Ratio test and Root test

Videos

Review Videos

Videos, Math Dr. Bob

Link to original

01 Theory

Theory 1

Ratio Test (RaT)

Applicability: Any series with nonzero terms.

Test Statement:

Suppose that |an+1an|⟶L as n→∞.

Then:

L<1:∑n=1∞anconverges absolutelyL>1:∑n=1∞andivergesL=1or DNE:test inconclusive

Extra - Ratio Test - a deeper look

To understand the ratio test better, first consider this series:

∑n=0∞2nn!=1+21!+222!+233!+⋯
  • The term 233! is created by multiplying the prior term by 23.
  • The term 244! is created by multiplying the prior term by 24.
  • The term an is created by multiplying the prior term by 2n.

When n>3, the multiplication factor giving the next term is necessarily less than 23. Therefore, when n>3, the terms shrink faster than those of a geometric series having r=23. Therefore this series converges.

Similarly, consider this series:

∑n=0∞10nn!=1+101!+1022!+1033!+⋯

Write Rn=anan−1 for the ratio from the prior term an−1 to the current term an. For this series, Rn=10n.

This ratio falls below 1011 when n>11, after which the terms necessarily shrink faster than those of a geometric series with r=1011. Therefore this series converges.

The main point of the discussion can be stated like this:

Rn→L<1asn→∞

Whenever this is the case, then eventually the ratios are bounded below some r<1, and the series terms are smaller than those of a converging geometric series.

Extra - Ratio Test proof

Let us write Rn=|an+1an| for the ratio to the next term from term n.

Suppose that Rn→L as n→∞, and that L<1. This means: eventually the ratio of terms is close to L; so eventually it is less than 1.

More specifically, let us define r=L+12. This is the point halfway between L and 1. Since Rn→L, we know that eventually Rn<r.

Any geometric series with ratio r converges. Set c=aN for N big enough that RN<r. Then the terms of our series satisfy |aN+n|≤crn, and the series starting from aN is absolutely convergent by comparison to this geometric series.

(Note that the terms a1,…,aN−1 do not affect convergence.)

Link to original

02 Illustration

Example - Ratio test

Ratio test examples

(a) Observe that ∑n=0∞10nn! has ratio Rn=10n+1 and thus Rn→0=L<1. Therefore the RaT implies that this series converges.

Simplify the ratio:

(10n+1(n+1)!)(10nn!)≫≫10n+1(n+1)!⋅n!10n≫≫10⋅10n(n+1)n!⋅n!10n≫≫10n+1⟶n→∞0

Notice this technique! We frequently use these rules:

10n+1=10n⋅10,(n+1)!=(n+1)n!

(To simplify ratios with exponents and factorials.)


(b) ∑n=1∞n22n has ratio Rn=(n+1)22n+1/n22n.

Simplify this:

(n+1)22n+1/n22n≫≫(n+1)22n+1⋅2nn2 ≫≫(n+1)2⋅2nn2⋅2⋅2n≫≫n2+2n+12n2⟶n→∞12=L

So the series converges absolutely by the ratio test.


(c) Observe that ∑n=1∞n2 has ratio Rn=n2+2n+1n2→1 as n→∞.

So the ratio test is inconclusive, even though this series fails the SDT and obviously diverges.


(d) Observe that ∑n=1∞1n2 has ratio Rn=n2n2+2n+1→1 as n→∞.

So the ratio test is inconclusive, even though the series converges as a p-series with p=2>1.


(e) More generally, the ratio test is usually inconclusive for rational functions; it is more effective to use LCT with a p-series.

Link to original

03 Theory

Theory 2

Root Test (RooT)

Applicability: Any series.

Test Statement:

Suppose that |an|1/n⟶L as n→∞.

Then:

L<1:∑n=1∞anconverges absolutelyL>1:∑n=1∞andivergesL=1or DNE:test inconclusive

Extra - Root test: explanation

The fact that |an|1/n→L and L<1 implies that eventually |an|1/n<L+12 for all high enough n. Set r=L+12 (the midpoint between L and 1).

Now, the equation |an|1/n<r is equivalent to the equation |an|<rn.

Therefore, eventually the terms |an| are each less than the corresponding terms of this convergent geometric series:

∑n=1∞rn=1+r+r2+r3+⋯
Link to original

04 Illustration

Example - Root test examples

Root test examples

(a) Observe that ∑n=1∞(1n)n has roots of terms:

|an|1/n=((1n)n)1/n=1n⟶n→∞0=L

Because L<1, the RooT shows that the series converges absolutely.


