WEBVTT

00:00:06.975 --> 00:00:07.600
JOEL LEWIS: Hi.

00:00:07.600 --> 00:00:09.200
Welcome back to recitation.

00:00:09.200 --> 00:00:11.570
One of the things you've been
learning about in lecture

00:00:11.570 --> 00:00:13.830
is how to solve some
max-min problems.

00:00:13.830 --> 00:00:16.950
How to find the maximum
or minimum of a given

00:00:16.950 --> 00:00:21.300
function given some constraints
and figure out where it reaches

00:00:21.300 --> 00:00:22.990
its largest or smallest value.

00:00:22.990 --> 00:00:26.760
So here I have a particular
example of such a problem.

00:00:26.760 --> 00:00:30.400
So we're going to
build a cardboard box.

00:00:30.400 --> 00:00:33.910
And our cardboard box has to
meet the following criteria.

00:00:33.910 --> 00:00:38.190
So its volume has to be 3 units.

00:00:38.190 --> 00:00:42.960
The front and back
of the cardboard box

00:00:42.960 --> 00:00:46.260
are going to be made just of
single-thickness cardboard.

00:00:46.260 --> 00:00:48.370
But the two sides,
the left and right,

00:00:48.370 --> 00:00:50.290
are going to be
made double-thick.

00:00:50.290 --> 00:00:53.162
And the bottom is going
to be made triple-thick.

00:00:53.162 --> 00:00:54.620
We're not going to
have a top, it's

00:00:54.620 --> 00:00:55.828
just going to be an open box.

00:00:55.828 --> 00:00:59.700
So it's going to have five
sides: two of single thickness,

00:00:59.700 --> 00:01:01.590
two of double thickness,
and with the bottom

00:01:01.590 --> 00:01:02.815
of triple thickness.

00:01:04.760 --> 00:01:06.510
So, the question is--
there are, you know,

00:01:06.510 --> 00:01:11.205
a lot of different shapes of
box with total volume 3 units.

00:01:11.205 --> 00:01:14.140
So the question is what
dimensions should this box

00:01:14.140 --> 00:01:18.530
have so that we use the minimum
amount of cardboard possible?

00:01:18.530 --> 00:01:19.030
All right?

00:01:19.030 --> 00:01:21.190
So we want the
dimensions of the box

00:01:21.190 --> 00:01:23.670
that use the least
total amount cardboard.

00:01:23.670 --> 00:01:26.380
So why don't you pause the
video, work on that for awhile,

00:01:26.380 --> 00:01:36.870
come back, and we can
work on it together

00:01:36.870 --> 00:01:39.290
I hope you had some fun
puzzling this one out.

00:01:39.290 --> 00:01:41.190
Let's have a go at it.

00:01:41.190 --> 00:01:42.970
So to start, I think
I'd just like to draw

00:01:42.970 --> 00:01:45.630
a little picture of a box.

00:01:45.630 --> 00:01:53.980
So here's my cardboard box.

00:01:53.980 --> 00:01:55.750
Its top is going
to be open, say.

00:01:55.750 --> 00:01:58.250
So let's give it--
we want to figure out

00:01:58.250 --> 00:02:01.200
what its dimensions
should be in order

00:02:01.200 --> 00:02:05.051
to minimize some expression,
so let's give them names.

00:02:05.051 --> 00:02:05.550
Right?

00:02:05.550 --> 00:02:07.730
So I think a natural
thing to call

00:02:07.730 --> 00:02:11.520
them is we can call
one of them x and one

00:02:11.520 --> 00:02:14.940
of them y and the height z.

00:02:14.940 --> 00:02:16.300
There.

00:02:16.300 --> 00:02:21.330
So if we give the box these
dimensions, x, y, and z,

00:02:21.330 --> 00:02:24.320
we need to ask, well first
of all, what is its volume?

00:02:24.320 --> 00:02:26.580
And second of all, how
much cardboard is used?

00:02:26.580 --> 00:02:28.170
And as long as we
can keep the volume

00:02:28.170 --> 00:02:30.960
equal to 3, what's the least
amount of cardboard we can use?

00:02:30.960 --> 00:02:34.610
So the volume of this box is
just x*y*z and that has to be

00:02:34.610 --> 00:02:35.170
3.

00:02:35.170 --> 00:02:43.790
So we know x*y*z is equal to
the volume, which is equal to 3.

