WEBVTT

00:00:07.210 --> 00:00:07.680
Hi.

00:00:07.715 --> 00:00:09.220
Welcome back to recitation.

00:00:09.255 --> 00:00:11.780
In class, you've been learning
about convergence tests for

00:00:11.815 --> 00:00:12.800
infinite series.

00:00:12.835 --> 00:00:17.050
I have here three examples of
series that I happen to like.

00:00:17.085 --> 00:00:21.510
So the first one is the sum from
n equals 0 to infinity of

00:00:21.545 --> 00:00:25.160
5n plus 2 divided by
n cubed plus 1.

00:00:25.195 --> 00:00:28.550
The second one is the sum from
n equals 1 to infinity of the

00:00:28.585 --> 00:00:32.590
quantity 1 plus the square root
of 5 over 2, all that to

00:00:32.625 --> 00:00:33.630
the nth power.

00:00:33.665 --> 00:00:36.590
And the third one is the sum
from n equals 1 to infinity of

00:00:36.625 --> 00:00:39.290
the natural log of n divided
by n squared.

00:00:39.325 --> 00:00:42.500
So what I'd like you to do for
each of these three series is

00:00:42.535 --> 00:00:44.290
to figure out whether or
not they converge.

00:00:44.325 --> 00:00:46.580
So whether they converge,
or whether they diverge.

00:00:46.615 --> 00:00:48.620
So why don't you pause the
video, take some time to work

00:00:48.655 --> 00:00:50.800
that out, come back, and we
can work on it together.

00:00:59.920 --> 00:01:00.640
Welcome back.

00:01:00.675 --> 00:01:03.130
Hopefully you had some luck
working on these series.

00:01:03.165 --> 00:01:04.920
So let's talk about
how we do them.

00:01:04.955 --> 00:01:06.970
So I'll start with the
first one here.

00:01:07.005 --> 00:01:10.232
So this first one, the function
that we're, that

00:01:10.267 --> 00:01:16.900
we're summing is this 5n plus
2 divided by n cubed plus 3.

00:01:16.935 --> 00:01:21.450
And so you want to know, does
that sum of that expression,

00:01:21.485 --> 00:01:24.410
of n goes from 1 to infinity,
does that converge?

00:01:24.445 --> 00:01:25.900
Does it reach some
finite value?

00:01:25.935 --> 00:01:27.380
Or does it diverge?

00:01:27.414 --> 00:01:31.160
Does it either, you know,
oscillate, or go to infinity,

00:01:31.195 --> 00:01:32.720
or something like that?

00:01:32.755 --> 00:01:37.580
So for one thing you can always
do, is something like a

00:01:37.615 --> 00:01:39.710
comparison test. And so what you
want to do there, is you

00:01:39.745 --> 00:01:42.009
want to extract the information
from the

00:01:42.045 --> 00:01:43.800
expression for the summand.

00:01:43.835 --> 00:01:46.700
You want to figure out, you
know, about how big this is.

00:01:46.735 --> 00:01:48.539
What are the important
parts of it?

00:01:48.575 --> 00:01:50.250
So in this case, you
have a ratio.

00:01:50.285 --> 00:01:51.850
And so when you have a ratio,
what you want to do, is you

00:01:51.884 --> 00:01:54.820
want to look at what's the
magnitude of the top, and

00:01:54.854 --> 00:01:56.410
what's the magnitude
of the bottom.

00:01:56.445 --> 00:01:59.539
So roughly speaking, the
magnitude of the top is of the

00:01:59.575 --> 00:02:01.000
order of n.

00:02:01.035 --> 00:02:01.290
Right?

00:02:01.325 --> 00:02:02.800
It's 5n.

00:02:02.835 --> 00:02:04.750
The 5 is just a constant
multiple.

00:02:04.785 --> 00:02:06.220
That's not going to make
a whole big difference.

00:02:06.255 --> 00:02:09.190
And the plus 2 is just much,
much, much smaller than the n

00:02:09.225 --> 00:02:11.080
when n gets very, very large.