(b) Observe that ∑n=1∞(−1)n(n2n+1)n has roots of terms:

|an|n=n2n+1⟶n→∞12=L

Because L<1, the RooT shows that the series converges absolutely.

Link to original

Example - Ratio test versus root test

Ratio test versus root test

Determine whether the series ∑n=1∞n24n5n+2 converges absolutely or conditionally or diverges.

Solution Before proceeding, rewrite somewhat the general term as (n5)2⋅(45)n.

Now we solve the problem first using the ratio test. By plugging in n+1 we see that

an+1=(n+15)2⋅(45)n+1

So for the ratio Rn we have:

an+1⋅1an=(n+15)2(45)n+1⋅(5n)2(54)n≫≫n2+2n+1n2⋅45⟶45<1 as n→∞

Therefore the series converges absolutely by the ratio test.

Now solve the problem again using the root test. We have for |an|n:

((n5)2(45)n)1/n=(n5)2/n45

To compute the limit as n→∞ we must use logarithmic limits and L’Hopital’s Rule. So, first take the log:

ln⁡((n5)2/n45)=2nln⁡n5+ln⁡45

Then for the first term apply L’Hopital’s Rule:

ln⁡n5⟶d/dx1n/515n/2⟶d/dx1/2≫≫1/n1/2≫≫2n⟶0 as n→∞

So the first term goes to zero, and the second (constant) term is the value of the limit. So the log limit is ln⁡45, and the limit (before taking logs) must be eln⁡45 (inverting the log using ex) and this is 45. Since 45<1, the root test also shows that the series converges absolutely.

Link to original

Series tests: strategy tips

Videos

Review Videos

Videos, Math Dr. Bob

Videos, Trefor Bazett

Link to original

05 Theory

Theory 1

It can help to associate certain “strategy tips” to find convergence tests based on certain patterns.

Matching powers → Simple Divergence Test

∑n=1∞n−12n+1

Use the SDT because we see the highest power is the same (=1) in numerator and denominator.

Rational or Algebraic → Limit Comparison Test

∑n=1∞n3+13n3+4n2+2

Use the LCT because we have a rational or algebraic function (positive terms).

Not rational, not factorials → Integral Test

∑n=1∞ne−n2

Use the IT because we do not have a rational/algebraic function, and we do not see factorials.

Rational, alternating → AST, and LCT or DCT

∑n=1∞(−1)nn2n4+1

Notice large n behavior is like n2/n4 or 1/n2. This converges. Use the LCT to show absolute convergence. Skip the AST because absolute convergence settles the matter. Lesson: check for absolute BEFORE applying AST, even when alternating!

Factorials → Ratio Test

∑n=1∞2nn!

Use the RaT because we see a factorial. (In case of alternating + factorial, use RaT first.)

Recognize geometric → LCT or DCT

∑n=1∞12+3n

Use the LCT or DCT comparing to 13n because we see similarity to 13n (recognize geometric).

Link to original

Power series: Radius and Interval

Videos

Review Videos

Videos, Math Dr. Bob

Link to original

06 Theory

Theory 1

A power series looks like this:

f(x)=a0+a1x+a2x2+a3x3+⋯

Power series are used to build and study functions. They allow a uniform “modeling framework” in which many functions can be described and compared. Power series are also convenient for computers because they provide a way to store and evaluate differentiable functions with numerical (approximate) values.

Small x needed for power series

The most important fact about power series is that they work for small values of x.

Many power series diverge for |x| too big; but even when they converge, for big |x| they converge more slowly, and partial sum approximations are less accurate.

The idea of a power series is a modification of the idea of a geometric series in which the common ratio r becomes a variable x, and each term has an additional coefficient parameter an controlling the relative contribution of different orders.

Link to original

07 Theory

Theory 2

Every power series has a radius of convergence and an interval of convergence.

Radius of convergence

Consider a power series centered at x=0:

f(x)=a0+a1x+a2x2+a3x3+⋯

Apply the ratio test:

limn→∞⁡|an+1xn+1anxn|≫≫(limn→∞⁡|an+1an|)|x|=L

Define the radius of convergence R∈[0,∞]:

R=1limn→∞⁡|an+1an|

Therefore:

|x|<R⟹L<1⟹converges|x|>R⟹L>1⟹diverges

We can build shifted power series for x near some other value c. Just replace the variable x with a shifted variable u=x−c:

a0+a1u+a2u2+a3u3+⋯≫≫a0+a1(x−c)+a2(x−c)2+a3(x−c)3+⋯

Now apply the ratio test to determine convergence:

limn→∞⁡|an+1|x−c|n+1an|x−c|n|≫≫(limn→∞⁡|an+1an|)|x−c|

Define the radius of convergence R∈[0,∞]:

