Absolute And Conditional Convergence

An improper integral converges absolutely when the integral of | f | converges, and converges conditionally when it converges only because positive and negative pieces cancel. Absolute convergence is the stronger of the two: it always implies convergence, and it is the version you can test with the comparison tests, which require a nonnegative integrand.

Item Statement
Converges absolutely a | f ( x ) | d x converges
Converges conditionally a f ( x ) d x converges but a | f ( x ) | d x diverges
Main theorem absolute convergence convergence, never the reverse
Engine of the proof 0 f ( x ) + | f ( x ) | 2 | f ( x ) |
Standard absolute test find g with | f ( x ) | g ( x ) and a g convergent
Model absolutely convergent 1 cos x x 2 d x
Model conditionally convergent 0 sin x x d x (value π / 2 )
The whole family 1 sin x x p d x : absolute for p > 1 , conditional for 0 < p 1 , divergent for p 0
Block fact used constantly n π ( n + 1 ) π | sin x | d x = 2
Tool for conditional convergence integration by parts, which trades 1 / x p for an absolutely convergent 1 / x p + 1 tail
Comparison tests need f 0 , so apply them to | f | first
Pitfall 0 sin x x 2 d x diverges at x = 0 , not at
Pitfall convergence of a f does not force f ( x ) 0
Useful corollary a f absolutely convergent and g bounded a f g absolutely convergent

The Two Kinds of Convergence

The comparison tests of the previous section require the integrand to be nonnegative. When f changes sign, we distinguish two kinds of convergence: one where the integral of | f | also converges, and one where convergence relies on cancellation between the positive and the negative parts.

Let f be integrable on every bounded subinterval of [ a , ) .

  1. The integral a f ( x ) d x converges absolutely if a | f ( x ) | d x converges.
  2. The integral a f ( x ) d x converges conditionally if a f ( x ) d x converges but a | f ( x ) | d x diverges.

Read the two clauses as a pair of independent questions about the same integrand: does f converge, and does | f | converge? Because | f | 0 , its integral either converges to a finite number or increases to + ; there is no oscillation to worry about. The two answers together produce exactly three possibilities, and the next theorem rules out a fourth.

The picture below is the example to keep in mind for the rest of the section. The graph of y = ( sin x ) / x crosses the axis at every multiple of π , so the area it encloses splits into lobes that alternate in sign. Write A n for the (unsigned) area of the n th lobe.

TikZ figure

The signed areas of the lobes of y = ( sin x ) / x on [ 0 , 6 π ] alternate, and they shrink, but only about as fast as 1 / n . Numerically A 1 1.8519 , A 2 0.4338 , A 3 0.2566 , A 4 0.1826 , A 5 0.1418 , A 6 0.1159 . Adding the unsigned areas gives + ; adding the signed areas gives π / 2 .

The same two definitions apply verbatim to Type II improper integrals (those with an infinite discontinuity at an endpoint) and to integrals over ( , ) : in every case, replace f by | f | and ask the same question.

Absolute Convergence Implies Convergence

The key fact is that absolute convergence is the stronger condition. It is stronger in the strict sense: it implies convergence, and there are convergent integrals that are not absolutely convergent.

Absolute Convergence Implies Convergence. If a | f ( x ) | d x converges, then a f ( x ) d x converges.

\bbox[8px, #E6F0FA, border: 3px solid #0066CC]{\int_a^\infty |f(x)|\,dx\ \text{converges}\ \implies\ \int_a^\infty f(x)\,dx\ \text{converges}}

In words: if the total unsigned area is finite, then the signed area is finite too. Cancellation can only help.

Proof

Since | f ( x ) | f ( x ) | f ( x ) | , adding | f ( x ) | throughout gives

0 f ( x ) + | f ( x ) | 2 | f ( x ) | .

The point of this step is that f + | f | is a nonnegative function, so the Direct Comparison Test applies to it even though it does not apply to f .

Because a 2 | f ( x ) | d x converges, the Direct Comparison Test gives convergence of a ( f ( x ) + | f ( x ) | ) d x . Writing f ( x ) = ( f ( x ) + | f ( x ) | ) | f ( x ) | and using linearity:

\begin{aligned} \int_a^\infty f(x)\,dx &= \int_a^\infty \bigl(f(x)+|f(x)|\bigr)\,dx \\ &\quad - \int_a^\infty |f(x)|\,dx, \end{aligned}

a difference of two convergent integrals. Therefore a f ( x ) d x converges.

Two consequences are worth stating separately.

  • The contrapositive. If a f ( x ) d x diverges, then a | f ( x ) | d x diverges. In other words, a divergent integral can never be absolutely convergent.
  • The converse is false. Divergence of a | f ( x ) | d x tells you nothing about a f ( x ) d x ; the integral may still converge conditionally, as 0 sin x x d x does. This is the single most common misreading of the theorem.