00:02:11.115 --> 00:02:14.680
So the, we can say that the
order of magnitude of the top

00:02:14.715 --> 00:02:18.070
here is of the order of n, and
similarly, the order of

00:02:18.105 --> 00:02:21.000
magnitude of the bottom is
of the order of n cubed.

00:02:21.035 --> 00:02:21.240
Right?

00:02:21.275 --> 00:02:24.210
The plus 1 is much, much, much
smaller than the n cubed, so

00:02:24.245 --> 00:02:28.000
it's not, it's almost
insignificant.

00:02:28.035 --> 00:02:33.579
So roughly speaking, we should
expect this to be of the same

00:02:33.615 --> 00:02:36.760
behavior as if we just
had n over n cubed,

00:02:36.795 --> 00:02:38.410
which is 1 over n squared.

00:02:38.445 --> 00:02:41.850
And we know that 1 over n
squared, the series 1 over n

00:02:41.885 --> 00:02:45.000
squared from n equals 1 to
infinity, converges.

00:02:45.035 --> 00:02:49.340
So that's sort of just
a sort of a way of

00:02:49.375 --> 00:02:50.079
thinking about this.

00:02:50.115 --> 00:02:53.060
And we can formalize it using
the limit comparison test.

00:02:53.095 --> 00:03:03.560
So the idea here for part A is
to use the limit comparison

00:03:03.595 --> 00:03:06.070
test. So what we want
to compute--

00:03:06.105 --> 00:03:08.820
so I'm going to put a 5 in.

00:03:08.855 --> 00:03:11.190
So I'm going to make it 5 over
n squared, because we

00:03:11.225 --> 00:03:12.940
had this 5 up here.

00:03:12.975 --> 00:03:21.079
So we have an is equal to 5n
plus 2 over n cubed plus 1.

00:03:21.115 --> 00:03:24.600
And I want to limit
compare it with--

00:03:24.635 --> 00:03:35.350
I'm going to compare my series
with the series sum from n

00:03:35.385 --> 00:03:39.460
equals 1 to infinity of
5 over n squared.

00:03:39.495 --> 00:03:41.180
Now--

00:03:41.215 --> 00:03:42.970
I'm talking about
this 5, right?

00:03:43.005 --> 00:03:44.190
This 5 doesn't matter.

00:03:44.225 --> 00:03:48.579
If this series-- if 1 over n
squared, that sum converges

00:03:48.615 --> 00:03:52.350
then 5 over n squared, that
sum converges to exactly 5

00:03:52.385 --> 00:03:53.130
times as much.

00:03:53.165 --> 00:03:56.210
And if, you know, if I wrote a
divergent thing here, and then

00:03:56.245 --> 00:03:58.530
multiplied it by 5, the
result multiplied by

00:03:58.565 --> 00:03:59.730
5 would still diverge.

00:03:59.765 --> 00:04:01.630
So this constant multiple
isn't going to matter.

00:04:01.665 --> 00:04:04.150
And I'm just choosing, I'm just
putting the 5 in there

00:04:04.185 --> 00:04:06.890
because of this 5 we
saw over here.

00:04:06.925 --> 00:04:07.590
So OK.

00:04:07.625 --> 00:04:09.700
So let's select work
it out, then.

00:04:09.735 --> 00:04:12.810
So the limit comparison
test says--

00:04:12.845 --> 00:04:18.240
so we look at the limit as n
goes to infinity of the ratio

00:04:18.274 --> 00:04:20.250
of the two things that
we're interested in.

00:04:20.285 --> 00:04:25.420
So in this case, this is 5n and
plus 2 over n cubed plus

00:04:25.455 --> 00:04:30.950
1, divided by 5 over
n squared.

00:04:30.985 --> 00:04:31.500
And you can--

00:04:31.535 --> 00:04:31.680
OK.

