variables larger than floats

So I am having trouble with numbers bigger than floats.

A few questions. Can the variable type

Code:

---

long long

---

hold decimal places. I think it doesnt.
I am running an iterative calculation in a for loop that runs 100 times. The first few answers are small enough for a float to hold - need up to 8 decimal places. But then they become to big for floats, so I wanted to hold them all in long long type variable. But it doesnt seem to store decimals.

Please can someone tell me a veriable type, that is very large, but can also hold decimals. And what the code is when printing or scanning for one.

i.e printf("%d", int);

the %d is the "code for integers".

Thanks

Alex

I tend to default to double, but there is also long double. What are your exact requirements, including precision and accuracy?

Floating point numbers aren't really that great for high precision. You'd be better off finding a way to convert over to integers, since there is no loss of precision there. Floating point numbers suffer from precision loss due to the way they're stored.


Quzah.

Basically for smaller programs I always used floats because they could hold decimals. I had no idea of a precision or accuracy. Now that the actual size of the numbers are too large, I need something like a float, that can hold a few decimal places - forget 8 if thats too many, but can also hold the number size.

I have tried long long, but i dont think it can hold decimals.

Quote:

---

Originally Posted by a.mlw.walker

Basically for smaller programs I always used floats because they could hold decimals. I had no idea of a precision or accuracy. Now that the actual size of the numbers are too large, I need something like a float, that can hold a few decimal places - forget 8 if thats too many, but can also hold the number size.

I have tried long long, but i dont think it can hold decimals.

---

If i were you i would try double,if it does not enough to satisfied then maybe long double..

Doubles are probably good enough; try them out. You read and print them with

Code:

---

double n;
scanf("%f", &n);  /* scanf: %f is for floats and doubles */
printf("%lf\n", n);  /* printf: %f is for floats, %lf is for doubles */

---

When you say to big for floats do you mean to many decimal places or the number is higher than the maximum value for a float?

Just because an integral type doesn't normally hold decimals doesn't mean you cant use it as if it did. For example if you wanted to hold 3 decimal places you can store the numbers multiplied by 1000. So a value of 3200 would mean 3.2 etc. You have to multiply of divide by 1000 when doing some things to keep the numbers in the right range, but it's not very hard.

Doubles are too small, what are the symbols for long doubles?

Really, what are you doing, and what range/number of decimals do you need?

--
Mats

Quote:

---

Originally Posted by a.mlw.walker

Doubles are too small, what are the symbols for long doubles?

---

%Lf in both cases.

You may also want to consider using a fixed precision math library.

Please clarify what you requirements are. What is the largest and smallest number you will need to represent. And what is the precision you need? I see you are an engineering student so you should know what i mean by precision.

A couple of decimal places would be nice enough, but the numbers get very large due the calculation being Lorenz' equation for his butterfly effect -so they get bigger each time - it managed to print out
11343638689789134000000000000000.000000
but couldnt get the next one.

the values are calculated based on any number the user enters, possibly ranging from 0.2 to maybe 1000.

thanks

Sounds like you need GMP (GNU MultiPrecision library) - it has functions to give as much precision as the machine can support in terms of memory - something like millions or billions of digits in a modern PC.

--
Mats