A convenient bound comes free with the theorem: when a | f | converges,

| a f ( x ) d x | a | f ( x ) | d x ,

which is the improper-integral version of the triangle inequality. It often gives a numerical estimate at no extra cost.

Worked Examples

The first example is the standard pattern for absolute convergence: bound | f | by a p -integral. The second and third are the standard pattern for conditional convergence: get convergence from integration by parts, then get divergence of | f | from a block-by-block comparison with the harmonic series.

Show that 1 cos x x 2 d x converges absolutely.

Solution

We have

| cos x x 2 | = | cos x | x 2 1 x 2 ,

because | cos x | 1 for every x . Since 1 d x x 2 = 1 converges (the p -test with p = 2 > 1 ), the Direct Comparison Test gives convergence of 1 | cos x | x 2 d x . By definition, the integral converges absolutely.

By the theorem above, the original integral 1 cos x x 2 d x therefore converges as well, and the triangle inequality bounds its value: | 1 ( cos x ) / x 2 d x | 1 .

Show that 0 sin x x d x converges conditionally.

Solution

That the integral converges was established with the comparison tests: the singularity at x = 0 is removable because ( sin x ) / x 1 as x 0 + , and integration by parts on [ π , ) turns the tail into an absolutely convergent integral. See the worked example on the Dirichlet integral. It remains to show that 0 | sin x x | d x diverges.

On each interval [ n π , ( n + 1 ) π ] , for n = 1 , 2 , 3 , , the function | sin x | traces one full arch, so

n π ( n + 1 ) π | sin x | d x = 2 ,

independently of n . Also 1 / x 1 / ( ( n + 1 ) π ) on this interval, since x never exceeds ( n + 1 ) π there. Replacing 1 / x by that smallest value can only decrease the integral:

\begin{aligned} \int_{n\pi}^{(n+1)\pi}\frac{|\sin x|}{x}\,dx &\ge \frac{1}{(n+1)\pi}\int_{n\pi}^{(n+1)\pi}|\sin x|\,dx \\ &= \frac{2}{(n+1)\pi}. \end{aligned}

The picture below is exactly this estimate: each shaded arch is the graph of | sin x | flattened down to the constant height 1 / ( ( n + 1 ) π ) , and it sits under the true curve.

TikZ figure

Summing the block estimate over n = 1 , 2 , , N gives a lower bound for the integral over [ π , ( N + 1 ) π ] :

\bbox[8px, #E6F0FA, border: 3px solid #0066CC]{\int_\pi^{(N+1)\pi}\frac{|\sin x|}{x}\,dx \ \ge\ \frac{2}{\pi}\sum_{n=1}^{N}\frac{1}{n+1}}

The sum on the right is a tail of the harmonic series, which diverges, so the right side tends to + as N . Therefore 0 | sin x x | d x diverges, and the convergence of 0 sin x x d x is conditional.

The next example runs the same two-step argument from scratch, with the integration by parts written out in full. It also shows that nothing about the method depends on the decay being a power of x .

Show that 2 sin x ln x d x converges conditionally.

Solution

(i) Convergence. Integrate by parts with u = 1 / ln x and d v = sin x d x , so that

d u = 1 x ( ln x ) 2 d x , v = cos x .

Then for every b > 2 ,

\begin{aligned} \int_2^b \frac{\sin x}{\ln x}\,dx &= \left[-\frac{\cos x}{\ln x}\right]_2^b \\ &\quad - \int_2^b \frac{\cos x}{x(\ln x)^{2}}\,dx. \end{aligned}

The boundary term is cos b ln b + cos 2 ln 2 , and | cos b ln b | 1 ln b 0 as b , so it tends to cos 2 ln 2 .

The remaining integral converges absolutely, because

| cos x x ( ln x ) 2 | 1 x ( ln x ) 2 and 2 d x x ( ln x ) 2 = [ 1 ln x ] 2 = 1 ln 2 .

Both pieces have limits, so 2 sin x ln x d x converges. Integration by parts did the essential work: it converted an integrand decaying like 1 / ln x into one decaying like 1 / ( x ( ln x ) 2 ) , which is small enough to be handled by comparison.

(ii) Non-absolute convergence. Since π > 2 , every interval [ n π , ( n + 1 ) π ] with n 1 lies inside [ 2 , ) . On such an interval ln x ln ( ( n + 1 ) π ) , so 1 / ln x 1 / ln ( ( n + 1 ) π ) , and

\begin{aligned} \int_{n\pi}^{(n+1)\pi}\frac{|\sin x|}{\ln x}\,dx &\ge \frac{1}{\ln\bigl((n+1)\pi\bigr)}\int_{n\pi}^{(n+1)\pi}|\sin x|\,dx \\ &= \frac{2}{\ln\bigl((n+1)\pi\bigr)} \\ &\ge \frac{2}{(n+1)\pi}, \end{aligned}

where the last step uses ln t t . Summing over n = 1 , , N again produces a multiple of a partial sum of the harmonic series, which grows without bound. Hence 2 | sin x | ln x d x diverges and the convergence is conditional.

