truncation

[amrutansu garanaik]

I am given the task to find the value of PI up to decimal poins 'x'
awhere 'x' is input through keyboard.
If I use c++ and use setprecision() defined in iomanip.h, I am getting
rounded off results. I need accurate truncated value. If I use c, I
can not use a variable in precision field in printf() . What can I do?

e.g. PI=3.1415(9265)
I am getting 3.1416

Please help

 

[Vyom]

Your approach is wrong.
You are just dividing 22 by 7 and expecting the results.

You need to develop some kind of loop to calculate the values 'manually'. Make a loop to keep dividing 22 by 7 and store the remainder in a variable.
After that its just a matter of executing loop until remainder isn't X.

 

[nims11]

if x can be large, you need to develop a manual algorithm. But in that case, the problem is not simply performing a manual division of 22 and 7, as pi is irrational. 22/7 is an approximation which i think is valid upto about 10 decimal places. If x is small enough, a better approach will be acos(-1). Anyways, back to your question, eg you want to print pi truncating after 2 decimal places, one thing that can be done is

int(pi*100)/100.0

 

[harshilsharma63]

All build in data types (signed or unsigned) have data limit. You need to create a new class/data type that has no limit on the number of digits that can be stored in it. Think over how to do that, it's really easy.

 

[Santa Maria!]

Proof that 22/7 exceeds pi

But I'm guessing you're not that concerned about insane amounts of precision.
If it's just for display purposes, why don't you just use a string representation obtained from your calculation? Calculate the various digits and keep appending them to a std::vector<int(or char)>.
If you intend to use your truncated value for calculations... well, you better read up on the subtleties of floating point arithmetic.
What Every Computer Scientist Should Know About Floating-Point Arithmetic
(I keep throwing this link around, but I still haven't gotten around to reading it fully myself )

 

[pranav0091]

Slightly off-topic, but Pi should be calculated as the sum of its infinite series and not as anything else if you want any arbitrary level of precision.

I am not much of a programmer, so this might be quite stupid, but why not use the infinite series and then calculate one digit at a time and store it as a text file or something. Have a look here: Chudnovsky algorithm - Wikipedia, the free encyclopedia

 

[Anorion]

130 mb text dump up to 100 mill digits here Pi to 100 million decimal places : QuickPi : Free Download & Streaming : Internet Archive
think that's the easiest way out