Hard Light Productions Forums

Off-Topic Discussion => General Discussion => Topic started by: Kosh on December 16, 2008, 12:58:36 am

Title: integral of......
Post by: Kosh on December 16, 2008, 12:58:36 am
ln(tan(x)) dx


I'm not even sure where to start with this. Anyone got some pointers so I can get going on it? Thanks.
Title: Re: integral of......
Post by: redsniper on December 16, 2008, 01:48:08 am
Well, you'll have to use the Chain Rule, I'm pretty sure.
Title: Re: integral of......
Post by: Kosh on December 16, 2008, 01:58:52 am
the chain rule is for derivatives, this is an integration problem.
Title: Re: integral of......
Post by: portej05 on December 16, 2008, 02:39:14 am
Where did you get this problem - it looks ugly.
Try mathematica:
]&random=false]http://integrals.wolfram.com/index.jsp?expr=Log[+Tan (http://integrals.wolfram.com/index.jsp?expr=Log[+Tan[x)
Title: Re: integral of......
Post by: ssmit132 on December 16, 2008, 02:58:24 am
We just did integrals in our final Maths B test...

...but that was only with Logs of base 10. What is this for, exactly?
Title: Re: integral of......
Post by: CP5670 on December 16, 2008, 03:08:23 am
You can't express this in terms of elementary functions. There are various additional, standard functions that allow you to write things like this in closed form. My guess is it would involve a polylog or Lerch function somewhere, just from looking at it.
Title: Re: integral of......
Post by: portej05 on December 16, 2008, 03:50:45 am
We just did integrals in our final Maths B test...

...but that was only with Logs of base 10. What is this for, exactly?

You sure it wasn't base e - these are much easier.

It's been too long since my last maths units....
Title: Re: integral of......
Post by: ssmit132 on December 16, 2008, 04:59:12 am
No, we did logs of base 10. Although we just did simple stuff. Like log xn = n log x and so forth. And most of the time on the calculator. We haven't done ln (base e) yet. Knowing some of the people in my class, I'm sure they had trouble with these logs, which, on our scientific calculators, can ONLY do base 10 logs.  ;)

This is Maths B in Queensland, and we haven't got a good reputation when it comes to smartness IIRC.  :hopping:
Title: Re: integral of......
Post by: portej05 on December 16, 2008, 05:59:33 am
Just wait 'til you see the new WA curriculum  :)

I thought you meant integrals of log10

Don't forget that ln x = log(x)/log(e)... Which scientific calculator do you have?

Keep up the good work - I'm originally from Queensland!
Title: Re: integral of......
Post by: ssmit132 on December 16, 2008, 06:05:09 am
A Casio fx-82MS.

And thank you, the same to you!
Title: Re: integral of......
Post by: Kosh on December 16, 2008, 07:15:09 am
We just did integrals in our final Maths B test...

...but that was only with Logs of base 10. What is this for, exactly?

Calculus homework, and it isn't log(tanx), it's ln(tanx).
Title: Re: integral of......
Post by: portej05 on December 16, 2008, 07:42:16 am
That's still just as ugly. Mathematica uses Log to mean the natural logarithm (as opposed to using ln(x)). Still Ugly.
Title: Re: integral of......
Post by: Kosh on December 16, 2008, 07:54:24 am
yeah I know....   :doubt:
Title: Re: integral of......
Post by: FreeSpaceFreak on December 16, 2008, 07:55:43 am
I think you'd have to use integration by parts, not sure though.
Title: Re: integral of......
Post by: Jeff Vader on December 16, 2008, 07:58:19 am
Undoubtedly, but even then that thing is a *****. Hopefully solving that problem is not a requirement for passing the course.
Title: Re: integral of......
Post by: portej05 on December 16, 2008, 08:32:14 am
Integration by parts isn't suitable for this question since you've got f(g(x)), not f(x)g'(x).

I think CP5670 summed it up best:
You can't express this in terms of elementary functions. There are various additional, standard functions that allow you to write things like this in closed form. My guess is it would involve a polylog or Lerch function somewhere, just from looking at it.

I've got no idea what he's on about (I really haven't done enough maths... and I can't remember all of it), but mathematica seems to spit out an answer along those lines (you gotta trust somebody :p)
Agreed with hoping that that's not a requirement for your course.
Title: Re: integral of......
Post by: Jeff Vader on December 16, 2008, 08:36:26 am
Integration by parts isn't suitable for this question since you've got f(g(x)), not f(x)g'(x).
Whoops. Correct you are. Pardon our confusion.

Edit: Hmm. I consulted The Holy BookTM and started to work on the following (if I'm spewing out lies, hit me with a stick or something). If we have g(f(x)), deriving it would be Dg(f(x)) = g'(f(x))f'(x). From there, we could extrapolate that

g(f(x)) = ∫ g ' (f(x)) f ' (x) dx

, where f(x) = tan(x) and g(x) = lnx. That way we'd get

ln(tan(x)) = ∫(1 / tanx) * (1 + tan2(x))dx

or something. Brainfreeze.


Edit: Damn, ignore that. Dunno what I was thinking.
Title: Re: integral of......
Post by: FreeSpaceFreak on December 16, 2008, 08:49:09 am
Integration by parts IS suitable here, I think.

What I meant is that (ln(tanx)) is the f(x), and you have to introduce a g'(x) equal to 1. Then start integrating by parts.
Title: Re: integral of......
Post by: redsniper on December 16, 2008, 09:20:44 am
Skip it.
Title: Re: integral of......
Post by: CP5670 on December 16, 2008, 10:16:30 am
Are you supposed to find the value of this integral in some interval or actually get an antiderivative? I'm guessing it's the former, since as I said earlier the latter cannot be done using only elementary functions.