Why Conditional Convergence Depends on Cancellation

It is worth seeing why the comparison tests cannot possibly settle 0 ( sin x ) / x d x . The natural bound is

| sin x x | 1 x ( x > 0 ) ,

and the graph of ( sin x ) / x really does live inside the envelope formed by y = 1 / x and y = 1 / x , touching it near each peak and trough.

TikZ figure

The curve y = ( sin x ) / x trapped between y = ± 1 / x on [ 1 , 6 π ] . The envelope controls the size of the oscillations but is itself not integrable.

The trouble is that the envelope is 1 / x , and 1 d x / x diverges (the p -test with p = 1 ). A divergent upper bound gives no information, and part (b) of the Direct Comparison Test does not apply either, because | sin x | / x is not bounded below by any fixed multiple of 1 / x (it vanishes at every multiple of π ). So no comparison against the envelope can decide the question, in either direction. Something finer is needed, and that finer thing is cancellation.

Here is the accounting, in terms of the lobe areas A n from the first figure. The unsigned total is

0 | sin x x | d x = A 1 + A 2 + A 3 + = + ,

because the block estimate shows A n + 1 2 / ( ( n + 1 ) π ) and the harmonic series diverges. The signed total is the alternating sum

0 sin x x d x = A 1 A 2 + A 3 A 4 + = π 2 1.5708 ,

which converges because the A n decrease steadily to 0 . This is precisely the situation of the alternating harmonic series 1 1 2 + 1 3 , which converges while 1 + 1 2 + 1 3 + does not. The integral 0 ( sin x ) / x d x is called the Dirichlet integral; its exact value π / 2 requires techniques beyond this text, but the convergence is elementary.

Why does the distinction matter in practice? Because an absolutely convergent integral behaves like a proper integral: its value does not depend on the order in which you accumulate area, and estimates such as | f | | f | are available. A conditionally convergent integral has none of that stability. Its value is the limit of a delicately balanced cancellation, and small changes can destroy it. For instance, multiplying ( sin x ) / x by the bounded function sin x gives

sin 2 x x = 1 cos 2 x 2 x ,

whose integral over [ 1 , ) diverges, since 1 d x 2 x diverges while 1 cos 2 x 2 x d x converges. Multiplying an absolutely convergent integrand by a bounded function never causes such a collapse, as you will prove in the exercises.

A Strategy for Integrands That Change Sign

The following procedure, the Absolute-First Strategy, is what to do when you meet an integrand that is not of one sign. It is ordered so that the cheapest test comes first.

  1. Locate every improper point. Check each endpoint and every interior point where the integrand blows up, and split the integral so that each piece is improper at one point only. A single divergent piece makes the whole integral divergent.
  2. Test | f | first. On each piece, apply the Direct Comparison Test or the Limit Comparison Test to | f | , which is nonnegative and therefore legal input for those tests. The usual comparison functions are the p -integrals 1 / x p and the exponentials e c x .
  3. If | f | converges, stop. The integral converges absolutely, hence converges, and | f | | f | gives you a bound for free.
  4. If | f | diverges, conclude nothing yet. Return to f itself and try integration by parts, taking d v to be the oscillating factor ( sin x d x or cos x d x ) and u the slowly decaying factor. If the boundary term has a limit and the new integral converges absolutely, the original integral converges conditionally.
  5. If integration by parts fails, look for divergence. The standard argument: find intervals [ c n , d n ] marching off to infinity on which f keeps one sign and c n d n f d x δ for some fixed δ > 0 . If a f d x converged to L , these increments would have to tend to 0 , so no such δ could exist.

Two traps.

  • Forgetting the origin. The integral 0 sin x x 2 d x has a perfectly well behaved tail, since | sin x / x 2 | 1 / x 2 . It still diverges, because near x = 0 the integrand behaves like 1 / x . Step 1 exists precisely to catch this.
  • Assuming the integrand must vanish. For series, a n convergent forces a n 0 . For integrals there is no such rule: 1 sin ( x 2 ) d x converges although sin ( x 2 ) keeps swinging between 1 and 1 forever. Convergence of an improper integral is a statement about accumulated area, not about the size of the integrand.

Five Integrals Classified

The table applies the strategy to five standard integrands. Read each row left to right: the two middle columns are the two independent questions, and the verdict is determined by the pair.

