WEBVTT

00:00:05.777 --> 00:00:06.610
PROFESSOR: Hi there.

00:00:06.610 --> 00:00:07.370
My name is Ana.

00:00:07.370 --> 00:00:09.090
Welcome to recitation.

00:00:09.090 --> 00:00:12.030
In lecture, you've been learning
about how to multiply matrices,

00:00:12.030 --> 00:00:13.780
and how to think about
that multiplication

00:00:13.780 --> 00:00:17.410
in different ways, and
also about when a matrix is

00:00:17.410 --> 00:00:20.580
invertible or not, and how
to compute the inverse when

00:00:20.580 --> 00:00:21.850
it is invertible.

00:00:21.850 --> 00:00:24.360
And that's what today's
problem is about.

00:00:24.360 --> 00:00:28.330
We have a matrix A that
has variables a and b

00:00:28.330 --> 00:00:29.290
instead of numbers.

00:00:29.290 --> 00:00:33.730
And the question is: what are
the conditions on a and b that

00:00:33.730 --> 00:00:36.150
make that matrix invertible?

00:00:36.150 --> 00:00:40.550
And when it is invertible,
what is A inverse?

00:00:40.550 --> 00:00:43.389
Why don't you hit Pause and
work on it for a little while.

00:00:43.389 --> 00:00:45.430
And then we'll come back
and work on it together.

00:00:53.910 --> 00:00:55.220
And we're back.

00:00:55.220 --> 00:00:59.040
I hope you had success
in solving that.

00:00:59.040 --> 00:01:01.730
Let's do it ourselves.

00:01:01.730 --> 00:01:06.090
So, remember from
lecture, we talked

00:01:06.090 --> 00:01:09.130
about-- well,
Professor Strang talked

00:01:09.130 --> 00:01:13.620
about how it's easy to spot--
some easy tests to spot when

00:01:13.620 --> 00:01:15.610
a matrix is not invertible.

00:01:15.610 --> 00:01:19.070
Those were: if you have a column
of 0's or a row of 0's, then

00:01:19.070 --> 00:01:20.340
the matrix is not invertible.

00:01:20.340 --> 00:01:22.272
Or if you have two
columns that are the same,

00:01:22.272 --> 00:01:23.980
or two rows that are
the same, the matrix

00:01:23.980 --> 00:01:25.650
is also not invertible.

00:01:25.650 --> 00:01:28.510
So let's see if this
matrix satisfies

00:01:28.510 --> 00:01:31.390
any of those easy conditions.

00:01:31.390 --> 00:01:34.410
Well, here you
have a row of a's.

00:01:34.410 --> 00:01:37.040
If a is equal to 0, you have
a row of 0's, the matrix

00:01:37.040 --> 00:01:38.730
is not invertible.

00:01:38.730 --> 00:01:51.420
So a is-- What
was the other one?

00:01:51.420 --> 00:01:54.800
Two of the same column
or of the same row.

00:01:54.800 --> 00:01:59.789
Well, if a is equal to b, then
all the entries in the matrix

00:01:59.789 --> 00:02:01.580
are the same, so all
the rows are the same,

00:02:01.580 --> 00:02:04.162
all the columns are the same,
the matrix is not invertible.

00:02:04.162 --> 00:02:05.620
So that's the other
easy condition.

00:02:05.620 --> 00:02:12.750
A is not invertible if a is
equal to 0 or a equals to b.

00:02:12.750 --> 00:02:15.120
There's not any other easy
condition that I can spot.

00:02:15.120 --> 00:02:18.310
So we have to do it in
a more systematic way.

00:02:18.310 --> 00:02:22.000
So to do that, actually what
we do is we start with A,

00:02:22.000 --> 00:02:24.750
we try to find its inverse,
and if in the process

00:02:24.750 --> 00:02:28.400
we run into something fishy,
those are our conditions.

00:02:28.400 --> 00:02:32.060
So remember how to find
the inverse of a matrix?

