Previous page : The integral of sin(x)/x dx from -infinity to infinity
Next page : The integral of e^(-x^2) from –infinity to infinity
This integral was introduced to me by one of my students. It is from some hard maths contest – I could guess from William Lowell Putnam Mathematical Competition. Anyhow, it took me about one and a half hours to figure this one out, and then some hours to redo it and verify all steps and so on – about four hours all in all.
Ok, so the integral is
It seems like no standard methods would work. One way is to use the trick introduced in the first integral in this section, and that is to insert another variable. This enables us to get rid of the logarithm. Let
The reason for putting the variable there is that it will enable us to get rid of the logarithm, as we will see.
Differentiate with respect to p
Then we differentiate with respect to p to get
I.e.
Here we have to remember that p is our variable and that we need to use the chain rule.
Partial fractions
To solve the integral we need to rewrite our integrand using partial fractions
We should find values for A, B and C such that
The reason for the numerator Bx+C is that, when we multiply away the denominators, the two terms in the right-hand side (RHS) should be of the same degree – as we will see is necessary.
Next, we multiply both sides with the denominator of the left-hand side (LHS) to get
Now comes a really neat trick – so if you have not worked with partial fractions then listen carefully (as I would have said if this would have been a lesson).
When x varies the constant part of the LHS and the RHS must be the same. The same with the part that varies with x and the part that varies with x2. Each of these parts must vary in unison with each other. We thus get
Here we just write the coefficient, since we don´t care about any particular value of x. From the first equation, we get that A= –C. This gives us
We may now multiply the first equation of this with p to get
Next, we subtract the second equation from the first to get
or
From this, and that A= –C we get
Finally, from
We get
Back to our rational function we thus get
One can check that this is correct by adding the fractions (find a common denominator) and then by expanding and simplification.
Back to the integral
We now have
Solving the three integrals
For the first integral, we get
The second one is
To solve this we can make the substitutions
to get
The third one gives us
Combining this we get
Undoing the Differentiation with respect to p
Ok, to get back to I(p) we need to take the anti-derivative of I’(p).
The second term is virtually the same as the third one above, so we get
(We wait with the constant of integration.)
The third term is like the second one above, and we have
The third integral may be the most difficult to figure out, but if we remember that
then we can rewrite the integral as
Putting it all together
This gives us, after adding the constant of integration we get from the indefinite integrals
Solving this for a value of p that we can solve it for, namely p=0 we get
or
So C=0.
Now we can let p=1 to get back to our original integral
or
It is rather fascinating to see that the two last terms turned out to be the same. That gives us
We have thus basically been able to turn the integral to itself – but happily enough with a minus sign in front of it. We thus have
or
This finally gives us
Phew. That was a tough one.
Up a level : IntegralsPrevious page : The integral of sin(x)/x dx from -infinity to infinity
Next page : The integral of e^(-x^2) from –infinity to infinityLast modified: