Hi-prec math for visual studio c#?

[devronious]

Is there a Hi-prec math for visual studio c#?

Thanks in advance,

Devin

[CRGreathouse]

There's at least one commercial package out there, but it's not cheap:
http://www.extremeoptimization.com/

What are your needs? What does the library need to do, and how fast does it need to work? Are you just using big numbers, or are you doing matrices? Basic operations (+, -, *, /, %, ^), transcendental (sin, sinh, asin, exp, log, erf, gamma), or specialized (bessel, hypergeom, prime operations, etc.)? Integer or floating-point?

[devronious]

I'm doing floating point with basic math ops, ie: +,-,*,/,^, sqrt

[CRGreathouse]

Does it need to be fast? Is anything running in a tight loop, or are you just writing (say) a calculator where an extra millisecond won't hurt?

[devronious]

Doesn't need to be fast.

[CRGreathouse]

You could write your own, then, or have someone write it. Quick code:

class Bignum {
protected ulong[] limb;
public Bignum (long initialValue) {
limb = new ulong[] {initialValue};
}

public Bignum (long[] initialValue) {
Array.Copy(limb, initialValue, initialValue.Length);
}

public Bignum (string initialValue) {
// Bad method, replace with a faster one if needed
limb = new ulong[] {long.Parse(initialValue.Substring(0, 18))};
for (int i = 18; i < initialValue.Length; ++i)
limb = limb * 10 + long.Parse(initialValue[i]);
}

public Bignum (Bignum initialValue) {
Array.Copy(limb, initialValue.limb, initialValue.limb.Length);
}

public static Bignum operator+ (Bignum a, Bignum b) {
long carry = 0;
int sz = Math.Max (a.limb.Length. b.limb.length);

Bignum result = new Bignum(a);
for (int i = 0; i < sz; ++i) {
long o1 = result.limb[i];
long o2 = b.limb[i];
limb[i] += b.limb[i] + carry;
if (result.limb[i] < o1 || result.limb[i] < o2)
carry = 1;
else
carry = 0;
}
if (carry == 1) {
Array.Copy(result.limb, result.limb, result.limb.Length + 1);
result.limb[sz] = 1;
}
return result;
}

public static Bignum operator* (Bignum a, Bignum b) {
// Code here
}

public static Bignum operator^ (Bignum a, uint b) {
Bignum result = new Bignum(a);
Bignum carry = new Bignum(0);
while (b > 1) {
if ((b&1) == 1)
carry *= result;
result *= result;
b >>= 1;
}
return result * carry;
}
}

I haven't compiled this (or even used an IDE) but this should be a good start. Subtraction is easy, multiplication not too hard, and division... well, code that as far as you need it. I decided to do the exponentiation so it wouldn't be too slow.

[devronious]

Cool, thanks for your time, effort & code. I like your usage of binary ops.

-Devin