WEBVTT

00:00:06.915 --> 00:00:07.540
JOEL LEWIS: Hi.

00:00:07.540 --> 00:00:09.330
Welcome back to recitation.

00:00:09.330 --> 00:00:12.510
I've got a nice exercise here
on Stokes' Theorem for you.

00:00:12.510 --> 00:00:14.380
Now this problem is
a little bit more

00:00:14.380 --> 00:00:16.110
sophisticated than
a lot of problems

00:00:16.110 --> 00:00:17.630
we've been doing in recitation.

00:00:17.630 --> 00:00:20.400
So it requires a little bit
more thought, and it also

00:00:20.400 --> 00:00:22.940
involves more mathematical
sophistication.

00:00:22.940 --> 00:00:28.230
So we're doing a clever kind
of proof here that I like.

00:00:28.230 --> 00:00:31.170
So hopefully you'll
like this one.

00:00:31.170 --> 00:00:33.670
It's a little bit in a different
style than some of the ones

00:00:33.670 --> 00:00:35.600
we've done.

00:00:35.600 --> 00:00:38.000
So I think I need to talk
about it a little bit before we

00:00:38.000 --> 00:00:38.630
get started.

00:00:38.630 --> 00:00:41.580
So let's let F be
the field [x,  y, z].

00:00:41.580 --> 00:00:45.870
So this is our radial field that
we've seen a lot in recitation.

00:00:45.870 --> 00:00:47.840
So what I'd like
you to do is prove

00:00:47.840 --> 00:00:52.200
that this field is not the
curl of any field G. All right.

00:00:52.200 --> 00:00:53.910
So I'd like you to
show that there's

00:00:53.910 --> 00:00:58.790
no field G such that F is
equal to the curl of G.

00:00:58.790 --> 00:01:02.370
Now, rather than just
saying that to you

00:01:02.370 --> 00:01:04.750
and letting you run
off, I have a suggestion

00:01:04.750 --> 00:01:07.750
for an interesting way
you could go about this.

00:01:07.750 --> 00:01:11.340
And this interesting way is
going to use Stokes' Theorem.

00:01:11.340 --> 00:01:16.400
So what I'd like you to do
is a proof by contradiction.

00:01:16.400 --> 00:01:18.430
OK, so what you're
going to do is

00:01:18.430 --> 00:01:21.740
you're going to assume
that F is a curl.

00:01:21.740 --> 00:01:22.240
OK?

00:01:22.240 --> 00:01:23.890
So you're going to
assume that there

00:01:23.890 --> 00:01:28.920
is some G such that F is
curl G. And then you're

00:01:28.920 --> 00:01:33.890
going to use that to get
a ridiculous conclusion.

00:01:33.890 --> 00:01:35.760
So you're going to
start with that premise,

00:01:35.760 --> 00:01:38.960
and you're going to end
up with a contradiction.

00:01:38.960 --> 00:01:43.540
So these two arrows
colliding into each other

00:01:43.540 --> 00:01:46.810
is a symbol that mathematicians
use for a contradiction.

00:01:46.810 --> 00:01:48.640
So you're going to
start from this premise,

00:01:48.640 --> 00:01:50.460
and you're going to
reach a contradiction.

00:01:50.460 --> 00:01:52.750
And what that's going to
show is that your premise

00:01:52.750 --> 00:01:54.071
couldn't be right.

00:01:54.071 --> 00:01:54.570
Yeah?

00:01:54.570 --> 00:01:57.260
Because if you start
from a true premise,

00:01:57.260 --> 00:02:00.050
well then all your conclusions
should be true as well.

00:02:00.050 --> 00:02:02.610
So if you reach a
false conclusion,

00:02:02.610 --> 00:02:05.470
then you must have
had a false premise.

00:02:05.470 --> 00:02:07.950
So what you're going
to do is you're

00:02:07.950 --> 00:02:11.590
going to take a sphere of
radius b centered at the origin.

00:02:11.590 --> 00:02:13.750
And a curve C on the sphere.

00:02:13.750 --> 00:02:16.640
You know, a simple,
closed curve.

00:02:16.640 --> 00:02:21.730
So assuming that F is this curl
of G, what I'd like you to do

00:02:21.730 --> 00:02:25.330
is use Stokes'
Theorem to interpret

00:02:25.330 --> 00:02:31.170
the line integral of G dot dr
over C in two different ways.

00:02:31.170 --> 00:02:32.110
OK?

