WEBVTT

00:00:05.564 --> 00:00:06.480
MARTINA BALAGOVIC: Hi.

00:00:06.480 --> 00:00:08.010
Welcome to recitation.

00:00:08.010 --> 00:00:10.590
Today's problem is
about change of basis.

00:00:10.590 --> 00:00:13.030
It says the vector
space of polynomials

00:00:13.030 --> 00:00:18.725
in x of degree up to 2 has
a basis 1, x, and x squared.

00:00:18.725 --> 00:00:20.350
That's the obvious
basis that you would

00:00:20.350 --> 00:00:22.260
write for that vector space.

00:00:22.260 --> 00:00:25.150
But today we're going to
consider another basis, w_1,

00:00:25.150 --> 00:00:26.910
w_2, and w_3.

00:00:26.910 --> 00:00:30.310
And we don't know what w_1,
w_2, and w_3 are explicitly.

00:00:30.310 --> 00:00:33.860
What we know is that
their values at x

00:00:33.860 --> 00:00:38.270
equals minus 1, 0, and 1 are
given by this table here.

00:00:38.270 --> 00:00:43.920
So they are 1, 0, 0;
0, 1, 0; and 0, 0, 1.

00:00:43.920 --> 00:00:45.740
We're asked to do the following.

00:00:45.740 --> 00:00:48.800
We're asked to express
this polynomial-- so y

00:00:48.800 --> 00:00:54.160
of x is minus x plus 5-- in
this basis, w_1, w_2, w_3.

00:00:54.160 --> 00:00:56.100
We're asked to find
the change of basis

00:00:56.100 --> 00:01:00.180
matrices between these two
bases, 1, x, x squared,

00:01:00.180 --> 00:01:02.850
and w_1, w_2, w_3.

00:01:02.850 --> 00:01:05.440
And finally, we're asked to
find the matrix of taking

00:01:05.440 --> 00:01:08.200
derivatives, which is a
linear map on this space,

00:01:08.200 --> 00:01:10.500
in both of these basis.

00:01:10.500 --> 00:01:15.200
And let me give you an extra
level of challenge, which

00:01:15.200 --> 00:01:18.800
is to try to do as much of this
as possible without explicitly

00:01:18.800 --> 00:01:21.910
finding w_1, w_2, and w_3.

00:01:21.910 --> 00:01:23.410
I'll let you think
about the problem

00:01:23.410 --> 00:01:25.650
and then you can come back
and compare your solution

00:01:25.650 --> 00:01:26.150
with mine.

00:01:33.710 --> 00:01:34.230
Hi.

00:01:34.230 --> 00:01:35.370
Welcome back.

00:01:35.370 --> 00:01:40.990
So to start with
the problem a, we

00:01:40.990 --> 00:02:02.550
need to find coefficients alpha,
beta, and gamma so that y of x

00:02:02.550 --> 00:02:05.740
is expressed through,
with this coefficient,

00:02:05.740 --> 00:02:08.940
in this new basis,
w_1, w_2, and w_3.

00:02:08.940 --> 00:02:11.610
now, one way to do
that would be to look

00:02:11.610 --> 00:02:15.230
at this table of
values, explicitly

00:02:15.230 --> 00:02:21.320
find w_1, w_2, and w_3, so
a quadratic polynomial is--

00:02:21.320 --> 00:02:22.820
all the information
we need about it

00:02:22.820 --> 00:02:26.740
is in values at three points.

00:02:26.740 --> 00:02:34.110
So say w_1 is a*1 plus b times
x plus c times x squared.

00:02:34.110 --> 00:02:35.560
Find a, b, and c.

00:02:35.560 --> 00:02:40.230
Find w_1, w_2, w_3 explicitly,
and then go back to this system

00:02:40.230 --> 00:02:43.220
and try to find
alpha, beta and gamma.

00:02:43.220 --> 00:02:45.420
However, there's a trick.

00:02:45.420 --> 00:02:48.480
Let's try to see if we
can do it without finding

00:02:48.480 --> 00:02:50.830
w_1, w_2, and w_3 explicitly.

00:02:50.830 --> 00:02:55.010
So let me try to see
what are the values of y

00:02:55.010 --> 00:02:56.060
at these points.

00:02:56.060 --> 00:02:58.870
So y is minus x plus 5.