00:02:43.790 --> 00:02:47.720
So we have a constraint
on x, y, and z here.

00:02:47.720 --> 00:02:50.980
And the thing that
we want to optimize--

00:02:50.980 --> 00:02:55.360
that we want to
find a minimum for--

00:02:55.360 --> 00:02:57.860
is the total amount
of cardboard used.

00:03:04.830 --> 00:03:07.500
So let's talk about how
much cardboard is used.

00:03:07.500 --> 00:03:11.800
So the bottom of the
box is triple-thick.

00:03:11.800 --> 00:03:15.230
So its area is x times y.

00:03:15.230 --> 00:03:18.345
So the total amount of
cardboard used is 3x*y.

00:03:21.150 --> 00:03:27.220
Now, the front and back of
the box are single thickness.

00:03:27.220 --> 00:03:31.190
And they have area x*y, but
there-- sorry, not x*y, x*z,

00:03:31.190 --> 00:03:32.280
right?

00:03:32.280 --> 00:03:35.270
This height is z,
so it's x times z--

00:03:35.270 --> 00:03:36.420
and there are two of them.

00:03:36.420 --> 00:03:41.180
So we have x*z total cardboard
in front and x*z total

00:03:41.180 --> 00:03:41.950
cardboard in back.

00:03:41.950 --> 00:03:45.820
Oh, I guess this has another
segment there in back

00:03:45.820 --> 00:03:47.580
because it's an open box.

00:03:47.580 --> 00:03:49.590
So x*z and x*z.

00:03:49.590 --> 00:03:54.280
So that's 2x*z coming
from the front and back.

00:03:54.280 --> 00:03:59.380
And from the two sides,
well each side has area y*z.

00:03:59.380 --> 00:04:01.470
There are two of them
and they're double-thick.

00:04:01.470 --> 00:04:08.760
So that contributes 4y*z to the
total amount of cardboard used.

00:04:08.760 --> 00:04:11.270
So this is the total
amount of cardboard used.

00:04:11.270 --> 00:04:14.590
And we know x*y*z is equal to 3.

00:04:14.590 --> 00:04:17.820
So we want to
minimize this, but we

00:04:17.820 --> 00:04:20.352
want to minimize
this taking this

00:04:20.352 --> 00:04:22.850
into account, taking this
restriction on the volume

00:04:22.850 --> 00:04:23.550
into account.

00:04:23.550 --> 00:04:25.680
So what we can do
is you can realize

00:04:25.680 --> 00:04:28.650
that we can just eliminate
one of these variables.

00:04:28.650 --> 00:04:29.210
Right?

00:04:29.210 --> 00:04:31.760
We can say z is
equal to 3 over x*y.

00:04:31.760 --> 00:04:33.490
That always has to be true.

00:04:33.490 --> 00:04:36.360
And if we make that
substitution in here,

00:04:36.360 --> 00:04:39.940
then we'll be able to minimize
this expression without regard

00:04:39.940 --> 00:04:41.700
to any constraint anymore.

00:04:41.700 --> 00:04:53.440
So from x*y*z equals 3, we
have z equals 3 over x*y.

00:04:53.440 --> 00:05:07.100
So we want to minimize what
we get when we plug z equals 3

00:05:07.100 --> 00:05:08.310
over x*y into here.

00:05:08.310 --> 00:05:19.810
So that's 3x*y, plus-- well, 2x
times 3 over x*y is 6-- over y,

00:05:19.810 --> 00:05:26.900
plus-- and 4y times 3
over x*y is 12-- over x.

00:05:26.900 --> 00:05:30.820
And one other thing I guess we
haven't mentioned explicitly

00:05:30.820 --> 00:05:33.420
is that this is a physical box.

00:05:33.420 --> 00:05:37.540
It has actual dimensions,
so its dimensions have

00:05:37.540 --> 00:05:39.490
to be, you know--
they're lengths,

00:05:39.490 --> 00:05:43.050
they have to be
positive numbers.

00:05:43.050 --> 00:05:50.570
So we want for x to be
positive and we want for y

00:05:50.570 --> 00:05:52.320
to be positive numbers.

00:05:52.320 --> 00:05:53.250
OK.

00:05:53.250 --> 00:05:55.540
So now we've got this function.

00:05:55.540 --> 00:05:57.920
And there aren't-- x
and y can be anything,

00:05:57.920 --> 00:05:59.800
any value of x and y we choose.

