Calculating Pi with a c# program.

[werto165]

Just wondering how I would be able to do this. I've looked into how they calculate pi. One method is this(mostly just infinite series): 

 

pi^4/90 = 1/1^4 + 1/2^4 + 1/3^4 + ... + 1/n^4 

 

and just re arrange the formula for pi once you've gone to the nth term that you want to go to. 

 

I'm just confused how I would store it, surely most data types wouldn't be able to store more than something like 100 decimal places, I want to be able to calculate at least 500 places or so. Or perhaps suggest a better way of calculating it. 

[Nineshadow]

Using tan^-1 is pretty cool, it's not infinite:

So, in C++ (pretty sure it's the same in C#):

 π  = 16 * atan2(1.0, 5.0) - 4 * atan2(1.0, 239.0)

 

Or with the the zeta function(  ) if you like zeta <3:

 

Your formula is Leibniz's. 

 

 

There's also this series :

π=4-4/3+4/5-4/7+4/9-4/11+.....

[DXMember]

roll your own datatype

store it as a string

use arbitrary lenght data type

why would you want to calculate AT LEAST FIVE FRIGGIN HUNDRED PLACES???

[werto165]

roll your own datatype

store it as a string

use arbitrary lenght data type

why would you want to calculate AT LEAST FIVE FRIGGIN HUNDRED PLACES???

Dunno well if super computers can calculate like 5 trillion or something insane, 500 didn't seem too hard. Just curious that's all, maths is just interesting and I need to get back into programming again as my course coming up in sept will use matlab so probably worth doing that. 

[DXMember]

Dunno well if super computers can calculate like 5 trillion or something insane, 500 didn't seem too hard. Just curious that's all, maths is just interesting and I need to get back into programming again as my course coming up in sept will use matlab so probably worth doing that. 

oh the meth lab is great, i did it recently, personally my favourite mission in GTA V

 

also the mathlab is quite a bit different from c# afaik

[,,,,,,,,,,,,,,,,,,,,,,,,,,]

Using tan^-1 is pretty cool, it's not infinite:

 

 

 

Or with the the zeta function(  ) if you like zeta <3:

 

Your formula is Leibniz's. 

 

 

There's also this series :

π=4-4/3+4/5-4/7+4/9-4/11+.....

 

Next time I have math homework, I'm coming to you OK?

[Nineshadow]

Next time I have math homework, I'm coming to you OK?

As long as it's not something I haven't studied yet, ok.

[Nuluvius]

C#/.NET in general is not a good choice, it's far too slow. This sort of thing should be done natively, at the lowest level possible. I'd be inclined to play with C++ AMP/CUDA for this.

[werto165]

C#/.NET in general is not a good choice, it's far too slow. This sort of thing should be done natively, at the lowest level possible. I'd be inclined to play with C++ AMP/CUDA for this.

I've not had any experience with those and I would feel so stupid trying to do it... Can I do CUDA stuff with a 660ti? I would assume so. Is C++ that much different from c#? Also what software/ide would  I need just like notepad ++ or does visual studio work  ?

[Nuluvius]

I've not had any experience with those and I would feel so stupid trying to do it... Can I do CUDA stuff with a 660ti? I would assume so. Is C++ that much different from c#? Also what software/ide would  I need just like notepad ++ or does visual studio work  ?

 

Hell yes you can. Yes C++ is a different animal to C#; evolved for a different problem domain. That said their evolution is becoming somewhat convergent in places given the recent decade with C++ 11 and the upcoming 14.

 

Quintessentially in this context it's speed: .NET utilizes JIT, C++ doesn't and applying GPU parallelism yields lot's of cores to crunch the problem with...

[werto165]

Hell yes you can. Yes C++ is a different animal to C#; evolved for a different problem domain. That said their evolution is becoming somewhat convergent in places given the recent decade with C++ 11 and the upcoming 14.

 

Quintessentially in this context it's speed: .NET utilizes JIT, C++ doesn't and applying GPU parallelism yields lot's of cores to crunch the problem with...

Thanks for the info, I will try and do it, probably get nowhere tho.... 

[Nuluvius]

Thanks for the info, I will try and do it, probably get nowhere tho.... 

 

At the very least a little exploration of C/C++ would be very useful.

[werto165]

At the very least a little exploration of C/C++ would be very useful.

Well afaik unreal engine uses it too which I want to mess with a bit, programming can be really enjoyable but sometimes I hate it with a passion, don't know perhaps it's not for me. 

[Nuluvius]

Well afaik unreal engine uses it too which I want to mess with a bit, programming can be really enjoyable but sometimes I hate it with a passion, don't know perhaps it's not for me. 

 

I would say that it's perfectly normal to feel that way about it. I too both hate it and love it at times and so does just about everyone I know/have known in the industry.

 

Perhaps take it as a positive sign that you're able to feel such strong polar feelings about it.

[Guest]

If you don't roll your own data type, you'll need some kind of big number library, e.g. https://gmplib.org/.

 

Coding your own big number library would be a pretty fun and educational project, though. The basic strategy is that you will store the values of the number in a contiguous byte array, but provide an interface to that array in the form of operations (functions) that implement the arithmetic. Your operations would need to iterate over the array, making sure to carry properly and avoid values that exceed the size of the native data types.

 

I recommend doing this in C, but an object oriented C++ approach could work. Worry about getting fancy with CUDA, etc. once you have a working solution.

[cluelessgenius]

you want your pc lose its mind? because that how you loose your mind.

 

just use math.pi or something