sinh(3391014490.0)

[secondrate]

Hi guys,

I want to evaluate sinh(3391014490.0) as i'm doing numerical simulation. It produces inf and RR: overflow in NTL (i used (0.5)(e^x + e^-x)). Anyone know of a way to evaluate this?

Thanks!

[Moschops]

e^-3391014490 is so close to zero that we can ignore it for purposes of this calculation.

e^3391014490 is really, really, really, really big.

Half of that number is still far, far too big. Too big for a 64 bit number. You'll have to use some kind of specialised library made for handling really, really, really big numbers.

Depending on what you're using it for, you might be able to box clever and simplify whatever your calculation actually is so that you don't have to evaluate such huge values.

[secondrate]

hi moschops, i know it's infinity and i've tried using NTL (number theory library) as i've heard that it can evaluate large floating point numbers. (in fact, it can evaluate exp(91014490) i.e., chopping the frst 2 three's.

So now, would you know of any other library?

[Moschops]

Maybe this one:

http://gmplib.org

[PanGalactic]

Even Google doesn't know the answer to that question:
http://www.google.com/search?q=e^3391014490%3D

It knows this though:
http://www.google.com/search?q=e^33%3D

If Google doesn't know, does the problem really exist? :-)

[Moschops]

Wolfram Alpha has an answer, but I don't like it very much... :)

Edit: I'm the number one suggestion on your google link. Neat :)

[Kyon]

Wolfram Alpha is neat. (I found a way to break it, though. :P)

Do this: Calculate the length of a function over two x coordinates which include the asymptote of said function.

Example (for tan(x)):
integrate √(1+cos(x)^-4) dx from x=-pi/2 to x=pi/2