00:05:59.800 --> 00:06:01.960
This describes the
amount of cardboard

00:06:01.960 --> 00:06:04.420
used in a box with
those dimensions

00:06:04.420 --> 00:06:07.390
on its base that has volume 3.

00:06:07.390 --> 00:06:07.970
All right?

00:06:07.970 --> 00:06:09.770
So we want to minimize that.

00:06:09.770 --> 00:06:13.990
So now we've finally got to
the point of our calculus.

00:06:13.990 --> 00:06:17.260
Let's call this function--
let's give it a name

00:06:17.260 --> 00:06:22.800
like, I don't know-- f of x, y.

00:06:22.800 --> 00:06:25.980
So we want to find the
global minimum of f.

00:06:25.980 --> 00:06:28.330
We want to find the absolute
least amount of cardboard

00:06:28.330 --> 00:06:30.110
that we can use.

00:06:30.110 --> 00:06:32.340
So there are several
possibilities

00:06:32.340 --> 00:06:34.570
for where a global
minimum can occur.

00:06:34.570 --> 00:06:40.990
It can occur on some critical
point of the function,

00:06:40.990 --> 00:06:43.720
or it can occur on the
boundary of the region

00:06:43.720 --> 00:06:45.400
where the function is defined.

00:06:45.400 --> 00:06:47.930
And in this case, that
includes the possibility

00:06:47.930 --> 00:06:52.790
that it can occur as x or y
or both go off to infinity.

00:06:52.790 --> 00:06:53.350
All right?

00:06:53.350 --> 00:06:57.290
So we have to look at
those possibilities.

00:06:57.290 --> 00:07:00.040
In particular, we have to
look at the critical points.

00:07:00.040 --> 00:07:02.010
That's one of the possibilities.

00:07:02.010 --> 00:07:03.880
So the critical points of f.

00:07:03.880 --> 00:07:07.680
So we need to find out
what those points are.

00:07:10.990 --> 00:07:13.490
So in order to find the critical
points, well what do we do?

00:07:13.490 --> 00:07:14.840
We do the usual thing.

00:07:14.840 --> 00:07:16.550
We look at its
partial derivatives.

00:07:16.550 --> 00:07:20.230
So we need the first
and second-- or sorry--

00:07:20.230 --> 00:07:23.230
the first partial derivatives
with respect to x and y,

00:07:23.230 --> 00:07:26.030
we need both of them
to be equal to 0.

00:07:26.030 --> 00:07:34.520
So we need f_x equals
0, and f_y equals 0.

00:07:34.520 --> 00:07:38.880
OK, so let's write down
what that means over here.

00:07:38.880 --> 00:07:40.330
So what is f_x?

00:07:40.330 --> 00:07:42.760
Well, we just take the partials.

00:07:42.760 --> 00:07:47.270
So it's 3y-- plus we take the
partial of 6y with respect

00:07:47.270 --> 00:07:50.540
to x and we get 0, and we
take the partial of 12 over x

00:07:50.540 --> 00:07:55.030
with respect to x-- and we
get minus 12 over x squared.

00:07:55.030 --> 00:08:01.550
So that's f_x, and we need
that to be equal to 0.

00:08:01.550 --> 00:08:09.490
And f_y is equal to 3x
minus 6 over y squared,

00:08:09.490 --> 00:08:13.030
and we also need that
to be equal to 0.

00:08:13.030 --> 00:08:15.340
So if we solve this first
equation, for example,

00:08:15.340 --> 00:08:17.820
we can solve it for
y in terms of x.

00:08:17.820 --> 00:08:22.490
And that tells me
that y is equal to--

00:08:22.490 --> 00:08:25.270
I need-- let's see-- 12
over x squared divided by 3,

00:08:25.270 --> 00:08:29.880
so that's 4 over x squared.

00:08:29.880 --> 00:08:33.370
Now I can plug this 4 over
x squared in down here,

00:08:33.370 --> 00:08:38.630
and so I get 3x minus 6.

00:08:38.630 --> 00:08:42.580
Well, 6 divided by 4 divided
by x squared quantity

00:08:42.580 --> 00:08:50.510
squared is 6 times x
squared over 4 squared.

00:08:50.510 --> 00:08:52.080
That has to be equal to 0.

00:08:52.080 --> 00:08:54.710
And I can-- let's see--
I can rewrite this.

