Thread: attempting to calculate e with spigot algorithm

Hello, everyone. I am trying to use a c program I found on the Internet:

Code:

int main(void){
  int a[3302],b=3301,*c=a,d,e,f;
  for(e=b; --e; *c++=1)
      ;
  *c=2;
  for(d=2001; d--; printf("%05d",f)) {
    for(c=a, e=b; e; f/=e--){
      f += *c * 1e5;
      *c++ = f%e;
    }
  }
}

(I did insert an extra line at the end after the compiler complained about the lack of it.)

I am trying to figure out why it does not seem to be giving me the right value for e. It did compile with some warnings but no fatal errors. The main warning were for the printf and for line 8 converting integer to floating point.

Personally, I can say that I used to write c programs and fortran 77 programs when I studied at uCsD 1995-97 as well at mathlab at Pomona college 1994-5. But I'm a little rusty and I'm not that clear on what all the syntax is supposed to be doing: a[3302], *c, the single ; for line 4, printf("%05d",f) does what?, and lastly, *c++ = f%e; has % and I don't know what that does, again.

I do have a basic grasp of the three for loops and what the basic math operations are doing, but without really understanding the program, I don't know how to adapt it to what I want to do, namely, to calculate even more digits of e (correctly, of course).

It took me a few hours of looking around on the Internet to finally find a spigot algorithm that I could put into a computer language to try to calculate the base of natural logarithms. There were many formulas that produce e and various similar values (e^x, 1/e, etc), and I read that e has only been calculated out to a billion decimal places, so if the spigot algorithm could allow me to pick up where I leave off, then I can keep adding to my collection of digits.

Help is greatly appreciated.

It gives the correct value of e, except without the decimal point. What value are you getting? The last few digits might be wrong, but that's a property of spigot algorithms in general.

Also, the algorithm appears to be O(n^2) or worse, so you've very little chance of computing a billion digits using such a method.

% is the modulus operator, which you can think of as taking the remainder after division.

This code is very very very bad and ugly -- don't learn coding style from it!

And it DOES seem to output the digits of e correctly.


Listen to the warnings. They may signal possible runtime errors or bugs.
If you want to get rid of the "convert double to int" warning, cast the double first:

Code:

      f += (int)(*c * 1e5);

Sure, blame the compiler - nice of you not to state which one that is....

> The point of this spigot algorithm is partly to be only 10 lines in length
How about blame the code? Terse, but wrong is still wrong.


First attempt at compiling your code

Code:

$ gcc -W -Wall -ansi -pedantic -O2 foo.c
foo.c: In function ‘main’:
foo.c:4: warning: ‘f’ may be used uninitialized in this function

OK, fix that problem, try again

Code:

$ gcc -W -Wall -ansi -pedantic -O2 foo.c
$ ./a.out
271828182845904523536 yada yada yada

My guess is your compiler doesn't initialise 'f' to zero like the code expects.
int a[3302],b=3301,*c=a,d,e,f=0;
Not that it's obliged to initialise it to anything at all unless you say so.
So if it works at all for some people, that's just one happy coincidence.

> My compiler is a freeware compiler I got from
Yes, it is old, and as such it's int's are 16-bit rather than 32 bit.
When I change my version to use 16-bit integers, it all goes horribly wrong as well.

Try a 'long' version which should result in 32-bit numbers for your fossil compiler.

Code:

#include <stdio.h>
int main(void){
  long int a[3302],b=3301,*c=a,d,e,f=0; /* added long */
  for(e=b; --e; *c++=1)
    ;
  *c=2;
  for(d=2001; d--; printf("%05ld",f)) { /* added l to the format */
    for(c=a, e=b; e; f/=e--){
      f += *c * 1e5;
      *c++ = f%e;
    }
  }
  return 0;
}

> but I didn't want to spend the money on a compiler yet.
Dev-c++ seems to be the free compiler of choice.
http://www.compilers.net/
http://www.thefreecountry.com/compilers/cpp.shtml

Dev-C++ is not a compiler, it's an IDE. It uses the MinGW compiler package.

Homepage

You don't need money; gcc (and Dev-C++) is 110% free of charge. It complies to current standards.