Integral Does f converge? Does | f | converge? Verdict
0 sin x x d x yes no converges conditionally
1 sin x x 3 / 2 d x yes yes converges absolutely
1 sin x x d x yes no converges conditionally
1 cos x x 2 d x yes yes converges absolutely
0 e x sin x d x yes yes converges absolutely

Notice that the answer "no" in the third column never by itself settles anything: rows one and three are conditional, but an integral such as 1 sin x d x also has 1 | sin x | d x divergent and is itself divergent. The fourth logical combination, | f | convergent while f diverges, never appears: it is exactly what the theorem of this section forbids.

Summary. An absolutely convergent improper integral behaves like a proper integral: the value is not affected by the order of integration or by cancellation between positive and negative regions. A conditionally convergent integral converges only because of cancellation, and it diverges if the integrand is replaced by its absolute value. When applying comparison tests, check whether f 0 ; if not, consider first whether | f | converges (absolute convergence) before working with f directly.

Exercises

Determine whether 1 sin x x 3 / 2 d x converges absolutely, conditionally, or diverges.

Answer

Converges absolutely.

Solution

We have | sin x x 3 / 2 | 1 x 3 / 2 and 1 x 3 / 2 d x converges (the p -test with p = 3 / 2 > 1 ). By the Direct Comparison Test, 1 | sin x x 3 / 2 | d x converges. Since absolute convergence implies convergence, the integral converges, and it converges absolutely.

Determine whether 1 sin x x d x converges absolutely, conditionally, or diverges.

Answer

Converges conditionally.

Solution

(i) Convergence. We use integration by parts with u = 1 / x and d v = sin x d x , giving d u = 1 / ( 2 x 3 / 2 ) d x and v = cos x . Then:

\begin{aligned} \int_1^b \frac{\sin x}{\sqrt{x}}\,dx &= \Bigl[-\frac{\cos x}{\sqrt{x}}\Bigr]_1^b \\ &\quad - \int_1^b \frac{\cos x}{2x^{3/2}}\,dx. \end{aligned}

As b : cos b / b 0 (since | cos b / b | 1 / b 0 ), and the remaining integral converges absolutely since | cos x / ( 2 x 3 / 2 ) | 1 / ( 2 x 3 / 2 ) and 1 x 3 / 2 d x converges. Therefore the integral converges.

(ii) Non-absolute convergence. On [ n π , ( n + 1 ) π ] we have 1 / x 1 / ( n + 1 ) π , so:

\begin{aligned} \int_{n\pi}^{(n+1)\pi}\frac{|\sin x|}{\sqrt{x}}\,dx &\ge \frac{1}{\sqrt{(n+1)\pi}}\int_{n\pi}^{(n+1)\pi}|\sin x|\,dx \\ &= \frac{2}{\sqrt{(n+1)\pi}}. \end{aligned}

Since n = 1 1 / n + 1 diverges (comparison with the harmonic series via 1 / n + 1 1 / ( n + 1 ) , or more directly since n 1 / 2 diverges), we conclude that 1 | sin x | x d x diverges. Therefore the convergence is conditional.

Show that 0 e x sin x d x converges absolutely, and evaluate it.

Answer

1 2 .

Solution

Since | e x sin x | e x and 0 e x d x = 1 converges, the Direct Comparison Test gives convergence of 0 e x | sin x | d x . Therefore the integral converges absolutely.

To evaluate the integral, we integrate by parts twice. Let I = 0 e x sin x d x .

First integration by parts. Let u = sin x and d v = e x d x , so d u = cos x d x and v = e x :

\begin{aligned} \int_0^b e^{-x}\sin x\,dx &= \bigl[-e^{-x}\sin x\bigr]_0^b + \int_0^b e^{-x}\cos x\,dx \\ &= -e^{-b}\sin b + \int_0^b e^{-x}\cos x\,dx. \end{aligned}

Second integration by parts. Let u = cos x and d v = e x d x , so d u = sin x d x and v = e x :

\begin{aligned} \int_0^b e^{-x}\cos x\,dx &= \bigl[-e^{-x}\cos x\bigr]_0^b - \int_0^b e^{-x}\sin x\,dx \\ &= (-e^{-b}\cos b + 1) - \int_0^b e^{-x}\sin x\,dx. \end{aligned}

Solving for I . Substituting back:

0 b e x sin x d x = e b sin b + ( e b cos b + 1 ) 0 b e x sin x d x .

Adding 0 b e x sin x d x to both sides:

2 0 b e x sin x d x = 1 e b ( sin b + cos b ) .

Since | e b ( sin b + cos b ) | 2 e b 0 as b , taking the limit gives 2 I = 1 , so I = 1 2 .

Determine whether 1 cos ( π x ) x d x converges absolutely, conditionally, or diverges.

Answer

Converges conditionally.

