WEBVTT

00:00:00.000 --> 00:00:00.730


00:00:00.730 --> 00:00:06.870
For any transformation that maps
from Rn to Rn, we've done

00:00:06.870 --> 00:00:09.590
it implicitly, but it's been
interesting for us to find the

00:00:09.590 --> 00:00:12.460
vectors that essentially just
get scaled up by the

00:00:12.460 --> 00:00:13.880
transformations.

00:00:13.880 --> 00:00:17.230
So the vectors that have the
form-- the transformation of

00:00:17.230 --> 00:00:20.950
my vector is just equal
to some scaled-up

00:00:20.950 --> 00:00:22.035
version of a vector.

00:00:22.035 --> 00:00:24.290
And if this doesn't look
familiar, I can jog your

00:00:24.290 --> 00:00:25.750
memory a little bit.

00:00:25.750 --> 00:00:27.690
When we were looking for
basis vectors for the

00:00:27.690 --> 00:00:29.140
transformation--
let me draw it.

00:00:29.140 --> 00:00:31.190
This was from R2 to R2.

00:00:31.190 --> 00:00:33.970


00:00:33.970 --> 00:00:37.110
So let me draw R2 right here.

00:00:37.110 --> 00:00:44.310
And let's say I had the
vector v1 was equal to

00:00:44.310 --> 00:00:45.870
the vector 1, 2.

00:00:45.870 --> 00:00:48.990
And we had the lines spanned
by that vector.

00:00:48.990 --> 00:00:52.000
We did this problem several
videos ago.

00:00:52.000 --> 00:00:55.350
And I had the transformation
that flipped across this line.

00:00:55.350 --> 00:01:01.230
So if we call that line l, T was
the transformation from R2

00:01:01.230 --> 00:01:05.410
to R2 that flipped vectors
across this line.

00:01:05.410 --> 00:01:13.210
So it flipped vectors
across l.

00:01:13.210 --> 00:01:15.740
So if you remember that
transformation, if I had some

00:01:15.740 --> 00:01:19.050
random vector that looked like
that, let's say that's x,

00:01:19.050 --> 00:01:21.640
that's vector x, then the
transformation of x looks

00:01:21.640 --> 00:01:22.410
something like this.

00:01:22.410 --> 00:01:24.640
It's just flipped across
that line.

00:01:24.640 --> 00:01:26.770
That was the transformation
of x.

00:01:26.770 --> 00:01:28.990
And if you remember that video,
we were looking for a

00:01:28.990 --> 00:01:31.670
change of basis that would allow
us to at least figure

00:01:31.670 --> 00:01:34.640
out the matrix for the
transformation, at least in an

00:01:34.640 --> 00:01:35.500
alternate basis.

00:01:35.500 --> 00:01:36.900
And then we could figure
out the matrix for the

00:01:36.900 --> 00:01:38.950
transformation in the
standard basis.

00:01:38.950 --> 00:01:42.790
And the basis we picked were
basis vectors that didn't get

00:01:42.790 --> 00:01:44.950
changed much by the
transformation, or ones that

00:01:44.950 --> 00:01:46.940
only got scaled by the
transformation.

00:01:46.940 --> 00:01:52.750
For example, when I took the
transformation of v1, it just

00:01:52.750 --> 00:01:54.320
equaled v1.

00:01:54.320 --> 00:01:59.380
Or we could say that the
transformation of v1 just

00:01:59.380 --> 00:02:02.800
equaled 1 times v1.

00:02:02.800 --> 00:02:06.780
So if you just follow this
little format that I set up

00:02:06.780 --> 00:02:08.860
here, lambda, in this
case, would be 1.

00:02:08.860 --> 00:02:11.360
And of course, the vector
in this case is v1.

00:02:11.360 --> 00:02:16.395
The transformation just
scaled up v1 by 1.

00:02:16.395 --> 00:02:18.860
In that same problem, we had
the other vector that

00:02:18.860 --> 00:02:22.450
we also looked at.

00:02:22.450 --> 00:02:28.270
It was the vector minus-- let's
say it's the vector v2,

00:02:28.270 --> 00:02:32.410
which is-- let's say
it's 2, minus 1.

00:02:32.410 --> 00:02:34.420
And then if you take the
transformation of it, since it

00:02:34.420 --> 00:02:36.250
was orthogonal to the
line, it just got

00:02:36.250 --> 00:02:37.840
flipped over like that.

