C#

harold aptroot wrote:

is that XOR-swap really faster than a normal swap?

PIEBALDconsult wrote:

How could it be?

harold aptroot wrote:

how that would help

Max Holder wrote:

Only up to n=200; n=201 result is bigger than ulong

checked { b += a; }

• n = 92: 7540113804746346429
• n = 93: 12200160415121876738
• n = 94: 1293530146158671551

private static ulong GetFib3(int n)
{
    ulong a = 0;
    ulong b = 1;

    if (n == 0) return a;
    if (n <= 2) return b;

    if ((n & 1) == 0)
    {
        a = 1;
        n--;
    }

    int i = n - 1;
    do
    {
        a += b;
        b += a;
        i -= 2;
    } while (i != 0);

    return b;
}

private static ulong
Fibo2
(
  int n
)
{
  ulong a = (ulong) n & 1 ;
  ulong b = a ^ 1 ;

  for ( n /= 2 ; n > 0 ; --n ) checked { a += b += a ; }

  return ( a ) ;
}

private static ulong[] a = new ulong [ 3 ] ;

private static ulong
Fibo
(
  int n
)
{
  a [ 0 ] = a [ 1 ] = 0 ;
  a [ 2 ] = 1 ;
  int x = 2 ;
  int y = 1 ;

  for ( ; n > 0 ; --n )
  {
    a [ 0 ] = a [ x ] += a [ y ] ;

    x ^= 3 ;
    y ^= 3 ;
  }

  return ( a [ 0 ] ) ;
}

Jacob Barton wrote:

There are many similar programs available on the web.

Mycroft Holmes wrote:

What is your unique requirement that these many apps can't service.

• Can nodes be re-used (ie serve as end points for multiple beams)?
  
  
  • Should join strength be a variable or ideal?
  • Is there a common joint, or are one connected to side of other?
• Can you have multiple load vectors on each beam?
• What types of load do you want to model (of the five classic)?
• Do you want to distinguish between, for want of a better term, primary (entered by user) and secondary (calculated and distributed over the structure) loads?

Orjan Westin wrote:

For instance, I'd probably keep nodes, members and primary and secondary loads separate in a relational database, and let the loads have a relationship to a member, which has a relationship to two nodes. So that's four tables of data to start you with.