00:04:31.715 --> 00:04:35.659
So this is a ratio of two
ratios, so we can rewrite it

00:04:35.695 --> 00:04:39.200
by multiplying upstairs, and we
get that this is equal to

00:04:39.235 --> 00:04:48.670
the limit as n goes to infinity
of 5n cubed plus 2 n

00:04:48.705 --> 00:04:53.620
squared over 5n cubed plus 5.

00:04:53.655 --> 00:04:56.800
And so we've seen before, when
we were dealing with limits,

00:04:56.835 --> 00:05:00.610
that when you have a ratio of
two polynomials, as the

00:05:00.645 --> 00:05:04.560
variable goes to infinity, the
limit is what you get just by

00:05:04.595 --> 00:05:08.340
comparing the leading terms. So
in this case, we just have

00:05:08.375 --> 00:05:12.570
to look at 5n cubed over 5n
cubed, and that's indeed 1.

00:05:12.605 --> 00:05:13.700
OK, so this is 1.

00:05:13.735 --> 00:05:19.050
So by the limit comparison test,
our series, the sum of

00:05:19.085 --> 00:05:23.230
this an, converges if and only
if this series converges.

00:05:23.265 --> 00:05:25.340
And we already said
that we know sum 1

00:05:25.375 --> 00:05:27.020
over n squared converges.

00:05:27.055 --> 00:05:29.550
So by the limit comparison
test, since sum 1 over n

00:05:29.585 --> 00:05:31.320
squared converges, our
series converges.

00:05:36.630 --> 00:05:37.140
Great.

00:05:37.175 --> 00:05:37.550
OK.

00:05:37.585 --> 00:05:39.870
So that's the first one.

00:05:39.905 --> 00:05:41.930
Let's look at the second one.

00:05:41.965 --> 00:05:45.450
So the second one here is the
sum from n equals 1 to

00:05:45.485 --> 00:05:49.520
infinity of 1 plus the square
root of 5 over 2.

00:05:49.555 --> 00:05:51.720
That whole thing to
the nth power.

00:05:51.755 --> 00:05:54.100
Now, if you look at this, the
thing you should recognize

00:05:54.135 --> 00:05:57.450
this as, is just a particular
geometric series.

00:05:57.485 --> 00:05:57.860
Right?

00:05:57.895 --> 00:06:01.090
This is, if you were to replace
1 plus the square root

00:06:01.125 --> 00:06:04.842
of 5 over 2 with x, this is just
x plus x squared plus x

00:06:04.877 --> 00:06:06.890
cubed plus dot dot
dot dot dot.

00:06:06.925 --> 00:06:10.800
It's just a geometric series
with constant ratio--

00:06:10.835 --> 00:06:11.620
well, x .

00:06:11.655 --> 00:06:13.760
1 plus the square root
of 5 over 2.

00:06:13.795 --> 00:06:18.510
So we know exactly when
geometric series converge.

00:06:18.545 --> 00:06:24.400
They converge exactly when the
constant ratio is between

00:06:24.435 --> 00:06:25.630
minus 1 and 1.

00:06:25.665 --> 00:06:28.250
So bigger than minus
1 and less than 1.

00:06:28.285 --> 00:06:31.570
So this series converges, then,
if and only if 1 plus

00:06:31.605 --> 00:06:34.520
the square root of 5 over 2
is between minus 1 and 1.

00:06:34.555 --> 00:06:37.300
So then we just have to think
about, is this number between

00:06:37.335 --> 00:06:38.980
minus 1 and 1.

00:06:39.015 --> 00:06:39.970
And OK.

00:06:40.005 --> 00:06:41.400
So this is not that
hard to do.

00:06:41.435 --> 00:06:44.520
Square root of 5 is bigger than
2, so 1 plus the square

00:06:44.555 --> 00:06:47.100
root of 5 is bigger than 3, so
1 plus the square root of 5

00:06:47.135 --> 00:06:49.460
over 2 is bigger than 3/2.