R=1limn→∞⁡|an+1an|

In the shifted setting, the radius of convergence limits the *distance from *:

|x−c|<R⟹converges|x−c|>R⟹diverges

Method: To calculate the interval of convergence of a power series, follow these steps:

  • Observe the center c of the shifted series (or c=0 for no shift).
  • Compute R using the limit of coefficient ratios.
  • Write down the preliminary interval (c−R,c+R).
  • Plug each endpoint, c−R and c−R, into the original series
    • Check for convergence
  • Add in the convergent endpoints. (4 possible scenarios.)
Link to original

08 Illustration

Example - Radius of convergence

Radius of convergence

Find the radius of convergence of the series:

(a) ∑n=0∞xn2n (b) ∑n=0∞x2n(2n)!

Solution

(a) Ratio of terms:

|an+1an|=|xn+12n+1xn2n|≫≫|xn+12n+1⋅2nxn|≫≫12|x|=L

This converges for L<1, or |x|<2. Therefore R=2.


(b) This power series skips the odd powers. Apply the ratio test to just the even powers:

|an+1an|=|x2n+2(2n+2)!x2n(2n)!|≫≫|x2n+2(2n+2)!⋅(2n)!x2n|≫≫1(2n+2)(2n+1)|x2|≫≫R=∞ Link to original

Example - Interval of convergence

Interval of convergence

Find the radius and interval of convergence of the following series:

(a) ∑n=1∞(x−3)nn (b) ∑n=0∞(−3)nxnn+1

Solution

(a) ∑n=1∞(x−3)nn

(1) Apply ratio test:

|an+1⋅an−1|=|(x−3)n+1n+1⋅n(x−3)n|≫≫nn+1|x−3|⟶n→∞|x−3|=L

Therefore R=1 and thus:

|x−3|<1⟹converges|x−3|>1⟹diverges

Preliminary interval: x∈(2,4).


(2) Check endpoints:

Check endpoint x=2:

∑n=1∞(2−3)nn≫≫∑n=1∞(−1)nn≫≫converges by AST

Check endpoint x=4:

∑n=1∞(4−3)nn≫≫∑n=1∞1n≫≫diverges as p-series

Final interval of convergence: x∈[2,4)


(b) ∑n=0∞(−3)nxnn+1

(1) Apply ratio test:

|an+1⋅an−1|=|(−3)n+1xn+1n+2⋅n+1(−3)nxn|≫≫|(−3)(−3)n|n+2⋅n+1|(−3)n||x|≫≫3n+1n+2|x|⟶n→∞3|x|=L

Therefore:

|x|<13⟹converges|x|>13⟹diverges

Preliminary interval: x∈(−13,13)


(2) Check endpoints:

Check endpoint x=−1/3:

∑n=0∞(−3⋅(−13))nn+1≫≫∑n=0∞1n+1≫≫diverges by LCT with bn=1/n

Check endpoint x=+1/3:

∑n=0∞(−3⋅(+13))nn+1≫≫∑n=0∞(−1)nn+1≫≫converges by AST

Final interval of convergence: x∈(−1/3,1/3]

Link to original

Exercise - Radius and interval

Radius and interval for a few series

Find the radius and interval of convergence of the following series:

(a) ∑n=0∞xn (b) ∑n=0∞n!xn

Link to original

Interval of convergence - further examples

Interval of convergence - further examples

Find the interval of convergence of the following series.

(a) ∑n=0∞n(x+2)n3n+1 (b) ∑n=1∞(4x+1)nn

Solution

(a) ∑n=0∞n(x+2)n3n+1

Ratio of terms:

n+13n+2⋅3n+1n|x+2|≫≫n+13n|x+2|⟶n→∞13|x+2|=L

This converges when L<1 or |x−(−2)|<3.

Therefore R=3 and the preliminary interval is x∈(−5,1).

Check endpoints: ∑n(−3)n3n+1 diverges and ∑n(3)n3n+1 also diverges.

Final interval is (−5,1).


(b) ∑n=1∞(4x+1)nn

Ratio of terms:

|an+1an|≫≫1n+11n|4x+1|≫≫nn+1|4x+1|⟶n→∞|4x+1|=L

This converges when L<1 or:

|4x+1|<1⟺|x+1/4|<1/4⟹converges|4x+1|>1⟺|x+1/4|>1/4⟹diverges

Preliminary interval: x∈(−1/2,0)

Check endpoints: ∑(4⋅−12+1)nn converges but ∑1n diverges.

Final interval of convergence: [−1/2,0)

Link to original