:56.640 --> 00:09:58.460
So that ends the first part.

00:09:58.460 --> 00:10:00.450
So now quickly, for
the second part,

00:10:00.450 --> 00:10:02.030
we're asked to
find the unit step

00:10:02.030 --> 00:10:09.270
response to the first system.

00:10:09.270 --> 00:10:11.030
So the unit step
response is just

00:10:11.030 --> 00:10:20.900
the solution to our original
ODE problem, x dot plus 2x.

00:10:20.900 --> 00:10:25.580
But now, f of t is the step
function, hence, the step

00:10:25.580 --> 00:10:31.900
here, with still rest
initial conditions.

00:10:31.900 --> 00:10:33.900
So really, the step
function is just

00:10:33.900 --> 00:10:36.140
0 everywhere before 0 minus.

00:10:36.140 --> 00:10:39.320
And it takes the
value of 1 after.

00:10:39.320 --> 00:10:41.260
And so basically, we
can just solve again

00:10:41.260 --> 00:10:50.090
for t positive, the modified
differential equation with u

00:10:50.090 --> 00:10:51.940
taking just the value of 1.

00:10:51.940 --> 00:10:57.680
And here, we just
get similar roots.

00:10:57.680 --> 00:10:59.039
So it would be a minus 2t.

00:10:59.039 --> 00:11:01.080
But then, we would have
a constant of integration

00:11:01.080 --> 00:11:02.650
to worry about.

00:11:02.650 --> 00:11:06.220
And a new particular
solution, a lucky guess

00:11:06.220 --> 00:11:07.290
would just be a constant.

00:11:07.290 --> 00:11:11.990
And that would give us the
one half from 1 over 2.

00:11:11.990 --> 00:11:15.680
So then, we just need to seek
the c_1 that would give us

00:11:15.680 --> 00:11:18.620
x of t equals to 0 at 0.

00:11:18.620 --> 00:11:20.180
Because in this
case, we don't need

00:11:20.180 --> 00:11:21.710
to modify the
initial conditions.

00:11:21.710 --> 00:11:24.540
They still remain the same,
rest initial conditions.

00:11:24.540 --> 00:11:34.940
And so we would get here just
c_1 equals minus one half.

00:11:34.940 --> 00:11:39.190
So we have 1 minus
exponential of minus 2t.

00:11:39.190 --> 00:11:41.120
So just here, something
I forgot to mention,

00:11:41.120 --> 00:11:42.670
what could this be modeling?

00:11:42.670 --> 00:11:44.350
So this could be
modeling, if I go back

00:11:44.350 --> 00:11:48.400
to my analogy of the radioactive
chemical in the tank,

00:11:48.400 --> 00:11:50.530
this would be telling me
that after a certain time,

00:11:50.530 --> 00:11:52.400
I start inputting
at a steady rate

00:11:52.400 --> 00:11:55.160
at a constant rate of u of
t, of one per unit of time,

00:11:55.160 --> 00:11:57.596
the amount in the tank
and then looking at how

00:11:57.596 --> 00:11:58.720
the system evolves to that.

00:11:58.720 --> 00:12:02.680
So there's no abrupt change
that introduces a discontinuity.

00:12:02.680 --> 00:12:06.360
So just to conclude, I just
want to sketch the solution

00:12:06.360 --> 00:12:10.210
just so that you see the
difference between the two.

00:12:10.210 --> 00:12:15.297
And here, what we have, again, I
could have introduced my u of t

00:12:15.297 --> 00:12:19.320
here if I want t
in R. So here, we

00:12:19.320 --> 00:12:24.010
have again a solution
that is 0 before.

00:12:24.010 --> 00:12:25.320
There is no discontinuity.

00:12:25.320 --> 00:12:27.870
It's still 0 at 0 minus.

00:12:27.870 --> 00:12:30.600
And then, we have
basically a solution

00:12:30.600 --> 00:12:36.380
that is growing until it
reaches an asymptote of one half

00:12:36.380 --> 00:12:38.514
when this exponential goes to 0.

00:12:38.514 --> 00:12:40.930
And so you can see that it's
a smooth transition because I

00:12:40.930 --> 00:12:42.600
just started
inputting the amount

00:12:42.600 --> 00:12:46.470
of the chemical in the
tank in a non-abrupt way.

00:12:46.470 --> 00:12:48.180
So this concludes this session.

00:12:48.180 --> 00:12:50.010
And the key here was
to really remember

00:12:50.010 --> 00:12:53.700
what do we mean by the
unit step response.

00:12:53.700 --> 00:12:55.550
What type of f of t
are we talking about?

00:12:55.550 --> 00:12:57.410
What initial
conditions do we need?

00:12:57.410 --> 00:13:00.910
And same thing for the
unit impulse response.