Solution

(i) Convergence. We use integration by parts with u = 1 / x and d v = cos ( π x ) d x , so d u = 1 / x 2 d x and v = sin ( π x ) / π . Then:

\begin{aligned} \int_1^b \frac{\cos(\pi x)}{x}\,dx &= \Bigl[\frac{\sin(\pi x)}{\pi x}\Bigr]_1^b \\ &\quad + \int_1^b \frac{\sin(\pi x)}{\pi x^{2}}\,dx. \end{aligned}

As b : sin ( π b ) / ( π b ) 0 , and sin ( π 1 ) / ( π 1 ) = 0 since sin π = 0 , so the boundary term tends to 0 . For the remaining integral, | sin ( π x ) / ( π x 2 ) | 1 / ( π x 2 ) and 1 d x / x 2 converges, so by the Direct Comparison Test the remaining integral converges absolutely. Therefore 1 cos ( π x ) x d x converges.

(ii) Non-absolute convergence. The function cos ( π x ) has period 2 , so on each interval [ n , n + 1 ] with n 1 it completes exactly half a period and | cos ( π x ) | traces one full arch. Hence n n + 1 | cos ( π x ) | d x = 2 π , the same value for every n . Also 1 / x 1 / ( n + 1 ) on [ n , n + 1 ] . Therefore:

n n + 1 | cos ( π x ) | x d x 1 n + 1 2 π .

Summing over n = 1 , , N :

1 N + 1 | cos ( π x ) | x d x 2 π n = 1 N 1 n + 1 .

Since the harmonic series diverges, 1 | cos ( π x ) | x d x diverges. The convergence of 1 cos ( π x ) x d x is therefore conditional.

Determine whether 1 cos x x 2 + 1 d x converges absolutely, conditionally, or diverges. If it converges, give a numerical bound for its value.

Answer

Converges absolutely, and its value is at most π / 4 0.7854 in absolute value.

Solution

The integrand changes sign, so we test | f | first. Since | cos x | 1 and x 2 + 1 x 2 > 0 for x 1 ,

| cos x x 2 + 1 | = | cos x | x 2 + 1 1 x 2 + 1 1 x 2 .

Since 1 d x x 2 = 1 converges (the p -test with p = 2 > 1 ), the Direct Comparison Test gives convergence of 1 | cos x x 2 + 1 | d x . The integral therefore converges absolutely, and by the theorem of this section it converges.

For the bound, use the triangle inequality together with the sharper comparison function 1 / ( x 2 + 1 ) :

\begin{aligned} \left|\int_1^\infty \frac{\cos x}{x^{2}+1}\,dx\right| &\le \int_1^\infty \frac{dx}{x^{2}+1} \\ &= \lim_{b\to\infty}\bigl[\arctan x\bigr]_1^b \\ &= \frac{\pi}{2}-\frac{\pi}{4} \\ &= \frac{\pi}{4}. \end{aligned}

This is a typical dividend of absolute convergence: the same comparison that proves convergence also produces an error bound.

Show that 0 e x cos x d x converges absolutely, and evaluate it.

Answer

1 2 .

Solution

Absolute convergence. Since | cos x | 1 we have | e x cos x | e x for x 0 , and

0 e x d x = lim b [ e x ] 0 b = lim b ( 1 e b ) = 1

converges. By the Direct Comparison Test, 0 e x | cos x | d x converges, so the integral converges absolutely.

Evaluation. Let J = 0 e x cos x d x . Integrate by parts with u = cos x , d v = e x d x , so d u = sin x d x and v = e x :

\begin{aligned} \int_0^b e^{-x}\cos x\,dx &= \bigl[-e^{-x}\cos x\bigr]_0^b - \int_0^b e^{-x}\sin x\,dx \\ &= \bigl(1-e^{-b}\cos b\bigr) - \int_0^b e^{-x}\sin x\,dx. \end{aligned}

Now integrate the remaining integral by parts with u = sin x , d v = e x d x , so d u = cos x d x and v = e x :

\begin{aligned} \int_0^b e^{-x}\sin x\,dx &= \bigl[-e^{-x}\sin x\bigr]_0^b + \int_0^b e^{-x}\cos x\,dx \\ &= -e^{-b}\sin b + \int_0^b e^{-x}\cos x\,dx. \end{aligned}

Substituting the second computation into the first:

\begin{aligned} \int_0^b e^{-x}\cos x\,dx &= 1-e^{-b}\cos b + e^{-b}\sin b \\ &\quad - \int_0^b e^{-x}\cos x\,dx. \end{aligned}

Moving the last integral to the left side gives

2 0 b e x cos x d x = 1 e b ( cos b sin b ) .

Since | e b ( cos b sin b ) | 2 e b 0 as b , we get 2 J = 1 , so J = 1 2 .

Note that this is the same value as 0 e x sin x d x from an earlier exercise; the two integrals are equal, although the functions are not.

Show that 1 sin ( x 2 ) d x converges, and decide whether the convergence is absolute or conditional. Comment on why the result is surprising.