00:06:49.495 --> 00:06:51.140
So 3/2 is bigger than 1.

00:06:51.175 --> 00:06:56.590
So this common ratio in this
series is bigger than 1.

00:06:56.625 --> 00:06:59.630
So the terms of this series
are blowing ip.

00:06:59.665 --> 00:07:02.070
You know, when you, when n gets
bigger and bigger, you're

00:07:02.105 --> 00:07:03.700
adding larger and larger
numbers here.

00:07:03.735 --> 00:07:04.540
This is blowing up.

00:07:04.575 --> 00:07:06.790
It's a divergent geometric
series.

00:07:06.825 --> 00:07:08.175
So this series does
not converge.

00:07:08.210 --> 00:07:14.570
So b, I'll just write
that here.

00:07:14.605 --> 00:07:27.430
B diverges, because it's
geometric with common ratio

00:07:27.465 --> 00:07:28.480
bigger than 1.

00:07:28.515 --> 00:07:31.890
So that's the reason that
part B diverges.

00:07:31.925 --> 00:07:34.250
Finally, we're left
with question C.

00:07:34.284 --> 00:07:38.330
So I'm going to come over and
write it over here again, so

00:07:38.365 --> 00:07:38.850
we can see it.

00:07:38.885 --> 00:07:49.390
So part C asks the sum for n
equals 1 to infinity of log n

00:07:49.425 --> 00:07:50.640
over n squared.

00:07:53.360 --> 00:07:55.270
So this one's a little trickier,
and it requires a

00:07:55.305 --> 00:07:55.906
little bit more thought.

00:07:55.941 --> 00:07:57.156
The thing to-- do let's
start just by--

00:08:02.490 --> 00:08:06.440
it can't be solved just by a
straightforward application of

00:08:06.475 --> 00:08:10.740
the limit comparison test
that we've learned.

00:08:10.775 --> 00:08:13.870
So we need to think a little bit
more about what ways could

00:08:13.905 --> 00:08:15.690
we solve it.

00:08:15.725 --> 00:08:19.400
So one thing to remember here
is, is we should think about,

00:08:19.435 --> 00:08:21.450
what are the magnitudes
of these things?

00:08:21.485 --> 00:08:24.530
So we know that sum 1 over
n squared converges.

00:08:24.565 --> 00:08:27.640
But log n is a function
that grows.

00:08:27.675 --> 00:08:31.410
So this individual term is
bigger than 1 over n squared.

00:08:31.445 --> 00:08:35.620
So we can't just compare
it to 1 over n squared.

00:08:35.655 --> 00:08:38.010
But log n grows very,
very slowly.

00:08:38.044 --> 00:08:38.880
How slowly?

00:08:38.914 --> 00:08:41.909
Well, it grows more slowly
than any power of x.

00:08:41.945 --> 00:08:43.440
Right?

00:08:43.475 --> 00:08:45.200
So, or-- sorry-- any power of
n in this case, because the

00:08:45.235 --> 00:08:46.720
variable is n.

00:08:46.755 --> 00:08:57.750
So if you remember, we know,
we've shown, that the limit of

00:08:57.785 --> 00:09:10.010
ln x over x to the p as x goes
to infinity is equal to 0 for

00:09:10.045 --> 00:09:13.440
any positive p.

00:09:13.475 --> 00:09:19.420
So ln, log n, ln n, is going to
infinity, but it's going to

00:09:19.455 --> 00:09:24.540
infinity much slower than
any power of n.

00:09:24.575 --> 00:09:25.390
In this case.

00:09:25.425 --> 00:09:27.660
Or x, down here.

00:09:27.695 --> 00:09:28.700
So OK.

00:09:28.735 --> 00:09:29.860
So what does that mean?

00:09:29.895 --> 00:09:32.070
Well, one thing you could do,
is you could say, oh, OK.

00:09:32.105 --> 00:09:35.980
So ln n over n, that's
getting really small.