00:02:32.060 --> 00:02:38.760
You start by writing a
giant matrix that has

00:02:38.760 --> 00:02:40.610
A and the identity next to it.

00:02:40.610 --> 00:02:44.860
And then you perform
elimination steps

00:02:44.860 --> 00:02:51.450
until-- let's put dots here,
because there's lots of steps--

00:02:51.450 --> 00:02:54.120
and you stop once you've
reached the identity

00:02:54.120 --> 00:02:56.302
matrix on the left side.

00:02:56.302 --> 00:02:58.510
And when you do that, what
you have on the right side

00:02:58.510 --> 00:03:01.020
will be your A inverse.

00:03:01.020 --> 00:03:01.800
Easy enough.

00:03:01.800 --> 00:03:04.636
So let's do the computations.

00:03:08.130 --> 00:03:20.670
a, b, b; a, a, b; a, a, a; and
my identity matrix next to it.

00:03:20.670 --> 00:03:25.240
And now I know you
learned about elimination

00:03:25.240 --> 00:03:26.400
in the past recitation.

00:03:26.400 --> 00:03:29.010
So I'll do the first
few steps slowly,

00:03:29.010 --> 00:03:33.140
and then I'll just write the
solution for the last steps.

00:03:33.140 --> 00:03:37.290
So we want to eliminate this a.

00:03:37.290 --> 00:03:38.680
We want to turn it into 0.

00:03:38.680 --> 00:03:42.300
So let's subtract the
first row from the second.

00:03:42.300 --> 00:03:45.190
Let's do row 2 minus row 1,
and write it instead of row 2.

00:03:45.190 --> 00:03:52.650
So a, b, b, 1, 0,
0; 0, so a minus a,

00:03:52.650 --> 00:04:01.885
a minus b, b minus b is 0, 0
minus 1, 1 minus 0, 0 minus 0.

00:04:01.885 --> 00:04:04.130
And then we want to do the
same with the third row.

00:04:04.130 --> 00:04:07.430
So let's do row 3 minus row 1.

00:04:07.430 --> 00:04:13.420
0, a minus b, it doesn't fit.

00:04:13.420 --> 00:04:14.470
a minus b.

00:04:14.470 --> 00:04:16.339
Can you understand that?

00:04:16.339 --> 00:04:21.609
And then 0 minus 1,
0 minus 0, 1 minus 0.

00:04:21.609 --> 00:04:22.950
All right.

00:04:22.950 --> 00:04:24.450
Let's continue up here.

00:04:28.520 --> 00:04:31.730
Now I want to eliminate
this a minus b.

00:04:31.730 --> 00:04:33.380
I want to turn it into a 0.

00:04:33.380 --> 00:04:35.932
So the first row stays the same.

00:04:35.932 --> 00:04:46.460
a, b, b, 1, 0, 0; 0, a
minus b, 0, -1, 1, 0.

00:04:46.460 --> 00:04:51.240
And now 0 minus 0, a minus
b minus a minus b is 0.

00:04:51.240 --> 00:04:54.790
a minus b minus 0, that's easy.

00:04:54.790 --> 00:04:58.520
-1 minus -1, 0 again.

00:04:58.520 --> 00:05:02.240
-1 and 1.

00:05:02.240 --> 00:05:03.270
Almost there.

00:05:03.270 --> 00:05:04.584
We already have 0's down here.

00:05:04.584 --> 00:05:06.750
It's looking more and more
like the identity matrix.

00:05:06.750 --> 00:05:10.120
So we have to turn all these
diagonal entries into 1's, so

00:05:10.120 --> 00:05:11.050
let's do that now.

00:05:13.590 --> 00:05:18.290
1, b over a, b over-- oh.

00:05:18.290 --> 00:05:19.850
I'm dividing by a.

00:05:19.850 --> 00:05:21.850
a better not be 0.

00:05:21.850 --> 00:05:23.750
Let me signal that here.