00:08:54.710 --> 00:08:57.265
Maybe I'll divide
through by 3 as

00:08:57.265 --> 00:08:58.690
that 3 isn't going to matter.

00:08:58.690 --> 00:09:03.540
So it's x minus-- so then
I'll be left with 2 times x

00:09:03.540 --> 00:09:13.250
to the fourth over 16, so
that's-- x to the fourth over 8

00:09:13.250 --> 00:09:14.170
equals 0.

00:09:14.170 --> 00:09:19.360
And you can see in this equation
that either x is equal to 0,

00:09:19.360 --> 00:09:23.140
or we can divide by x and
then we get x cubed equals 8.

00:09:23.140 --> 00:09:28.120
So x equals 0 or
x cubed equals 8.

00:09:28.120 --> 00:09:32.160
The only solution is x equals 2.

00:09:32.160 --> 00:09:34.320
Now, it's easy to see
that we can't have

00:09:34.320 --> 00:09:37.770
a box with x equal to 0, right?

00:09:37.770 --> 00:09:42.520
Our function is not actually
defined at x equals 0 over

00:09:42.520 --> 00:09:43.610
here, you'll see, right.

00:09:43.610 --> 00:09:47.480
Our function had
a 12 over x in it,

00:09:47.480 --> 00:09:48.880
so we can't have x equal to 0.

00:09:48.880 --> 00:09:50.630
So that's not going
to be a critical point

00:09:50.630 --> 00:09:52.250
of this function.
x equals 0 isn't

00:09:52.250 --> 00:09:54.000
going to lead to a
critical point, so, OK,

00:09:54.000 --> 00:09:56.645
so our only critical
points are going

00:09:56.645 --> 00:09:59.000
to happen when x is equal to 2.

00:09:59.000 --> 00:10:01.200
And when x is equal to
2, we can go back up here

00:10:01.200 --> 00:10:04.240
and we see that y
was equal to 4 over x

00:10:04.240 --> 00:10:05.780
squared at our critical points.

00:10:05.780 --> 00:10:21.370
So the only critical
point is (2, 1).

00:10:21.370 --> 00:10:22.210
OK.

00:10:22.210 --> 00:10:23.700
So that's our critical point.

00:10:23.700 --> 00:10:26.310
So now, we want
to think about is

00:10:26.310 --> 00:10:31.730
this point going to be a
maximum or a minimum or what?

00:10:31.730 --> 00:10:36.010
And where is the global
minimum of this function

00:10:36.010 --> 00:10:37.300
going to occur?

00:10:37.300 --> 00:10:39.900
So remember that the global
minimum of this function

00:10:39.900 --> 00:10:44.170
can occur either on the critical
points or on the boundary

00:10:44.170 --> 00:10:49.790
or at a point where x or
y is going to infinity--

00:10:49.790 --> 00:10:51.470
in the limit, I guess I mean.

00:10:51.470 --> 00:10:55.300
So lets go back here and take
a look at our function here.

00:10:55.300 --> 00:10:59.590
So we can see from the
expression for this function

00:10:59.590 --> 00:11:03.270
that if x and y are going to
infinity, this is no good.

00:11:03.270 --> 00:11:05.030
This part is going
to go to infinity.

00:11:05.030 --> 00:11:07.446
3x*y is going to go to infinity,
and these are going to be

00:11:07.446 --> 00:11:10.160
positive, so f is going
to go to infinity.

00:11:10.160 --> 00:11:14.980
OK, so as the dimensions of
our box get very, very large,

00:11:14.980 --> 00:11:19.010
the total amount of cardboard
used becomes infinite.

00:11:19.010 --> 00:11:22.140
Also, if x or y gets
closer and closer

00:11:22.140 --> 00:11:25.230
to 0-- that's at
the boundary-- well,

00:11:25.230 --> 00:11:27.020
if x gets closer
and closer to 0,

00:11:27.020 --> 00:11:31.240
than 12 over x gets closer
and closer to infinity.

00:11:31.240 --> 00:11:35.020
And if y gets closer and
closer to 0, then 6 over y

00:11:35.020 --> 00:11:36.530
gets closer and
closer to infinity.

00:11:36.530 --> 00:11:38.030
And again, all these
terms are going

00:11:38.030 --> 00:11:40.210
to be positive because x
and y have to be positive.