00:02:32.110 --> 00:02:35.520
And interpreting this line
integral in two different ways,

00:02:35.520 --> 00:02:37.130
you're going to reach
a contradiction,

00:02:37.130 --> 00:02:40.130
and that will show that
F really isn't a curl.

00:02:40.130 --> 00:02:41.760
So that's what I'd
like you to do.

00:02:41.760 --> 00:02:43.460
So why don't you
pause the video,

00:02:43.460 --> 00:02:45.966
go ahead and see if you can
work that out, come back,

00:02:45.966 --> 00:02:47.340
and we'll talk
about it together.

00:02:55.660 --> 00:02:57.910
I hope you enjoyed
working on this problem.

00:02:57.910 --> 00:02:59.320
Let's get started on it.

00:02:59.320 --> 00:03:03.940
So as I was saying before we
started, what we're going to do

00:03:03.940 --> 00:03:06.280
is we're looking for a
proof by contradiction.

00:03:06.280 --> 00:03:08.738
So as the problem says, we're
going to start with a sphere.

00:03:11.520 --> 00:03:14.170
And I'm going to
take this curve C--

00:03:14.170 --> 00:03:17.940
some simply connected closed
curve that's going to go around

00:03:17.940 --> 00:03:21.354
the back of the sphere, and
it's going to be oriented

00:03:21.354 --> 00:03:22.895
one way or the
other-- and it's going

00:03:22.895 --> 00:03:25.580
to divide this sphere
into two pieces.

00:03:25.580 --> 00:03:30.050
So there's the one cap
on one side of it, S_1.

00:03:30.050 --> 00:03:33.380
And then there's-- whatever the
other piece on the other side

00:03:33.380 --> 00:03:35.430
of it is, S_2.

00:03:35.430 --> 00:03:36.030
OK.

00:03:36.030 --> 00:03:38.720
And so what we're going to
do is we're going to think

00:03:38.720 --> 00:03:40.921
about, what is
this line integral?

00:03:40.921 --> 00:03:41.420
OK.

00:03:41.420 --> 00:03:45.340
So this is our curve
C here on the sphere.

00:03:45.340 --> 00:03:54.400
So the integral
over C of G dot dr.

00:03:54.400 --> 00:03:56.910
So this is what the problem
suggests we think about.

00:03:56.910 --> 00:03:59.780
So this is a line
integral of a field dot dr

00:03:59.780 --> 00:04:01.450
over the boundary of a surface.

00:04:01.450 --> 00:04:03.640
Well, actually, it's the
boundary of two surfaces.

00:04:03.640 --> 00:04:04.140
Right?

00:04:08.440 --> 00:04:11.610
C is the boundary of S1,
and C-- if we orient it

00:04:11.610 --> 00:04:15.120
the other way-- is the
boundary of S_2, when

00:04:15.120 --> 00:04:18.730
we orient them both outwards.

00:04:18.730 --> 00:04:21.660
OK, so what is this?

00:04:21.660 --> 00:04:23.429
So Stokes' Theorem
tells us something

00:04:23.429 --> 00:04:24.470
about this line integral.

00:04:24.470 --> 00:04:28.500
So let's first think about
this as the top cap-- that

00:04:28.500 --> 00:04:32.770
cap S1-- with boundary
C oriented so that they

00:04:32.770 --> 00:04:33.720
agree with each other.

00:04:33.720 --> 00:04:36.210
So the normal is
outwards on the sphere,

00:04:36.210 --> 00:04:38.730
and C is proceeding
in the direction

00:04:38.730 --> 00:04:40.390
that I've drawn the arrow here.

00:04:40.390 --> 00:04:43.140
Well, in that
circumstance, we have

00:04:43.140 --> 00:04:48.600
that the integral around C of
G dot dr, by Stokes' Theorem,

00:04:48.600 --> 00:05:00.950
is equal to the surface integral
over S_1 of curl of G dot

00:05:00.950 --> 00:05:04.710
n with respect to surface area.

00:05:04.710 --> 00:05:05.210
Right?

00:05:05.210 --> 00:05:06.290
So this is just Stokes' Theorem.

00:05:06.290 --> 00:05:08.020
Stokes' Theorem says
the line integral

00:05:08.020 --> 00:05:11.680
of G around the boundary
curve is equal to the surface

00:05:11.680 --> 00:05:14.980
integral of the curl
of G over the region,

00:05:14.980 --> 00:05:17.370
provided all of our
orientations are correct.

00:05:17.370 --> 00:05:17.870
OK.

00:05:17.870 --> 00:05:20.820
Well, we know though
what curl of G

00:05:20.820 --> 00:05:24.820
is, because by assumption,
F is equal to curl of G. OK,

00:05:24.820 --> 00:05:36.050
so this is equal to the surface
integral over S1 of F dot n dS.