00:05:23.750 --> 00:05:25.825
a different from 0.

00:05:29.900 --> 00:05:33.480
1 over a, 0, 0.

00:05:33.480 --> 00:05:34.440
0.

00:05:34.440 --> 00:05:36.000
I want to make that into a 1.

00:05:36.000 --> 00:05:41.890
So 1 over a minus b, and--
you guessed it-- a minus

00:05:41.890 --> 00:05:43.310
b has to be different from 0.

00:05:47.915 --> 00:05:48.415
0.

00:05:52.170 --> 00:05:56.790
-1 over a minus b.

00:05:56.790 --> 00:05:58.570
1 over a minus b.

00:05:58.570 --> 00:06:05.850
0, 0, 1, 0, -1 over a minus b.

00:06:05.850 --> 00:06:07.262
1 over a minus b.

00:06:09.860 --> 00:06:11.484
AUDIENCE: [INAUDIBLE]

00:06:11.484 --> 00:06:13.150
PROFESSOR: Someone's
pointing a mistake.

00:06:13.150 --> 00:06:16.430
AUDIENCE: Yeah, the (2, 2)
entry, you should have a 1.

00:06:16.430 --> 00:06:18.250
PROFESSOR: Oh, you're
absolutely right.

00:06:22.320 --> 00:06:22.820
Thank you.

00:06:25.760 --> 00:06:27.230
All right.

00:06:27.230 --> 00:06:29.580
So not much left to do.

00:06:29.580 --> 00:06:31.870
We only have to eliminate
b over a and b over a.

00:06:31.870 --> 00:06:35.460
That's a little bit
computationally heavy.

00:06:35.460 --> 00:06:39.260
So what do you have to do?

00:06:39.260 --> 00:06:41.450
You just have to
basically subtract--

00:06:41.450 --> 00:06:47.670
replace row 1 by row 1 minus b
over a times row 2 plus row 3.

00:06:47.670 --> 00:06:50.090
Let me write that down.

00:06:50.090 --> 00:06:59.130
row 1 minus b over a
times row 2 plus row 3.

00:06:59.130 --> 00:07:02.540
You want that to
go into your row 1.

00:07:02.540 --> 00:07:03.470
These stay the same.

00:07:07.970 --> 00:07:09.870
-1 over a minus b.

00:07:09.870 --> 00:07:11.590
1 over a minus b.

00:07:21.090 --> 00:07:23.710
This is where I'm
going to go to my notes

00:07:23.710 --> 00:07:30.310
and tell you that these
numbers are 1 over a minus

00:07:30.310 --> 00:07:37.190
b, 0 minus b over a, a minus b.

00:07:37.190 --> 00:07:39.210
That looks pretty awful.

00:07:39.210 --> 00:07:41.410
But that is A inverse.

00:07:41.410 --> 00:07:43.300
And we're done.

00:07:43.300 --> 00:07:45.290
Except it looks terrible.

00:07:45.290 --> 00:07:48.260
So let me just write
it in a nicer way.

00:07:48.260 --> 00:07:50.860
A inverse equals-- see,
you're dividing everything

00:07:50.860 --> 00:07:52.960
by a minus b, so
pull out that factor.

00:07:57.680 --> 00:08:06.120
1, 0, minus b over a;
-1, 1, 0; 0, -1, 1.

00:08:06.120 --> 00:08:07.710
Looks much better now.

00:08:07.710 --> 00:08:09.870
So here's your A inverse.

00:08:09.870 --> 00:08:13.720
And the other question was: what
are the conditions on a and b

00:08:13.720 --> 00:08:16.480
for the matrix A
to be invertible?

00:08:16.480 --> 00:08:20.695
Well, a has to be non-zero, and
a has to be different from b.

00:08:20.695 --> 00:08:21.320
And we're done.

00:08:21.320 --> 00:08:22.370
That's it for today.

00:08:22.370 --> 00:08:24.360
See you next time.