Can You Integrate (ln ln x)^n Using a Taylor Series?

[flouran]

How do you integrate (ln ln x)^n for any n?

 

[nicksauce]

Well since mathematica isn't able to find a formula for n=2, I'm going to say "numerically".

 

[flouran]

Well since mathematica isn't able to find a formula for n=2, I'm going to say "numerically".

What do you mean exactly by "numerically"? Do you mean that i should evaluate it as a definite integral?

 

[nicksauce]

Well I just mean if you want to find a general closed for expression for the indefinite integral, you are out of luck. Therefore the only way I can conceive of doing an integral with this expression would be to do a definite integral numerically.

 

[22/7]

there is one place where the integral can be evaluated explicitly that I know of.
\int-log[-log[x]]dx=Euler Constant (.577...)
This follows from differentionation the gamma function in its product and integral forms and making a change of variables.

 

[flouran]

Well I just mean if you want to find a general closed for expression for the indefinite integral, you are out of luck. Therefore the only way I can conceive of doing an integral with this expression would be to do a definite integral numerically.

Although this method would be painful, since I cannot express this integral in terms of elementary functions, could I represent (ln ln x)^(n) as a Taylor polynomial (what is the Taylor series for ln ln x, anyways?) and then integrate that and leave it as a Taylor Series?