00:02:58.870 --> 00:03:05.300
So the values are 6, 5, and 4.

00:03:05.300 --> 00:03:09.300
And let me, instead of
considering this equation,

00:03:09.300 --> 00:03:18.840
let me evaluate it at x is
minus 1, x is 0, and x is 1.

00:03:18.840 --> 00:03:26.660
What I get through this is
that w at minus 1-- actually,

00:03:26.660 --> 00:03:27.480
let me write this.

00:03:30.420 --> 00:03:35.120
I get w at minus 1,
which is a number,

00:03:35.120 --> 00:03:40.970
equals alpha times
w_1 at minus 1,

00:03:40.970 --> 00:03:45.870
which is a number, plus
beta times w_2 at minus 1,

00:03:45.870 --> 00:03:51.100
plus gamma times w_3 at minus 1.

00:03:51.100 --> 00:03:58.110
And similarly at 0,
and similarly at 1.

00:03:58.110 --> 00:04:02.090
And now, let me think of
this as a linear system that

00:04:02.090 --> 00:04:05.850
has unknowns alpha,
beta, and gamma

00:04:05.850 --> 00:04:11.120
coefficients, these values
here at minus 1, 0, and 1.

00:04:11.120 --> 00:04:12.840
And the right-hand
side, well, what's

00:04:12.840 --> 00:04:16.500
written here on the left-hand
side. y at minus 1, y at 0,

00:04:16.500 --> 00:04:18.640
and y at 1.

00:04:18.640 --> 00:04:24.350
If you write this in a matrix
and read off coefficients

00:04:24.350 --> 00:04:26.740
from there, you get
the following system.

00:04:33.210 --> 00:04:38.380
So this is the matrix of the
system read off from here.

00:04:38.380 --> 00:04:39.350
These are the unknowns.

00:04:42.480 --> 00:04:48.699
And these are the values
of the right-hand side.

00:04:48.699 --> 00:04:50.990
And I hope you'll agree that
this is a very easy system

00:04:50.990 --> 00:04:51.490
to solve.

00:04:51.490 --> 00:04:55.900
We just get alpha is 6,
beta is 5, and gamma is 4.

00:04:55.900 --> 00:05:02.670
So the solution to the first
part is y equals 6*w_1 plus

00:05:02.670 --> 00:05:08.420
5*w_2 plus 4*w_3.

00:05:08.420 --> 00:05:10.510
And let's notice another thing.

00:05:10.510 --> 00:05:14.180
No matter what
values we put here,

00:05:14.180 --> 00:05:16.080
this matrix is always
going to stay the same.

00:05:16.080 --> 00:05:18.570
It's only the right-hand
side that's going to change.

00:05:18.570 --> 00:05:22.110
So if we're given
any other polynomial

00:05:22.110 --> 00:05:25.245
now to express in a
basis w_1, w_2, and w_3,

00:05:25.245 --> 00:05:26.620
we don't have to
do any thinking.

00:05:26.620 --> 00:05:28.410
We don't have to do
any computations.

00:05:28.410 --> 00:05:30.890
What we do is go back to
our table at the beginning

00:05:30.890 --> 00:05:33.170
and just read off-- let's
go back to the table,

00:05:33.170 --> 00:05:35.640
and just read these values.

00:05:35.640 --> 00:05:39.940
So in this case, y is 6
times w_1, 5 times w_2,

00:05:39.940 --> 00:05:40.960
and 3 times w_3.

00:05:43.530 --> 00:05:45.660
And that's already
a hint to solving

00:05:45.660 --> 00:05:48.170
the b part, which is
find the change of basis

00:05:48.170 --> 00:05:53.930
matrices between 1, x, x
squared and w_1, w_2, w_3.

00:05:53.930 --> 00:05:56.060
Change of basis matrices
means expressing

00:05:56.060 --> 00:05:58.180
one basis in terms of another.

00:05:58.180 --> 00:06:00.620
So as a part of the
problem, we will

00:06:00.620 --> 00:06:04.110
have to express 1,
x, and x squared

00:06:04.110 --> 00:06:06.930
in terms of w_1, w_2, w_3.

00:06:06.930 --> 00:06:10.360
So let's just find their
values at these three points.