Quote
That's still just as ugly. Mathematica uses Log to mean the natural logarithm (as opposed to using ln(x)). Still Ugly.

It's conventionally written as log in pretty much all math beyond the elementary level. They only differ by a constant factor anyway.

Quote
I've got no idea what he's on about (I really haven't done enough maths... and I can't remember all of it), but mathematica seems to spit out an answer along those lines (you gotta trust somebody )

Yeah, I just put it in there and it does involve a polylog as I expected. This (http://en.wikipedia.org/wiki/Polylog) is the thing I am referring to.

This is one of numerous "special functions" that can't be expressed using the standard, elementary functions. They come up a lot in various places though and are useful enough to be considered as basic objects in themselves.
Title: Re: integral of......
Post by: Mika on December 16, 2008, 04:06:17 pm
Arrived in slightly different special function, maybe someone could explain why?
And how the hell you can put up those neato integrate signs here?

Integrate:
ln(tan x) dx  || Substituted tan(x)=u and solved dx = 1/(1+u^2) du

Arrived at:
INT { ln(u)*[1/(1+u^2)] }du, partial integration resulted in

ln(u)*arctan(u) - INT { arctan(u)/u }du, which results in a special function also.

The handbook of mathematical functions, (Abramowitz-Stegun) states that:
INT { arctan(u) / u } du = INT from 0 to INF { exp(-u*t)*si(t) }dt,
where si is a special function also, defined on the same book.

It was probably something like INT from Z to INF { (sin t) / t }dt.

Mika
Title: Re: integral of......
Post by: CP5670 on December 16, 2008, 04:40:43 pm
Quote
Arrived in slightly different special function, maybe someone could explain why?

Many of them are related to each other. There are three or four fundamental classes of functions (hypergeometric, zeta/Lerch and elliptic come to mind) and the others are all basically combinations or special cases of those. The si (sine integral) and polylog are both of the hypergeometric variety.

There are a lot of essentially redundant functions in use since they often came from different contexts. I know si for example is used in speech recognition, even though it's just a slight variant of the arguably more basic ei.

Quote
And how the hell you can put up those neato integrate signs here?

You can find it in charmap. Getting the superscripts and subscripts to line up can be annoying though. I think it's clearer to just write something like Integral[ log(tan(x)) , {x,0,pi}].
Title: Re: integral of......
Post by: Mongoose on December 16, 2008, 11:30:33 pm
This is exactly why I despised integrals so much.  Derivatives are such beautiful, elegant things...learn a few simple rules, and you can derive out even the most complex function.  But as soon as you get into integration, you realize that the rules are largely useless, as there are gobs of types out there that are flat-out impossible to do.  This is about the point in my schooling when I decided to yell a very hearty "**** you!" to math.
Title: Re: integral of......
Post by: CP5670 on December 17, 2008, 12:02:35 am
Quote
This is exactly why I despised integrals so much.  Derivatives are such beautiful, elegant things...learn a few simple rules, and you can derive out even the most complex function.  But as soon as you get into integration, you realize that the rules are largely useless, as there are gobs of types out there that are flat-out impossible to do.  This is about the point in my schooling when I decided to yell a very hearty "**** you!" to math.

The funny thing is that this is not true in general. It just happens that all the elementary functions (rational functions, exp and log, and combinations of those) have derivatives that can be easily written out.

The most basic function you encounter beyond that setting is the gamma function (factorial), and there you already run into a situation where you have to introduce a new function for every derivative.
Title: Re: integral of......
Post by: redsniper on December 17, 2008, 12:40:42 am
Oh, kind of like: Heaviside(x)'= delta(x) ?
Title: Re: integral of......
Post by: CP5670 on December 17, 2008, 03:01:48 am
Yes, that's another example, although the derivative there is only meaningful in a weak/distributional sense.
Title: Re: integral of......
Post by: Kosh on December 17, 2008, 08:53:33 am
Quote
ln(tan x) dx  || Substituted tan(x)=u and solved dx = 1/(1+u^2) du

I should have figured it was something like that. Thanks.
Title: Re: integral of......
Post by: Mika on December 17, 2008, 01:49:06 pm
No problem. Physicists know these type of integrals by heart.

Mika
Title: Re: integral of......
Post by: Mongoose on December 17, 2008, 07:52:02 pm
No problem. Physicists know these type of integrals by heart.
Good physicists do.  I, unfortunately, do not.

And I suppose it's a good thing I got out of math when I did, CP.  I don't need another excuse to dislike its higher workings. :p
Title: Re: integral of......
Post by: Dark RevenantX on December 17, 2008, 08:06:35 pm
****.  I'll have to do this in AP Calculus BC for my Senior year, won't I?



Though, I have done harder things...  Such as creating a single formula that calculates the exact horizontal speed that a projectile must have in order to hit a certain point, taking into effect the change in elevation, gravity, actual speed of the projectile, etc (not including air resistance and such).
Title: Re: integral of......
Post by: Mika on December 18, 2008, 01:29:28 pm
Quote
Though, I have done harder things...  Such as creating a single formula that calculates the exact horizontal speed that a projectile must have in order to hit a certain point, taking into effect the change in elevation, gravity, actual speed of the projectile, etc (not including air resistance and such).

I don't think that is yet too bad. Try solving the equation of the projectile in rotating or accelerating coordinate system. Especially rotating coordinate system (like Earth) introduces several rather interesting forces into the play. Did it once, spherical coordinate system was a must on Earth's case. Also do try showing that the effect of the Coriolis force is neglible when the projetile travels 300 m for example...

Lagrange's formulation of mechanics helps a lot of these analysis. The only inherent problem in it is that it cannot handle friction.

Mika