00:05:36.050 --> 00:05:38.300
So in the first step,
we use Stokes' Theorem.

00:05:38.300 --> 00:05:41.090
In the second step,
we use our assumption

00:05:41.090 --> 00:05:45.410
that curl G is equal
to F. Well, now what?

00:05:45.410 --> 00:05:46.550
But we know what F is.

00:05:46.550 --> 00:05:47.050
Right?

00:05:47.050 --> 00:05:49.430
F is this radial
field [x, y,  z].

00:05:49.430 --> 00:05:52.760
So F and n are pointing
in the same direction.

00:05:52.760 --> 00:05:55.380
They're parallel to each
other. n is a unit vector,

00:05:55.380 --> 00:05:58.290
so this is just the
length of F. This F dot

00:05:58.290 --> 00:06:00.360
n is just the length
of F. And since we're

00:06:00.360 --> 00:06:04.310
on a sphere of radius
b, this is just b.

00:06:04.310 --> 00:06:06.680
OK, so the integrand is just b.

00:06:06.680 --> 00:06:14.530
So this is the
integral over S_1 of b

00:06:14.530 --> 00:06:23.580
dS, which is b times
the area of S_1.

00:06:23.580 --> 00:06:24.930
OK.

00:06:24.930 --> 00:06:27.440
One thing I'd like you to notice
is that in particular, this

00:06:27.440 --> 00:06:29.250
is a positive number.

00:06:29.250 --> 00:06:32.340
b is positive and
the area is positive.

00:06:32.340 --> 00:06:32.840
OK.

00:06:32.840 --> 00:06:35.150
So that's our first
interpretation.

00:06:35.150 --> 00:06:40.910
So we took our field G
that we suppose exists,

00:06:40.910 --> 00:06:44.260
and we integrate it
around this curve C,

00:06:44.260 --> 00:06:47.130
and we apply Stokes'
Theorem, and then

00:06:47.130 --> 00:06:49.720
the fact that we
know what F is means

00:06:49.720 --> 00:06:54.290
that we know what F dot n is,
and so that makes our surface

00:06:54.290 --> 00:06:55.860
integral very easy to compute.

00:06:55.860 --> 00:06:59.330
And it turns out to be b
times the area of S, which,

00:06:59.330 --> 00:07:03.360
I just happened to notice,
is a positive number.

00:07:03.360 --> 00:07:04.140
OK.

00:07:04.140 --> 00:07:06.790
Well, now we can
do the same trick

00:07:06.790 --> 00:07:08.981
on the other half of the sphere.

00:07:08.981 --> 00:07:09.480
Right?

00:07:09.480 --> 00:07:11.160
So we just did the top cap here.

00:07:11.160 --> 00:07:12.100
We did S_1.

00:07:12.100 --> 00:07:14.920
So now we have the
bottom cap, or whatever.

00:07:14.920 --> 00:07:17.951
All the rest of the sphere, S_2.

00:07:17.951 --> 00:07:18.450
OK.

00:07:18.450 --> 00:07:23.050
So we can also get that the
integral over C of G dot dr,

00:07:23.050 --> 00:07:25.200
we can interpret it in
terms of Stokes' Theorem.

00:07:25.200 --> 00:07:28.224
But notice then
that C-- we still

00:07:28.224 --> 00:07:29.390
want to use the same normal.

00:07:29.390 --> 00:07:31.140
We like outwards
pointing normals.

00:07:31.140 --> 00:07:34.450
So we're going to have to
orient C the other way in order

00:07:34.450 --> 00:07:36.480
to make Stokes'
Theorem make sense.

00:07:36.480 --> 00:07:40.619
So let's walk over here where
we have some empty board space.

00:07:40.619 --> 00:07:42.160
So we want to orient
C the other way.

00:07:42.160 --> 00:07:44.810
So in other words, we're
going to take the negative

00:07:44.810 --> 00:07:47.070
of this line integral.

00:07:47.070 --> 00:07:51.950
So it's minus G
dot dr. And if we

00:07:51.950 --> 00:07:54.980
apply Stokes' Theorem to
this line integral-- so this

00:07:54.980 --> 00:07:58.460
is the same line integral, but
with the opposite orientation

00:07:58.460 --> 00:08:03.690
on C and so with the opposite
sign-- by Stokes' Theorem,

00:08:03.690 --> 00:08:14.470
this is equal to the integral
over S_2 of the curl of G

00:08:14.470 --> 00:08:16.520
dot n dS.