00:09:36.015 --> 00:09:38.260
And then what we're left
with is 1 over n.

00:09:38.295 --> 00:09:39.050
So you can say, OK.

00:09:39.085 --> 00:09:42.030
So this is much smaller
than 1 over n.

00:09:42.065 --> 00:09:45.480
The problem is that the
sum 1 over n diverges.

00:09:45.515 --> 00:09:45.830
Yeah?

00:09:45.865 --> 00:09:47.630
So that doesn't help
us, really, right?

00:09:47.665 --> 00:09:52.620
So we've shown this is bigger
than the sum 1 over n squared,

00:09:52.655 --> 00:09:56.430
which converges, and it's
smaller than the sum 1 over n,

00:09:56.465 --> 00:09:57.570
which diverges.

00:09:57.605 --> 00:09:59.160
But that still doesn't
tell us, you know?

00:09:59.195 --> 00:10:02.590
It could be something that,
you know, there's a lot of

00:10:02.625 --> 00:10:06.390
room bigger than a particular
convergent series, and smaller

00:10:06.425 --> 00:10:08.130
than a particular divergent
series.

00:10:08.165 --> 00:10:10.800
And in particular, there are
both convergent and divergent

00:10:10.835 --> 00:10:11.950
series in between.

00:10:11.985 --> 00:10:14.450
So we still need, we need
either something that

00:10:14.485 --> 00:10:18.220
converges that our thing is
less than, or we need

00:10:18.255 --> 00:10:21.560
something that diverges that
our thing is bigger than.

00:10:21.595 --> 00:10:21.760
Right?

00:10:21.795 --> 00:10:24.960
If we can bound our series above
by something convergent,

00:10:24.995 --> 00:10:26.470
then our series converges.

00:10:26.505 --> 00:10:27.250
Because [UNINTELLIGIBLE]

00:10:27.285 --> 00:10:29.330
has positive terms.
This is important.

00:10:29.365 --> 00:10:32.390
Or if we can bound it below by
something that diverges, then

00:10:32.425 --> 00:10:33.830
we would know it diverges.

00:10:33.865 --> 00:10:37.090
And so far, we haven't
been able to do that.

00:10:37.125 --> 00:10:41.955
But maybe we can think of a-- so
I said we could write this,

00:10:41.990 --> 00:10:46.490
a second ago, I said we could
write ln n over n squared is

00:10:46.525 --> 00:10:51.420
equal to ln n and over
n times 1 over n.

00:10:51.455 --> 00:10:55.280
So that was what we just said
a minute ago, at which we

00:10:55.315 --> 00:11:00.220
showed is eventually
less than 1 over n.

00:11:00.255 --> 00:11:03.150
So this is true, but it wasn't
useful, because the sum 1 over

00:11:03.185 --> 00:11:04.830
n diverges.

00:11:04.865 --> 00:11:07.070
But maybe we can, we can
do something even

00:11:07.105 --> 00:11:08.800
a little more tricky.

00:11:08.835 --> 00:11:13.540
Because here we saw that x, we
could use it in the base, we

00:11:13.575 --> 00:11:17.350
could use any power of the
variable, any positive power.

00:11:17.385 --> 00:11:20.070
So here, we tried it with
the power n to the 1.

00:11:20.105 --> 00:11:23.260
We tried to split this 2 as 1
plus 1, and we kept one of the

00:11:23.295 --> 00:11:25.940
n's to knock out the log
n, and we kept the

00:11:25.975 --> 00:11:27.620
other n over here.

00:11:27.655 --> 00:11:30.060
But we don't need a whole
power of n to

00:11:30.095 --> 00:11:31.210
knock out the log n.

00:11:31.245 --> 00:11:35.180
Any positive power
of n would do.

00:11:35.215 --> 00:11:38.760
So in particular, we could split
this using, say, just a

00:11:38.795 --> 00:11:42.550
small power of n, even smaller
than the first power here, and

00:11:42.585 --> 00:11:44.050
leave more of it over here.