00:06:10.360 --> 00:06:14.110
One is a constant, it just
takes value 1 everywhere.

00:06:14.110 --> 00:06:19.310
x takes value minus 1 at
minus 1, 0 at 0, and 1 at 1.

00:06:19.310 --> 00:06:26.140
And x squared takes values 1,
0, and 1 at minus 1, 0, and 1.

00:06:26.140 --> 00:06:30.630
And from this we can already
conclude the part b here,

00:06:30.630 --> 00:06:39.206
we can conclude that 1
equals w_1 plus w_2 plus w_3.

00:06:39.206 --> 00:06:46.680
That x equals
minus w_1 plus w_3.

00:06:46.680 --> 00:06:53.800
And that x squared
equals w_1 plus w_3.

00:06:53.800 --> 00:06:56.390
And from this, we
can immediately

00:06:56.390 --> 00:06:59.790
write one change
of basis matrix.

00:06:59.790 --> 00:07:06.520
Namely, since we know
how to express 1, x,

00:07:06.520 --> 00:07:12.820
and x squared in terms
of w_1, w_2, and w_3,

00:07:12.820 --> 00:07:15.200
we can just copy
this information over

00:07:15.200 --> 00:07:24.750
to this matrix, getting 1, 1,
1; minus 1, 0, 1; and 1, 0, 1.

00:07:24.750 --> 00:07:26.480
So which matrix is this?

00:07:26.480 --> 00:07:29.130
This is a matrix--
so we have 1, x,

00:07:29.130 --> 00:07:34.030
and x squared expressed in
terms of w_1, w_2, and w_3.

00:07:34.030 --> 00:07:37.750
So if we feed this matrix
something expressed

00:07:37.750 --> 00:07:41.360
in the basis 1, x, and
x squared, say a, b,

00:07:41.360 --> 00:07:44.600
and c, what it's
going to throw out

00:07:44.600 --> 00:07:49.460
is the same polynomial expressed
in this basis here, w_1, w_2,

00:07:49.460 --> 00:07:50.680
and w_3.

00:07:50.680 --> 00:07:57.137
So I'm going to just write that
this is a matrix of this basis

00:07:57.137 --> 00:07:57.637
change.

00:08:02.570 --> 00:08:04.170
How do we get the other one?

00:08:04.170 --> 00:08:08.210
Well, very easy.

00:08:08.210 --> 00:08:11.910
We know it's just
the inverse of A.

00:08:11.910 --> 00:08:18.220
So this is going to
be the matrix that

00:08:18.220 --> 00:08:24.240
takes something
written in this basis

00:08:24.240 --> 00:08:27.050
and transfers it to this basis.

00:08:27.050 --> 00:08:28.770
I'm not going to
calculate the inverse

00:08:28.770 --> 00:08:30.070
of a matrix in front of you.

00:08:30.070 --> 00:08:32.760
Instead I'm going to
consult my oracle.

00:08:32.760 --> 00:08:34.900
Sorry about that.

00:08:34.900 --> 00:08:38.650
And my oracle says
that the inverse

00:08:38.650 --> 00:08:52.336
should be 0, 1, 0; minus 1/2,
0, 1/2; and 1/2, minus 1, 1/2.

00:08:52.336 --> 00:08:53.460
And that solves the b part.

00:08:56.240 --> 00:08:57.420
Let's go into the c part.

00:08:57.420 --> 00:09:01.170
The c part required us to find
a matrix of taking derivatives,

00:09:01.170 --> 00:09:04.030
which is a linear map in
the space of polynomials,

00:09:04.030 --> 00:09:06.520
in both of these basis.

00:09:06.520 --> 00:09:10.570
So let's first do the
1, x, x squared basis

00:09:10.570 --> 00:09:12.860
because that one's easier.

00:09:12.860 --> 00:09:14.830
I'm going to call it D_x.

00:09:18.430 --> 00:09:22.760
So I'm going to work in
basis 1, x, x squared.

00:09:22.760 --> 00:09:26.370
And what I want to express
is the transformation

00:09:26.370 --> 00:09:27.400
of taking derivatives.

00:09:27.400 --> 00:09:31.250
So here I'm going
to write the vector

00:09:31.250 --> 00:09:35.425
to which taking derivatives maps
the polynomial 1, which is 0.