00:08:16.520 --> 00:08:18.830
And so the way I've set
this up, this is still

00:08:18.830 --> 00:08:20.730
my outward-pointing normal.

00:08:20.730 --> 00:08:21.550
OK.

00:08:21.550 --> 00:08:23.950
But again, we can
use our assumption

00:08:23.950 --> 00:08:29.220
and we can get curl
of G is equal to F,

00:08:29.220 --> 00:08:31.760
because we're assuming
that G has this property.

00:08:31.760 --> 00:08:38.745
So this is equal to
the integral over S_2

00:08:38.745 --> 00:08:44.340
of F dot n with respect
to surface area.

00:08:44.340 --> 00:08:46.360
And again, n is the
outward-pointing normal,

00:08:46.360 --> 00:08:50.350
and F is parallel to it.

00:08:50.350 --> 00:08:52.090
So this dot product
is just the length

00:08:52.090 --> 00:08:54.930
of F. The outward
pointing unit normal.

00:08:54.930 --> 00:08:57.050
So this is just the
length of F, which is b.

00:08:57.050 --> 00:09:04.120
So this is equal to b
times the area of S_2,

00:09:04.120 --> 00:09:06.880
which is also positive.

00:09:06.880 --> 00:09:08.730
So what have we just shown?

00:09:08.730 --> 00:09:10.970
Well, we started
from the assumption

00:09:10.970 --> 00:09:14.840
that there exists a G such
that F is the curl of G.

00:09:14.840 --> 00:09:17.480
And starting from
that assumption--

00:09:17.480 --> 00:09:22.960
let's look-- we showed that
the line integral around C of G

00:09:22.960 --> 00:09:29.170
dot dr is equal to
some positive number.

00:09:29.170 --> 00:09:33.990
And we also showed, over
here, that the negative

00:09:33.990 --> 00:09:37.220
of the line integral
of G around C

00:09:37.220 --> 00:09:39.300
is equal to some
positive number.

00:09:39.300 --> 00:09:40.700
Well, this is clearly absurd.

00:09:40.700 --> 00:09:42.420
That can't be true.

00:09:42.420 --> 00:09:46.480
So starting from our assumption
that F was the curl of G--

00:09:46.480 --> 00:09:48.990
that there is a G such
the F is the curl of G--

00:09:48.990 --> 00:09:50.590
we reached an absurd conclusion.

00:09:50.590 --> 00:09:52.610
We reached a conclusion
that the same number

00:09:52.610 --> 00:09:55.210
is both positive and negative.

00:09:55.210 --> 00:09:56.280
But that can't happen.

00:09:56.280 --> 00:09:58.730
So that means our
premise had to be false.

00:09:58.730 --> 00:09:59.360
OK.

00:09:59.360 --> 00:10:00.485
So this is a contradiction.

00:10:22.284 --> 00:10:24.230
So our assumption is false.

00:10:24.230 --> 00:10:25.940
And our assumption
was-- that we used

00:10:25.940 --> 00:10:28.180
to get this whole thing
started-- that F was

00:10:28.180 --> 00:10:31.170
the curl of some G. All right.

00:10:31.170 --> 00:10:32.310
So what have we shown?

00:10:32.310 --> 00:10:35.340
So we used a nice argument
here with Stokes' Theorem

00:10:35.340 --> 00:10:38.360
in order to show that
certain fields aren't

00:10:38.360 --> 00:10:39.930
the curl of other fields.

00:10:39.930 --> 00:10:43.090
So Stokes' Theorem limits
the kind of fields that

00:10:43.090 --> 00:10:44.630
can be curls of other fields.

00:10:44.630 --> 00:10:48.060
Now perhaps, you may have
thought of other theorems

00:10:48.060 --> 00:10:51.210
that you can use that also
limit what sorts of fields

00:10:51.210 --> 00:10:51.930
can be curls.

00:10:51.930 --> 00:10:55.090
And so there are other ways
to reach this true conclusion

00:10:55.090 --> 00:10:58.250
that our field F-- whose
components are x, y, and z--

00:10:58.250 --> 00:11:00.310
is not the curl of any field.

00:11:00.310 --> 00:11:02.700
This isn't the only way
to reach that conclusion.

00:11:02.700 --> 00:11:06.000
But this is a nice way that
shows that Stokes' Theorem puts

00:11:06.000 --> 00:11:08.970
some limitations on what fields
can behave like if they're

00:11:08.970 --> 00:11:10.380
going to be curls.

00:11:10.380 --> 00:11:11.972
I'll stop there.