00:11:40.210 --> 00:11:45.180
So in those limiting cases,
the function f of x, y

00:11:45.180 --> 00:11:46.890
gets infinitely large.

00:11:46.890 --> 00:11:49.635
So along the boundary and
as x and y go to infinity,

00:11:49.635 --> 00:11:50.940
this function blows up.

00:11:50.940 --> 00:11:52.160
It gets very, very big.

00:11:52.160 --> 00:11:55.010
So that means the only place
it could have a global-- so it

00:11:55.010 --> 00:11:58.850
has a global minimum-- the only
place that global minimum can

00:11:58.850 --> 00:12:01.080
be is at a critical point.

00:12:01.080 --> 00:12:03.190
And we saw, if we
walk back over here

00:12:03.190 --> 00:12:07.240
again, that the only critical
point is the point (2, 1).

00:12:07.240 --> 00:12:07.740
All right?

00:12:07.740 --> 00:12:09.870
So that means this point
has to be a global minimum

00:12:09.870 --> 00:12:10.910
for the function.

00:12:10.910 --> 00:12:12.617
Now, you might,
out of curiosity,

00:12:12.617 --> 00:12:14.950
you might ask how much cardboard
are we actually talking

00:12:14.950 --> 00:12:17.170
about in that case?

00:12:17.170 --> 00:12:19.901
Since that's really
the one part--

00:12:19.901 --> 00:12:22.710
I guess it wasn't really
part of the question,

00:12:22.710 --> 00:12:25.310
but it's an interesting
thing to ask.

00:12:25.310 --> 00:12:35.560
So in this case, we use
f of 2, 1 equals-- well,

00:12:35.560 --> 00:12:44.180
let's see-- so that's 6 plus 6
plus 6 equals 18 units squared

00:12:44.180 --> 00:12:46.560
of cardboard.

00:12:46.560 --> 00:12:49.910
Now, I guess one
thing you might notice

00:12:49.910 --> 00:12:52.400
is we didn't use the second
derivative test here.

00:12:52.400 --> 00:12:54.760
And the question
is why-- I mean,

00:12:54.760 --> 00:12:58.490
we concluded it was a global
maximum without ever using

00:12:58.490 --> 00:13:01.200
the second derivative test.

00:13:01.200 --> 00:13:02.250
Sorry, global minimum.

00:13:02.250 --> 00:13:03.269
Pardon me.

00:13:03.269 --> 00:13:05.060
And the second derivative
test is the thing

00:13:05.060 --> 00:13:06.870
that tells us whether
things are minimum

00:13:06.870 --> 00:13:08.036
or maximum or saddle points.

00:13:08.036 --> 00:13:09.260
So why didn't we use it?

00:13:09.260 --> 00:13:11.200
Well, the answer is the
second derivative test

00:13:11.200 --> 00:13:15.630
tells you whether something
is a local minimum or maximum

00:13:15.630 --> 00:13:18.220
or whether it's a saddle point.

00:13:18.220 --> 00:13:20.440
So if we had applied the
second derivative test

00:13:20.440 --> 00:13:22.610
at this critical point,
what we would have learned

00:13:22.610 --> 00:13:25.480
is that this critical
point is a local minimum.

00:13:25.480 --> 00:13:28.486
But being a local minimum
isn't enough to guarantee

00:13:28.486 --> 00:13:29.610
that it's a global minimum.

00:13:29.610 --> 00:13:31.600
Because we could have
on the boundary--

00:13:31.600 --> 00:13:33.544
or as x or y goes
to infinity-- we

00:13:33.544 --> 00:13:35.960
could have that the function
value get smaller and smaller

00:13:35.960 --> 00:13:36.830
without bound.

00:13:36.830 --> 00:13:39.720
Now in this case, we showed that
the function doesn't do that.

00:13:39.720 --> 00:13:41.660
It gets larger and
larger without bound.

00:13:41.660 --> 00:13:45.170
And so that meant that
that minimum point really

00:13:45.170 --> 00:13:47.430
is the global minimum.

00:13:47.430 --> 00:13:49.300
But the second
derivative test isn't

00:13:49.300 --> 00:13:51.690
enough to conclude
that something

00:13:51.690 --> 00:13:53.030
is a global minimum on its own.

00:13:53.030 --> 00:13:56.600
You really do need that
extra analysis that we did.

00:13:56.600 --> 00:13:57.872
I'll end there.