00:09:38.150 --> 00:09:41.930
And that's this expressed in
the basis 1, x, x squared.

00:09:41.930 --> 00:09:48.190
In the second column I'm
going to write x prime,

00:09:48.190 --> 00:09:51.240
the vector to which
D_x sends to vector

00:09:51.240 --> 00:09:56.120
x, and that's equal to 1,
which expressed in this basis

00:09:56.120 --> 00:09:58.070
is 1, 0, 0.

00:09:58.070 --> 00:10:02.050
And here I'm going
to write x squared

00:10:02.050 --> 00:10:07.700
prime, which is 2x, which,
expressed in this basis,

00:10:07.700 --> 00:10:10.750
is just this.

00:10:10.750 --> 00:10:12.060
That one was easy.

00:10:12.060 --> 00:10:15.270
For the other one,
well we could calculate

00:10:15.270 --> 00:10:19.430
w_1, w_2, w_3 explicitly,
take the derivatives,

00:10:19.430 --> 00:10:23.180
go back to the table
and repeat the procedure

00:10:23.180 --> 00:10:24.050
that we did already.

00:10:24.050 --> 00:10:27.769
So expressing these derivatives
in terms of w_1, w_2, w_3,

00:10:27.769 --> 00:10:28.810
and that's a lot of work.

00:10:28.810 --> 00:10:32.020
But we pretty much already
did most of this work.

00:10:32.020 --> 00:10:35.230
So we know how to take
derivatives in this basis,

00:10:35.230 --> 00:10:38.059
and we know how to go
between these two basis.

00:10:38.059 --> 00:10:40.350
So if we want to take a
derivative of something written

00:10:40.350 --> 00:10:48.780
in the basis w_1, w_2,
w_3, well, let's first

00:10:48.780 --> 00:10:53.872
write this something in
basis 1, x, x squared.

00:10:53.872 --> 00:10:55.330
Then let's take a
derivative of it.

00:10:57.880 --> 00:11:00.380
And then let's write it
back in the original basis

00:11:00.380 --> 00:11:02.050
that we want.

00:11:02.050 --> 00:11:05.060
So it's multiplication
of three matrices.

00:11:05.060 --> 00:11:07.990
We have all three-- matrix
multiplication is easy.

00:11:07.990 --> 00:11:11.350
And my oracle, again, says
that this should be minus 3

00:11:11.350 --> 00:11:21.470
over 2, 2, minus 1 over 2;
minus 1 over 2, 0, 1 over 2;

00:11:21.470 --> 00:11:28.080
and 1 over 2, minus 2, 3 over 2.

00:11:28.080 --> 00:11:29.681
And that solves the problem.

00:11:29.681 --> 00:11:31.930
Now, one thing that I would
want to discuss in the end

00:11:31.930 --> 00:11:34.470
is how did you do with
respect to my challenge, which

00:11:34.470 --> 00:11:36.580
was let's try to do as
much of it as possible

00:11:36.580 --> 00:11:39.650
without finding w_1,
w_2, and w_3 explicitly.

00:11:39.650 --> 00:11:41.250
And it seems like
we did really well.

00:11:41.250 --> 00:11:43.930
There's nowhere on the
board written w_1 equals,

00:11:43.930 --> 00:11:46.100
w_2 equals, w_3 equals.

00:11:46.100 --> 00:11:47.780
But is it really so?

00:11:47.780 --> 00:11:48.740
It's not.

00:11:48.740 --> 00:11:51.550
We calculated the matrix
of A inverse here.

00:11:51.550 --> 00:11:59.360
And what this really means is
that w_1, w_2, and w_3, written

00:11:59.360 --> 00:12:03.760
in the basis 1, x, and x
squared, are as follows.

00:12:03.760 --> 00:12:08.640
w_1 is minus 1/2 x
plus 1/2 x squared.

00:12:08.640 --> 00:12:12.150
w_2 is 1 minus x squared.

00:12:12.150 --> 00:12:16.300
And w_3 is 1/2 x
plus 1/2 x squared.

00:12:16.300 --> 00:12:19.270
So you can check your work
with the help of this matrix

00:12:19.270 --> 00:12:22.500
in case you did find w_1,
w_2, and w_3 explicitly.

00:12:22.500 --> 00:12:24.928
And that's all I
wanted to say today.