Answer

Converges conditionally. It is surprising because sin ( x 2 ) does not tend to 0 as x .

Solution

Substitution. On any finite interval [ 1 , b ] , put u = x 2 . Then d u = 2 x d x , and since x = u > 0 we have d x = d u 2 u . The limits become u = 1 and u = b 2 , so

1 b sin ( x 2 ) d x = 1 2 1 b 2 sin u u d u .

As b we also have b 2 , so the left side has a limit exactly when the right side does:

1 sin ( x 2 ) d x = 1 2 1 sin u u d u .

Convergence. The integral 1 sin u u d u was shown to converge in the exercise on sine over the square root, by integration by parts. Hence 1 sin ( x 2 ) d x converges.

Not absolutely. The same substitution applied to | sin ( x 2 ) | gives

1 | sin ( x 2 ) | d x = 1 2 1 | sin u | u d u ,

and part (ii) of that same exercise shows the right side is + . So the convergence is conditional.

Why it is surprising. For an infinite series, convergence of a n forces a n 0 . Here sin ( x 2 ) returns to 1 whenever x = π / 2 + 2 k π and to 1 whenever x = 3 π / 2 + 2 k π , so it never settles down. What does happen is that the successive zeros of sin ( x 2 ) , namely x = n π , crowd together as x grows: consecutive zeros are about π 2 n π apart. The lobes become narrow fast enough that their areas shrink, and the alternating signs finish the job. Numerically the integral equals about 0.3164 .

Let x denote the greatest integer less than or equal to x . Determine whether 1 ( 1 ) x x d x converges absolutely, conditionally, or diverges.

Answer

Converges conditionally.

Solution

Reduce to a series. For n x < n + 1 we have x = n , so the integrand equals ( 1 ) n / x on that interval. The integrand has a jump at each integer but is bounded, so it is integrable on every bounded subinterval, and

n n + 1 ( 1 ) x x d x = ( 1 ) n n n + 1 d x x = ( 1 ) n ln ( 1 + 1 n ) .

Adding these for n = 1 , , N 1 gives

1 N ( 1 ) x x d x = n = 1 N 1 ( 1 ) n ln ( 1 + 1 n ) .

Convergence. The terms a n = ln ( 1 + 1 n ) are positive, strictly decreasing (because 1 + 1 n decreases), and tend to ln 1 = 0 . By the Alternating Series Test the series converges, so the integrals over integer upper limits converge to some number L . For a general upper limit b with N b < N + 1 ,

| N b ( 1 ) x x d x | N N + 1 d x x = ln ( 1 + 1 N ) 0 ,

so 1 b has the same limit L as b through all real values. The integral converges.

Not absolutely. Since | ( 1 ) x | = 1 for every x ,

1 | ( 1 ) x x | d x = 1 d x x ,

which diverges by the p -test with p = 1 . Therefore the convergence is conditional.

This integral is the exact analogue of the alternating harmonic series: it converges for precisely the same reason, and it fails to converge absolutely for precisely the same reason. Its value can be evaluated with Wallis's product and equals ln ( 2 / π ) 0.4516 .

Determine whether 0 sin x x 2 d x converges or diverges. Be careful to check both ends of the interval.

Answer

Diverges, because of the singularity at x = 0 . The tail 1 sin x x 2 d x by itself converges absolutely.

Solution

Step 1: split at the two improper points. The integral is improper at x = 0 , where the integrand is unbounded, and at + . Split at x = 1 :

0 sin x x 2 d x = 0 1 sin x x 2 d x + 1 sin x x 2 d x .

Step 2: the tail is harmless. For x 1 we have | sin x x 2 | 1 x 2 , and 1 d x x 2 = 1 converges. By the Direct Comparison Test the tail converges absolutely.

Step 3: the origin is not. Near x = 0 we have sin x x , so the integrand behaves like x / x 2 = 1 / x . Confirm this with the Limit Comparison Test using g ( x ) = 1 / x , which is legitimate because sin x > 0 on ( 0 , 1 ] , so the integrand is positive there:

lim x 0 + ( sin x ) / x 2 1 / x = lim x 0 + sin x x = 1.

Since L = 1 lies in ( 0 , ) and 0 1 d x x diverges (the p -test for Type II integrals with p = 1 ), the piece 0 1 sin x x 2 d x diverges.

Conclusion. One divergent piece makes the whole integral divergent, so 0 sin x x 2 d x diverges.

Compare this with 0 sin x x d x , where the extra factor of x in the denominator is absent and the singularity at the origin is removable, since ( sin x ) / x 1 . Raising the exponent from 1 to 2 improves the behaviour at infinity and ruins it at the origin.

True or false: if f is continuous on [ a , ) and a f ( x ) d x converges, then lim x f ( x ) = 0 . Justify your answer, and state a condition under which the conclusion does hold.

