Improper Integrals over Unbounded Interva

Improper integrals for infinite intervals are defined in a similar way

Suppose we want to calculate the integral of a continuous function over an unboundede interval of the from [a, +∞), (−∞, b] or (−∞, −∞). Like the intergral with unbounded integrand that we studied in the last section, these integrals are aid to be improper.

Definition 6.7.1. Let f : [a, +∞] ⟶ ℝ continuous. We define

a + f ( x ) d x = lim ω + a ω f ( x ) d x

If the limit exits finite then f is said integrable on [a, +∞) or that the integral is convergent. If the limit is ±∞ we say that the integral diverges. If, however, no such limit exists, we say that the improper integral does not exists.

An analogous definition is given if f: (−∞, b].

b f ( x ) d x = lim ω ω b f ( x ) d x

Finally if f: (−∞, ∞) ⟶ ℝ is continuos, we let

+ f ( x ) d x = c f ( x ) d x + c f ( x ) d x

for an arbitrary point c.

Example 6.7.2. Calculate the following imporper integral.

0 d x 1 + x 2

we first calculate the integral

0 ω d x 1 + x 2

|tan−1|0ω = tan−1 ω − tan−1 0 = tan−1 ω

The limω ⟶ ∞ tan−1 ω = π/2. So the integral converges to π/2.  ■

Example 6.7.3. Evaluate the improper integral

+ 1 1 + x 2 d x

Solution. To evaluate this integral, we break the interval into two pieces (−∞, +∞) = (−∞, 0) ∪ [0, +∞). The choice of zero is arbitrary;

+ 1 1 + x 2 d x = lim ω 1 1 + x 2 d x = lim ω [ arctan ( x ) ] 0 ω = lim ω ( arctan ω 0 ) = π / 2

The second integral evaluates to π/2 in a similar manner. We conclude that the original integral over the entire real line is convergent and that its value is π/2 + π/2 = π.  ■

Example 6.7.3. Evaluate the integral

1 + 1 x α d x α > 0

Case α = 1. We have

1 ω 1 x α d x = [ log x ] 1 ω = log ω

Since limω ⟶ +∞ log ω = +∞, the integral is divergent.

Case α ≠ 1. We have

1 ω 1 x α d x = 1 1 α [ x 1 α ] 1 ω = 1 1 α ( ω 1 α 1 )

Thus

1 + 1 x α d x = lim ω + 1 1 α ( ω 1 α 1 ) = { + if  α < 1 1 α 1 if  α > 1

Hence we have

1 + 1 x α d x is  { divergent to  if  α 1 convergent to  1 α 1 if  α > 1

Divergence of the harmonic series

When we studied infinite series we stated that the harmonic series Σ 1/n is divergent; We shall now prove it. From figure 1, it can be easily seen that

1 N 1 x d x < n = 1 N 1 1 n
harmonic series integral convergence
Fig.1. The yellow region under the red curve is equal to ∫16 1/x dx whilst the sum of area of the rectangles is equal to ∑5n=1 1/n

Since ∫N1 1/x dx = log N ⟶ +∞ if N ⟶ ∞, then ∑Nn=1 1/n as well ⟶ +∞ if N ⟶ +∞, thus the harmonic series diverges.

Congergence of the generalized harmonic series

We studied the generalized harmonic series

n = 1 1 n α

converges for α > 1 (and diverges for α ≤ 1). We have proved for α ≥ 2 (by comparison with the Mengoli's series) and we are now going to provide a proof for any α > 1, in a similar manner used for the proof of conergence of the harmonic series. From fig. 2 it is clear that

n = 2 N 1 n α 1 N d x x α
generalized harmonic series integral convergence
Fig.2. The series is trapped under the function

Since for α > 1 and N ⟶ ∞, the integral converges the series converges as well.

«Epoxides Index Sulfides»