00:02:37.840 --> 00:02:39.760
And that was a pretty
interesting vector force as

00:02:39.760 --> 00:02:44.960
well, because the transformation
of v2 in this

00:02:44.960 --> 00:02:47.050
situation is equal to what?

00:02:47.050 --> 00:02:48.930
Just minus v2.

00:02:48.930 --> 00:02:50.270
It's equal to minus v2.

00:02:50.270 --> 00:02:54.920
Or you could say that the
transformation of v2 is equal

00:02:54.920 --> 00:02:58.230
to minus 1 times v2.

00:02:58.230 --> 00:03:01.870
And these were interesting
vectors for us because when we

00:03:01.870 --> 00:03:06.390
defined a new basis with these
guys as the basis vector, it

00:03:06.390 --> 00:03:09.280
was very easy to figure out
our transformation matrix.

00:03:09.280 --> 00:03:12.000
And actually, that basis was
very easy to compute with.

00:03:12.000 --> 00:03:14.390
And we'll explore that a little
bit more in the future.

00:03:14.390 --> 00:03:16.620
But hopefully you realize that
these are interesting vectors.

00:03:16.620 --> 00:03:21.750
There was also the cases where
we had the planes spanned by

00:03:21.750 --> 00:03:23.630
some vectors.

00:03:23.630 --> 00:03:25.820
And then we had another vector
that was popping out of the

00:03:25.820 --> 00:03:27.040
plane like that.

00:03:27.040 --> 00:03:29.320
And we were transforming things
by taking the mirror

00:03:29.320 --> 00:03:31.200
image across this and we're
like, well in that

00:03:31.200 --> 00:03:34.360
transformation, these red
vectors don't change at all

00:03:34.360 --> 00:03:35.960
and this guy gets
flipped over.

00:03:35.960 --> 00:03:38.290
So maybe those would make
for good bases.

00:03:38.290 --> 00:03:40.250
Or those would make for
good basis vectors.

00:03:40.250 --> 00:03:41.240
And they did.

00:03:41.240 --> 00:03:44.850
So in general, we're always
interested with the vectors

00:03:44.850 --> 00:03:47.240
that just get scaled up
by a transformation.

00:03:47.240 --> 00:03:49.080
It's not going to be
all vectors, right?

00:03:49.080 --> 00:03:51.320
This vector that I drew here,
this vector x, it doesn't just

00:03:51.320 --> 00:03:54.650
get scaled up, it actually gets
changed, this direction

00:03:54.650 --> 00:03:56.730
gets changed.

00:03:56.730 --> 00:04:00.360
The vectors that get scaled up
might switch direct-- might go

00:04:00.360 --> 00:04:03.020
from this direction to that
direction, or maybe

00:04:03.020 --> 00:04:04.430
they go from that.

00:04:04.430 --> 00:04:07.270
Maybe that's x and then the
transformation of x might be a

00:04:07.270 --> 00:04:08.460
scaled up version of x.

00:04:08.460 --> 00:04:09.710
Maybe it's that.

00:04:09.710 --> 00:04:12.050


00:04:12.050 --> 00:04:16.970
The actual, I guess, line that
they span will not change.

00:04:16.970 --> 00:04:19.350
And so that's what we're going
to concern ourselves with.

00:04:19.350 --> 00:04:21.019
These have a special name.

00:04:21.019 --> 00:04:23.660
And they have a special name and
I want to make this very

00:04:23.660 --> 00:04:25.050
clear because they're useful.

00:04:25.050 --> 00:04:27.360
It's not just some mathematical
game we're

00:04:27.360 --> 00:04:29.970
playing, although sometimes
we do fall into that trap.

00:04:29.970 --> 00:04:31.250
But they're actually useful.

00:04:31.250 --> 00:04:34.140
They're useful for defining
bases because in those bases

00:04:34.140 --> 00:04:36.730
it's easier to find
transformation matrices.

00:04:36.730 --> 00:04:38.950
They're more natural coordinate
systems. And

00:04:38.950 --> 00:04:41.700
oftentimes, the transformation
matrices in those bases are

00:04:41.700 --> 00:04:43.620
easier to compute with.

00:04:43.620 --> 00:04:47.060
And so these have
special names.

00:04:47.060 --> 00:04:50.040
Any vector that satisfies this
right here is called an

00:04:50.040 --> 00:04:57.810
eigenvector for the
transformation T.