Answer

False. The conclusion does hold if, in addition, f is monotone on [ a , ) , and also if f is uniformly continuous on [ a , ) .

Solution

A first counterexample. By the exercise on 1 sin ( x 2 ) d x , that integral converges, yet sin ( x 2 ) takes the value 1 at x = π / 2 + 2 k π and the value 1 at x = 3 π / 2 + 2 k π for every k . So f ( x ) = sin ( x 2 ) has no limit at all as x , let alone the limit 0 .

A second counterexample, with spikes. Let f be the function that equals 1 on each interval [ n , n + 2 n ] for n = 1 , 2 , 3 , and equals 0 elsewhere. (These intervals are disjoint, since 2 n < 1 .) On any bounded interval f has finitely many discontinuities, so it is integrable there, and for every b ,

1 b f ( x ) d x n = 1 2 n = 1.

The function b 1 b f is nondecreasing and bounded above, so it has a limit; the integral converges (to 1 ). But f ( n ) = 1 for every integer n , so f ( x ) 0 . Replacing the rectangles by narrow continuous triangles of height 1 and base 2 n makes the same example continuous.

When the conclusion is true. If f is monotone on [ a , ) and a f converges, then f ( x ) 0 : a monotone function has a limit L at infinity, and L 0 would force the integral of a function eventually bounded away from 0 to diverge. The same conclusion holds if f is uniformly continuous on [ a , ) : uniform continuity stops the spikes from becoming arbitrarily narrow, since if | f ( x k ) | ε for some sequence x k , then | f | ε / 2 on an interval of one fixed length around each x k , and those intervals contribute the same positive amount of area infinitely often.

The moral is that convergence of an improper integral is a statement about accumulated area, not about the size of the integrand. This is the main way integrals differ from series.

Suppose a f ( x ) d x converges absolutely and g is integrable on every bounded subinterval of [ a , ) with | g ( x ) | M for all x a . Prove that a f ( x ) g ( x ) d x converges absolutely, and show by example that the statement fails if a f is only conditionally convergent.

Answer

The product integral converges absolutely, with | a f g | M a | f | . The statement fails for conditionally convergent f : take f ( x ) = ( sin x ) / x and g ( x ) = sin x on [ 1 , ) .

Solution

Proof. The product f g is integrable on every bounded subinterval, being a product of two such functions. For every x a ,

| f ( x ) g ( x ) | = | f ( x ) | | g ( x ) | M | f ( x ) | ,

and M | f | is nonnegative with a M | f ( x ) | d x = M a | f ( x ) | d x convergent by hypothesis. Since | f g | 0 , the Direct Comparison Test applies and gives convergence of a | f ( x ) g ( x ) | d x . By the theorem of this section, a f ( x ) g ( x ) d x converges, and it converges absolutely. Taking absolute values through the limit gives the bound

| a f ( x ) g ( x ) d x | a | f ( x ) g ( x ) | d x M a | f ( x ) | d x .

Failure for conditional convergence. Let a = 1 , f ( x ) = sin x x and g ( x ) = sin x , so | g | 1 is bounded and 1 f converges conditionally. Their product is

f ( x ) g ( x ) = sin 2 x x = 1 cos 2 x 2 x ,

using the identity sin 2 x = 1 2 ( 1 cos 2 x ) . Now 1 cos 2 x 2 x d x converges (integrate by parts exactly as for ( cos x ) / x ), while 1 d x 2 x diverges. A divergent integral minus a convergent one is divergent, so 1 sin 2 x x d x diverges.

Multiplying by a bounded function therefore destroyed the convergence. This is the precise sense in which conditional convergence is fragile: it depends on cancellation, and a bounded factor can remove the cancellation.

Classify each integral as absolutely convergent, conditionally convergent, or divergent.

  1. 1 cos x x 1 / 3 d x
  2. 1 arctan x x 2 sin x d x
  3. 1 x sin x d x
Answer

(a) converges conditionally. (b) converges absolutely. (c) diverges.

Solution

(a) First test | f | . On [ n π , ( n + 1 ) π ] the graph of cos x runs from ± 1 down through a zero at the midpoint to 1 , so | cos x | traces two half-arches whose areas together give n π ( n + 1 ) π | cos x | d x = 2 , and x 1 / 3 ( ( n + 1 ) π ) 1 / 3 there. Hence

n π ( n + 1 ) π | cos x | x 1 / 3 d x 2 ( ( n + 1 ) π ) 1 / 3 ,

and n ( n + 1 ) 1 / 3 diverges (its terms exceed those of the harmonic series), so 1 | cos x | x 1 / 3 d x diverges. Absolute convergence fails.