00:11:44.085 --> 00:11:54.260
So we also know, we also have
that ln n over n squared is

00:11:54.295 --> 00:11:59.180
equal to, for example,
ln n over--

00:11:59.215 --> 00:12:05.230
well, you know, for example,
n to the 1/2 times 1

00:12:05.265 --> 00:12:09.180
over n to the 3/2.

00:12:09.215 --> 00:12:14.980
And now we know that ln n over
n to the 1/2 goes to 0 as n

00:12:15.015 --> 00:12:15.780
gets large.

00:12:15.815 --> 00:12:17.610
So this is thing is
getting smaller

00:12:17.645 --> 00:12:18.650
and smaller and smaller.

00:12:18.685 --> 00:12:23.520
So as this gets smaller, we have
that this has to be less

00:12:23.555 --> 00:12:27.130
than 1 over n to the 3/2.

00:12:27.165 --> 00:12:31.240
So this thing that we're adding
up here is smaller than

00:12:31.275 --> 00:12:33.150
1 over n to the 3/2.

00:12:33.185 --> 00:12:34.550
Well, what's the significance
of that?

00:12:34.585 --> 00:12:38.600
Well, we know 1 over n to the
3/2, that series converges.

00:12:38.635 --> 00:12:39.330
Yeah?

00:12:39.365 --> 00:12:46.990
We know the sum 1 over n to
the 3/2 from n equals 1 to

00:12:47.025 --> 00:12:54.050
infinity converges, because
3/2 is bigger than 1.

00:12:54.085 --> 00:12:54.620
OK?

00:12:54.655 --> 00:12:55.370
So what does that mean?

00:12:55.405 --> 00:12:59.130
Well, we have our series, and
we've shown that the terms of

00:12:59.165 --> 00:13:02.660
our series are eventually
bounded below 1

00:13:02.695 --> 00:13:04.290
over n to the 3/2.

00:13:04.325 --> 00:13:08.150
And we know that the sum of 1
over n to the 3/2 converges,

00:13:08.185 --> 00:13:12.000
so our series is bounded above
by a convergent series.

00:13:12.035 --> 00:13:14.510
So whenever you have a series
bounded above, a series of

00:13:14.545 --> 00:13:17.850
non-negative terms bounded
above, by a convergent series,

00:13:17.885 --> 00:13:21.460
that means your series
also has to converge.

00:13:21.495 --> 00:13:24.388
OK, so this converges.

00:13:24.423 --> 00:13:27.230
So it converges by a comparison

00:13:27.265 --> 00:13:28.340
to this other series.

00:13:28.375 --> 00:13:32.980
Using this cute trick that we
can replace a log with any

00:13:33.015 --> 00:13:36.840
small positive power of n that
happens to be convenient.

00:13:36.875 --> 00:13:39.290
And of course, if you wanted to,
maybe you could have made

00:13:39.325 --> 00:13:42.540
this 1/2 a 1/10 and that would
have been fine, or you

00:13:42.575 --> 00:13:45.720
could've even made it a 9/10,
and then you would here be

00:13:45.755 --> 00:13:49.280
left with n to the 11/10, and
that would still be OK.

00:13:49.315 --> 00:13:53.030
Because that would be 11/10, and
11/10, then it would still

00:13:53.065 --> 00:13:54.540
be bigger than 1.

00:13:54.575 --> 00:13:55.900
All right.

00:13:55.935 --> 00:13:59.810
So this is a nice use of a, of
a comparison test. We didn't

00:13:59.845 --> 00:14:02.300
use exactly the limit comparison
test as it was

00:14:02.335 --> 00:14:06.220
described in lecture, but it's a
very closely related process

00:14:06.255 --> 00:14:10.020
that we went through to show
that this second series--

00:14:10.055 --> 00:14:11.540
sorry, this third series--

00:14:11.575 --> 00:14:13.480
also converges.

00:14:13.515 --> 00:14:15.330
So I'll end there.