00:04:57.810 --> 00:05:01.680
And the lambda, the multiple
that it becomes-- this is the

00:05:01.680 --> 00:05:12.410
eigenvalue associated with
that eigenvector.

00:05:12.410 --> 00:05:16.870


00:05:16.870 --> 00:05:19.590
So in the example I just gave
where the transformation is

00:05:19.590 --> 00:05:24.020
flipping around this line,
v1, the vector 1, 2 is an

00:05:24.020 --> 00:05:27.210
eigenvector of our
transformation.

00:05:27.210 --> 00:05:31.080
So 1, 2 is an eigenvector.

00:05:31.080 --> 00:05:33.960


00:05:33.960 --> 00:05:36.305
And it's corresponding
eigenvalue is 1.

00:05:36.305 --> 00:05:42.170


00:05:42.170 --> 00:05:43.820
This guy is also an
eigenvector-- the

00:05:43.820 --> 00:05:45.270
vector 2, minus 1.

00:05:45.270 --> 00:05:47.520
He's also an eigenvector.

00:05:47.520 --> 00:05:50.440
A very fancy word, but all it
means is a vector that's just

00:05:50.440 --> 00:05:51.920
scaled up by a transformation.

00:05:51.920 --> 00:05:55.030
It doesn't get changed in any
more meaningful way than just

00:05:55.030 --> 00:05:56.270
the scaling factor.

00:05:56.270 --> 00:06:03.860
And it's corresponding
eigenvalue is minus 1.

00:06:03.860 --> 00:06:05.580
If this transformation--
I don't know what its

00:06:05.580 --> 00:06:06.750
transformation matrix is.

00:06:06.750 --> 00:06:07.990
I forgot what it was.

00:06:07.990 --> 00:06:10.820
We actually figured it
out a while ago.

00:06:10.820 --> 00:06:16.490
If this transformation matrix
can be represented as a matrix

00:06:16.490 --> 00:06:18.180
vector product-- and it should
be; it's a linear

00:06:18.180 --> 00:06:22.940
transformation-- then any
v that satisfies the

00:06:22.940 --> 00:06:27.610
transformation of-- I'll say
transformation of v is equal

00:06:27.610 --> 00:06:32.520
to lambda v, which also would
be-- you know, the

00:06:32.520 --> 00:06:33.180
transformation of [? v ?]

00:06:33.180 --> 00:06:36.380
would just be A times v.

00:06:36.380 --> 00:06:39.390
These are also called
eigenvectors of A, because A

00:06:39.390 --> 00:06:41.570
is just really the matrix
representation of the

00:06:41.570 --> 00:06:43.090
transformation.

00:06:43.090 --> 00:06:51.560
So in this case, this would be
an eigenvector of A, and this

00:06:51.560 --> 00:06:53.690
would be the eigenvalue
associated with the

00:06:53.690 --> 00:06:54.940
eigenvector.

00:06:54.940 --> 00:06:58.700


00:06:58.700 --> 00:07:00.940
So if you give me a matrix that
represents some linear

00:07:00.940 --> 00:07:01.880
transformation.

00:07:01.880 --> 00:07:03.880
You can also figure
these things out.

00:07:03.880 --> 00:07:05.730
Now the next video we're
actually going to figure out a

00:07:05.730 --> 00:07:07.080
way to figure these
things out.

00:07:07.080 --> 00:07:10.320
But what I want you to
appreciate in this video is

00:07:10.320 --> 00:07:13.920
that it's easy to say,
oh, the vectors that

00:07:13.920 --> 00:07:15.130
don't get changed much.

00:07:15.130 --> 00:07:16.620
But I want you to understand
what that means.

00:07:16.620 --> 00:07:19.860
It literally just gets scaled up
or maybe they get reversed.

00:07:19.860 --> 00:07:22.060
Their direction or the
lines they span

00:07:22.060 --> 00:07:23.460
fundamentally don't change.

00:07:23.460 --> 00:07:26.400
And the reason why they're
interesting for us is, well,

00:07:26.400 --> 00:07:28.790
one of the reasons why they're
interesting for us is that

00:07:28.790 --> 00:07:32.590
they make for interesting basis
vectors-- basis vectors

00:07:32.590 --> 00:07:36.530
whose transformation matrices
are maybe computationally more

00:07:36.530 --> 00:07:41.610
simpler, or ones that make for
better coordinate systems.

00:07:41.610 --> 00:07:42.094