Now test 1 f itself by parts, with u = x 1 / 3 and d v = cos x d x , so d u = 1 3 x 4 / 3 d x and v = sin x :

\begin{aligned} \int_1^b \frac{\cos x}{x^{1/3}}\,dx &= \left[\frac{\sin x}{x^{1/3}}\right]_1^b \\ &\quad + \frac{1}{3}\int_1^b \frac{\sin x}{x^{4/3}}\,dx. \end{aligned}

The boundary term tends to sin 1 , since | sin b | / b 1 / 3 b 1 / 3 0 . The remaining integral converges absolutely because | sin x / x 4 / 3 | x 4 / 3 and 4 / 3 > 1 . So the integral converges, and since | f | diverges the convergence is conditional.

(b) Since 0 < arctan x < π 2 for x 1 and | sin x | 1 ,

| arctan x x 2 sin x | π / 2 x 2 ,

and 1 d x x 2 converges. By the Direct Comparison Test the integral converges absolutely. (This is also the previous exercise applied with f ( x ) = sin x / x 2 , which is absolutely convergent, and the bounded factor g ( x ) = arctan x .)

(c) Here the oscillation grows instead of decaying. An antiderivative of x sin x is sin x x cos x , as differentiating confirms:

d d x ( sin x x cos x ) = cos x cos x + x sin x = x sin x .

Therefore

1 b x sin x d x = ( sin b b cos b ) ( sin 1 cos 1 ) .

Taking b = 2 N π gives 2 N π for the first bracket, while b = ( 2 N + 1 ) π gives + ( 2 N + 1 ) π + . The partial integrals oscillate with unbounded amplitude, so no limit exists and the integral diverges.

Let p be a real number. Classify

1 sin x x p d x

as absolutely convergent, conditionally convergent, or divergent, according to the value of p .

Answer

Absolutely convergent for p > 1 ; conditionally convergent for 0 < p 1 ; divergent for p 0 .

Solution

Case 1: p > 1 , absolute convergence. Since | sin x | 1 ,

| sin x x p | 1 x p ( x 1 ) ,

and 1 x p d x converges by the p -test because p > 1 . The Direct Comparison Test gives convergence of 1 | sin x x p | d x , so the integral converges absolutely.

Case 2: 0 < p 1 , conditional convergence. Two things must be shown.

Convergence. Integrate by parts with u = x p and d v = sin x d x , so d u = p x p 1 d x and v = cos x :

\begin{aligned} \int_1^b \frac{\sin x}{x^{p}}\,dx &= \left[-\frac{\cos x}{x^{p}}\right]_1^b \\ &\quad - p\int_1^b \frac{\cos x}{x^{p+1}}\,dx. \end{aligned}

Because p > 0 , the boundary term satisfies | cos b b p | 1 b p 0 , so it tends to cos 1 . For the remaining integral, | cos x x p + 1 | 1 x p + 1 with p + 1 > 1 , so 1 cos x x p + 1 d x converges absolutely. Both parts have limits, so the integral converges. This is the payoff of integration by parts: it raises the exponent from p to p + 1 , moving it across the threshold 1 .

Non-absolute convergence. On [ n π , ( n + 1 ) π ] we have x p ( ( n + 1 ) π ) p , so

\begin{aligned} \int_{n\pi}^{(n+1)\pi}\frac{|\sin x|}{x^{p}}\,dx &\ge \frac{1}{\bigl((n+1)\pi\bigr)^{p}}\int_{n\pi}^{(n+1)\pi}|\sin x|\,dx \\ &= \frac{2}{\pi^{p}(n+1)^{p}}. \end{aligned}

Summing over n = 1 , , N bounds π ( N + 1 ) π | sin x | x p d x below by 2 π p n = 1 N 1 ( n + 1 ) p , and the series ( n + 1 ) p diverges for p 1 . Hence 1 | sin x x p | d x diverges and the convergence is conditional.

Case 3: p 0 , divergence. Write q = p 0 , so the integrand is x q sin x . On [ 2 n π , ( 2 n + 1 ) π ] we have sin x 0 and x q 1 (since x 1 and q 0 ), so

2 n π ( 2 n + 1 ) π x q sin x d x     2 n π ( 2 n + 1 ) π sin x d x = 2

for every n 1 . Now suppose the integral converged, say 1 b sin x x p d x L as b . Then the difference of the values at b = ( 2 n + 1 ) π and b = 2 n π would tend to L L = 0 . But that difference is exactly the integral above, which is at least 2 for every n . This contradiction shows the integral diverges.

Summary of the trichotomy. Reading the three cases together: the exponent p = 1 separates absolute from conditional convergence, and the exponent p = 0 separates conditional convergence from divergence. The case p = 1 is the tail of the Dirichlet integral of this section, and p = 1 / 2 is the exercise on sine over the square root. Every classification question about 1 ( sin x ) / x p d x reduces to locating p relative to these two thresholds.